Integration of Hydrological and Economical Aspects

for Water Management in Tropical Regions.

Case Study: Middle Magdalena Valley, Colombia.

MARIA CRISTINA ARENAS BAUTISTA, [Link]

Universidad Nacional de Colombia

School of Engineering
Doctoral Program in Civil Engineering
Bogota D.C., Colombia
2020
Integration of Hydrological and Economical Aspects
for Water Management in Tropical Regions.
Case Study: Middle Magdalena Valley, Colombia.

MARIA CRISTINA ARENAS BAUTISTA, [Link]

A dissertation submitted as partial fulfillment of the requirements for the degree of:

Doctor in Engineering, Ph.D

in the Program of Civil Engineering

Research Line in Water and Environment Engineering

Hydrodynamics of Natural-Media Research Group - HYDS

Adviser:
Leonardo David Donado Garzón, Ph.D
Associate Professor

Universidad Nacional de Colombia

School of Engineering
Graduate Programs in Civil and Agricultural Engineering
Bogota D.C., Colombia
2020
 iii

This line intentionally left white

This work was funded by Colciencias and Universidad Nacional de Colombia with the
Financing Program 647.

This work was also supported by Universidad Nacional de Colombia Research Program, with
National Call for the International Mobility N. 9599.

Views and conclusions contained in this document are those of the authors and should not be
interpreted as necessarily representing the social policies or endorsements, either expressed
or implied any of the funding institutions.
iv

This line intentionally left white

A mi familia, lo más importante en mi vida.

A mi mamita, Ligita, quien nunca me dejo desfallecer.
A Laura, a quien espero dejarle como ejemplo que
educarse si vale la pena.
A Pedro, mi compañero de vida, quien pese a las
dificultades me apoyó incondicionalmente.
A Leonardo, mi Director y amigo, por creer en mí.
 v

Acknowledgment

I want to give the special thanks to the guidance given by professors Monica Riva and Alberto
Guadagnini from the Politecnico di Milano (Italy) during my internship. Also, I want to
thank the members of the Hydrodynamics of Natural-Media - HYDS and GIREH research
groups, whose questions, discussions, and examples gave me new ideas to improve each part
of this work. Juan Pescador, I not only want to thank you I also want to recognize your
work in the consolidation of the geological model; This would not be possible without your
help. Thanks to Edwin Saavedra, Pedro Arboleda, Alejandro Proaño and Sarah Coral for
your help in style review of this document. Thanks to Universidad Nacional de Colombia,
that partially funded my Ph.D studies.

I would like to thank my advisor Leonardo Donado for his patience and unconditional support
in my doctoral training process. Thank you Leonardo, for showing me a different path of
life, for contributing to improve my professional skills and as a friend, for your kind guidance
in difficult moments.
vi

Graphical Abstract
 vii

Resumen

El recurso hídrico es un factor determinante en el desarrollo económico y social de las

comunidades dada la amplia necesidad que se genera en torno a su aprovechamiento. El
uso del agua ha generado a través de los años, presiones sobre su disponibilidad, las cuales
usualmente se resolvían aumentando el suministro, explorando y desarrollando nuevas fuentes
y, expandiendo las extracciones de las ya existentes. En Colombia, la concesión de agua es
el derecho al uso limitado del agua, otorgado para desarrollar una actividad económica.
Esta concesión debe estar relacionada con la disponibilidad de agua para garantizar la
preservación y el uso eficiente del agua. Para asignar el recurso hídrico de manera eficiente
se requieren herramientas que ayuden a tomar decisiones mediante el análisis del régimen
hidrológico integral (aguas superficiales y subterráneas), lo cual es un problema complejo
en áreas que carecen de datos hidrológicos confiables; y que permiten una evaluación sobre
la disponibilidad en un contexto económico. En este orden de ideas, el objetivo principal
de esta investigación fue proporcionar una aproximación metodológica que permita integrar
aspectos hidrológicos, hidrogeológicos y económicos en la asignación de agua entre diferentes
usuarios, priorizando las necesidades humanas y los procesos ecosistemicos, para establecer
estrategias de gestión a escala regional.

En consecuencia, esta investigación realizó una integración de aspectos hidrológicos,

hidrogeológicos y económicos, utilizando la cuenca geológica del Valle Medio del Magdalena
(MMV) como laboratorio a escala real. Debido a que esta zona es centro de abastecimiento
para la pequeñas poblaciones rurales y se desarrollan de manera conjunta actividades
económicas relacionadas con minería, agricultura, acuicultura, producción pecuaria,
industria, generación hidroeléctrica, servicios de seguridad y emergencia y exploración y
producción de hidrocarburos, se analizó el comportamiento hídrico y económico del sistema
respecto a la disponibilidad de agua superficial y subterránea, y su asignación a los diferentes
usuarios. Para ello, se definieron tres etapas: (1.) caracterización del sistema hidrológico,
(2.) caracterización del sistema hidrogelógico, y (3.) su integración en un modelo de
optimización económica.

En la primera etapa, se analizó mediante herramientas numéricas el comportamiento

hidrológico del sistema para caracterizar la oferta hídrica. En adición, se identificaron las
zonas de recarga y se investigaron las alteraciones hidrológicas que afectan la cantidad de
agua en el sistema. La modelación hidrológica permitió realizar una evaluación exhaustiva
de la interacción entre la dinámica del ciclo hidrológico y las condiciones climáticas. Luego,
se realizaron los análisis de sensibilidad e incertidumbre para evaluar la influencia de los
 viii

principales parámetros asociados con el modelo y a partir de ello, se validó una metodológia
que permite: (i) seleccionar valores apropiados para los parámetros de los modelos y (ii)
evaluar en qué medida la variación de estos parámetros afecta una respuesta simulada.

En la segunda etapa, se integraron la caracterización geológica, la hidrológica y la

hidráulica en un modelo hidrogeológico para estimar el volumen de agua y la descripción del
sistema de flujo subterráno. El resultado de esta etapa permitió consolidar una metodología
para restringir asertivamente un modelo inverso altamente parametrizado con pocos o
sesgados datos de campo, estimar parámetros hidráulicos de acuíferos y analizar la variación
espacial y temporal que presentan estos parámetros a escala regional.

Finalmente, en la tercera etapa, se integraron los aspectos hidrológicos superficiales y

subterráneos desarrollados anteriormente, en un marco de optimización económica para
determinar la asignación conjunta de agua y la gestión de la calidad del recurso hídrico. Este
aparte tuvo como objetivo principal análizar el beneficio de uso del agua en un modelo de
flujo regional que integra múltiples ofertas de agua y de demandas por parte de los usuarios.
Aquí, se analizó el modelo de asignación desde una escala regional con el fin de consolidar
tipologías de uso por sector económico y determinar estrategias de gestión a escala regional
que permitan reducir los conflictos por uso y calidad, y fortalecer la gestión y administración
del recurso hídrico en la zona.

Los resultados generales de esta investigación permitieron identificar y evaluar de manera

conjunta los problemas y estrategias de gestión, en una cuenca tropical con escases de
información. Adicionalmente, se concluyó sobre cómo la cuantificación de la oferta hídirica
afecta el proceso de asignación entre diferentes usuarios y este proceso a su vez, esta en
función de la calidad. Como parte de la etapa final de esta investigación, se analizó a través
de escenarios futuros el comportamiento del sitema hídrico.
 ix

Abstract

Water resources are a determining factor in the economic and social development of
communities, given the need that is generated around its use. Over the years, this use has
generated pressure on water availability, which were solved by increasing supply, exploring
and developing new water sources, and expanding the existing extractions. In Colombia,
water concession is the right to the limited use of water, and it is granted to develop economic
activity. This concession must be related to water availability to ensure the preservation
and efficient water use. However, to allocation water resources efficiently, tools that help
to make decisions by analyzing the hydrological regime (surface and groundwater) in areas
with lacking reliable data on water availability in an economic context are required. In this
context, the main goal of this research was to provide a methodological approximation that
allows integrating hydrological, hydrogeological, and economic aspects in water allocation
between different users, prioritizing human needs and ecosystem processes, to establish
management strategies at a regional scale.

In this research performed an integration of diverse hydrological, hydrogeological, and

economical aspects, using the Middle Magdalena Valley (MMV) geological basin as a real
scale laboratory. Because this area is a primary supply center for the Colombian population,
and economic activities related to mining, agriculture, aquaculture, livestock, industrial,
services, and O&G exploration and exploitation are developed in conjunction; hydric and
economic behavior of the system was analyzed. This analysis was carried out in regards to
water availability (surface and groundwater), and its allocation to different stakeholders.
For it, three phases were defined: (1.) to characterize the hydrological system, (2.)
to characterize the hydrogeological system, and (3.) its integration into an economic
optimization model.

In the first phase, the hydrologic system behavior was analyzed through a numeric tool,
to characterize the water supply, the recharge zones were identified, and the hydrologic
alterations affecting the water supply were evaluated. The hydrological modeling allowed to
perform an exhaustive interaction analysis between the hydrologic cycle dynamic and the
weather condition and land use. Then, it was made an analysis of uncertainty and sensitivity
to evaluate the influence of the principal parameters associated with the model. From this
analysis, it was validated a methodology allowing to: (i) select proper values for the model
parameters, and (ii) evaluate how the model parameters variations influence a simulated
response.
 x

In the second phase, the geological, hydrological and hydraulic characterization was
integrated into a hydrogeological model to estimate the water volume and groundwater
flow system description. The result of this phase allowed to consolidate a methodology to
assertively restrict a highly parameterized inverse model with lack of information, estimate
hydraulic parameters of aquifers and analyze the spatial and temporal variation presented
by these parameters at the regional scale.

Finally, in the third phase, the hydrological aspects (surface and groundwater) were
integrated with an economic optimization framework. This allows them to determine water
allocation and water resources quality management. The main objective of this phase was
to analyze the water use profit in a regional flow model, integrating multiple water supplies
(surface and groundwater) and multiple demands. Here, the allocation model was analyzed
from a regional scale in order to consolidate typologies of use by economic sector, and
determine management strategies at a regional level.

The general results of this research allowed to identify problems and evaluate management
strategies, in a tropical basin at the regional level. Additionally, it was concluded that the
quantification of water supply affects the allocation process between different stakeholders
and this process, in turn, is a function of water quality. As part of the final stage of this
research, the water system behavior was analyzed through future scenarios.
 xi

Keywords This line intentionally left white

Keywords: Hydro-economic Optimization, Integrated Water Resources Management,

Economic Aspects, Water Allocation, Management Strategies, Hydrological modeling,
hydrogeological Model, Tropical Regions, uncertainty, sensitivity analysis, Sobol and AMA
indices, pilot-points technique, PEST, inverse parameterization.

Palabras clave: Optimización Hidroeconómica, Gestión Integral del Recurso Hídrico,

aspectos económicos, asignación de agua, estrategias de gestión, Modelación hidrológica,
Modelación hidrogeológicoa, incertidumbre, análisis de sensibilidad, índices Sobol y AMA,
técnica de puntos piloto, PEST, parametrización inversa.
 xii

Related Publications
Papers
Arenas-Bautista, M.C., Pescador-Arévalo, J.P., Donado, L.D., Saavedra-Cifuentes, E.Y.
and Arboleda-Obando, P.F. (2020). Hydrogeological Modeling in Tropical Regions via
FeFlow. Earth Sciences Research Journal. (Accepted for Publication).

Arenas-Bautista, M.C., Saavedra-Cifuentes, E.Y., Arboleda-Obando, P.F and Donado,

L.D. (2020). Impact of pilot points number and plausibility for determination of hydraulic
properties in a real regional model of groundwater. In preparation to Groundwater Journal.

Arenas-Bautista, M.C., Gardeazabal, N., Arboleda-Obando, P.F, Guadganini, A., Riva,

M. and Donado, L.D. (2020). Comparing Global Sensitivity Analysis approaches of a
Semi-Distributed Hydrological Model in Tropical Regions. Study case: Middle Magdalena
Valley, Colombia. In preparation to Water Resources Management Journal.

Arenas-Bautista, M.C. and Donado, L.D. (2020). Hydro-economic Optimization of the

Assignment and Management of Water Quality in Tropical Basin. In preparation to Water
Resources and Economics Journal.

Conferences
Arenas-Bautista, M. C. and Donado, L.D. (2020). Integrated Water Resources
Optimization Model a tropical basin in Colombia. In European Geosciences Union -EGU, 3
- 8 may. Vienna (Austria).

Arenas-Bautista, M. C., Pescador-Arévalo, J.P. and Donado, L.D. (2019). Pilot-Points

Impact in Determination of Hydraulic Properties in a Tropical Basin. Study Case: Middle
Magdalena valley . In Darcy Lecture -ICGW, 22 November. Bogotá (Colombia).

Donado, L.D., Arenas-Bautista, M.C. and Pescador-Arévalo, J.P. (2018).

Hydrogeological Characterization in Tropical Regions with lack of information subject to
competing uses of groundwater. In American Geophysical Union -AGU, 10-14 December.
Washington (USA). DOI: 10.13140/RG.2.2.12349.10727

Arenas-Bautista, M. C., Pescador-Arévalo, J.P. and Donado, L.D. (2018).

Three-dimensional geological model applied for groundwater flow simulations in the Middle
 xiii

Magdalena Valley, Colombia. In American Geophysical Union -AGU, 10-14 December.

Washington (USA). DOI: 10.13140/RG.2.2.19059.99365

Arenas-Bautista, M.C., Gardeazabal, N., Arboleda-Obando, P.F, Guadganini, A., Riva,

M. and Donado, L.D. (2018). Sensitivity Analysis to uncertain Parameters of TopModel
in Tropical Regions with application to the Middle Magdalena Valley (Colombia). In
European Geosciences Union -EGU, 8-13 April. Vienna (Austria). DOI: 10.13140/RG.2.2.29
463.68008.

Arenas-Bautista, M. C., Gardeazabal, N., Arboleda-Obando, P.F, Guadganini, A., Riva,

M. and Donado, L.D. (2017). Hydrological Modeling in Tropical Regions via TopModel.
Study Case: Central Sector of the Middle Magdalena Valley - Colombia. In European
Geosciences Union -EGU, 23-28 April. Vienna (Austria). DOI: 10.20944/preprints201807.
0210.v1.

Arenas-Bautista, M.C., Suescún-Casallas, L.C., Guadganini, A., Riva, M. and Donado,

L.D. (2017). Hydrogeological model of the Middle Magdalena Valley – Colombia. In
International Conference on Groundwater -ICGW, 28-31 August. Bogotá (Colombia). ISBN:
978-958-59856-0-5.

Arenas-Bautista, M.C., Gardeazabal, N., Arboleda-Obando, P.F, Guadganini, A., Riva,

M. and Donado, L. D. (2017). Hydrological Modeling the Middle Magdalena Valley
(Colombia). In American Geophysical Union -AGU, 11-15 December. New Orleans (USA).

R Package - Tool

García-Echeverri, C., Arenas-Bautista, M. C, and Donado, L. D. (2020).

[Link]: Global Sensitivity Analysis Tool - R package (Version v1.0.0). Zenodo.
[Link] [Link]
Contents

Resumen vii

Abstract ix

List of Symbols xvi

Acronyms xvii

1. General Introduction 1

2. The Middle Magdalena Valley (Geological System) 6

2.1. Study Area . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7
2.2. Hydrology Context . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8
2.3. Geology Context . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13

3. Hydrological Model 15
3.1. Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 16
3.2. Methodology . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 17
3.2.1. Hydrological Model: TopModel . . . . . . . . . . . . . . . . . . . . . 17
3.2.2. Global Sensitivity Analysis . . . . . . . . . . . . . . . . . . . . . . . . 21
3.3. Results and Discussion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 24
3.3.1. Parameter estimation . . . . . . . . . . . . . . . . . . . . . . . . . . . 25
3.3.2. Global Sensitivity Analysis . . . . . . . . . . . . . . . . . . . . . . . . 27
3.3.3. Model Performance . . . . . . . . . . . . . . . . . . . . . . . . . . . . 32
3.3.4. Recharge . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 36
3.3.5. Limitations and Conclusions . . . . . . . . . . . . . . . . . . . . . . . 38

4. Hydrogeological Model 39
4.1. Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 40
4.2. Methodology . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 42
4.2.1. Governing Equations and Modeling . . . . . . . . . . . . . . . . . . . 42
4.2.2. Boundary Conditions . . . . . . . . . . . . . . . . . . . . . . . . . . . 43
4.2.3. Regularization and Parametrization . . . . . . . . . . . . . . . . . . . 45
4.2.4. Estimation and Objective Function . . . . . . . . . . . . . . . . . . . 47
Contents xv

4.3. Model Development . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 50

4.3.1. Geological Model . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 50
4.3.2. Hydrogeological Conceptual Model . . . . . . . . . . . . . . . . . . . 54
4.3.3. Numerical Implementation . . . . . . . . . . . . . . . . . . . . . . . . 58
4.4. Results and Discussions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 61
4.4.1. CZ and PP . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 61
4.4.2. Regularization Methods . . . . . . . . . . . . . . . . . . . . . . . . . 66
4.4.3. Plausibility Effect . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 71
4.5. Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 75

5. Integrated Water Resources Optimization Model 77

5.1. Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 78
5.2. Methodology . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 79
5.2.1. Parameters Associated with Water Usage . . . . . . . . . . . . . . . . 80
5.2.2. Water Demand and Quality . . . . . . . . . . . . . . . . . . . . . . . 81
5.3. Baseline and Optimization Model . . . . . . . . . . . . . . . . . . . . . . . . 87
5.4. Modeling Results and Discussions . . . . . . . . . . . . . . . . . . . . . . . . 89
5.5. Further Model Development and Limitations . . . . . . . . . . . . . . . . . . 94
5.6. Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 95

6. Water Resources Management and Planning 96

6.1. Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 97
6.2. Management Strategies . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 98
6.3. Strategies for Planning and Management of Water Resources. . . . . . . . . 100
6.4. Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 102
6.4.1. Management strategies proposed for allocation process. . . . . . . . . 102
6.4.2. Management Strategies Analysis. . . . . . . . . . . . . . . . . . . . . 106
6.5. Discussion and Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . 109

7. Conclusions and Outlook 111

7.1. Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 111
7.2. Outlook . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 114

A. Annex: Initial Data 116

B. Annex: Statistical Analysis of the Hydrological Model 133

C. Annex: Hydrogeologic Model 135

D. Annex: Optimization Model Data 142

References 151
List of Symbols

Symbol Units Description

◦
T C Temperature
P LT −1 Precipitation
Q L3 T−1 Flow
E LT−1 Evapotranspiration
ETo LT−1 Potential Evapotranspiration
Ea LT−1 Real Evapotranspiration
Rs LT−1 Solar Radiation
Ra LT−1 Extraterrestrial Radiation
ai L Upslope Contributing Area
βi radians Average Slope
D L Local Storage Deficit
h L Hydraulic Head
K LT−1 Hydraulic Conductivity
k L2 Permeability
∇h L Hydraulic Gradient
η - Porosity
SS L−1 Specific Storativity
t T Time
T L2 T−1 Transmisivity
−Dimensionless
Acronyms

COD Chemical Oxygen Demand

CZ Constant Zones
DTA Digital Terrain Analysis
DO Dissolved Oxygen
EC Electrical Conductivity
ERA Environmental Regional Authority
FDC Flow Duration Curve
GLUE Generalized Likelihood Uncertainty Estimation
GSA Global Sensitivity Analysis
IWRM Integrated Water Resources Management
ITCZ Inter-Tropical Confluence Zone
LSA Local Sensitivity Analysis
MCB Magdalena-Cauca Basin
MSE Mean Square Error
MMV Middle Magdalena Valley
MESD Ministry of Environment and Sustainable Development
NWS National Water Study
NSE Nash Sutcliffe Efficiency
O&G Oil and Gas
PP Pilot-points
PIA Price Index Analysis
pdf Probability Density Function
RMSE Root Mean Square Error
SVD Singular Value Decomposition
TWI Topographic Wetness Index
TSS Total Suspend Solids
WQI Water Quality Index
1. General Introduction

The current dynamics of the human civilization generates a constant rise in demand for
water. This increase is due to the increment of: population, city sizes, industrialization,
energy needs, and human consumption. The world will face a 40% water shortage if the
current water usage is kept (Sordo-Ward et al., 2019). This world water crisis will be a
matter of management and will not be related to resource availability (Vieira, 2009), but,
taking into account the upraise in needs for water from different uses the latter will boost
the possibility of localized conflicts and will drive more difficult choices regarding water
resource allocation. This situation will also limit the growth of crucial economic sectors for
development. In this context, water–food–energy relation will pose major political decisions
even though each sector is managed separately, and will imply concessions around allocation
and prioritization of water resources (Ki-moon et al., 2014; Schwabe et al., 2017). The social,
political, economic and legal connotations in water management are undoubtedly important,
as it allows the understanding of the lifestyle, traditions, and cultural development of the
population.

Water resources have to be managed according to laws and plans that regulate the water
allocation to different stakeholders including all productive sectors. According to its
economic nature, water is conceived as a limited value asset that economic sectors use
to create goods (Young, 1996; Gunawardena et al., 2018). In an economically efficient
allocation, the maximum amount of money a user is willing to pay for water use must
be equal for all sectors for the sake of maximizing social well-being. If this does not occur,
society will benefit allocating water in the sector where payments the higher are (Kirshen
et al., 2018; De O. Torres et al., 2016). The objectives of equitable systems refer to justice
in water allocation within different economic sectors. However, it is possible that these
objectives are not consistent with objectives related to efficiency (Cai, 2008).

Water management as an economic asset is an important strategy to achieve efficient and

equitable profitability and to favor the protection and preservation of water (Groves et al.,
2008; Räsänen et al., 2017; Wu et al., 2016; Cobourn et al., 2017). Thus, the economic theory
provides tools for decision making through the increase or decrease price analysis. These
tools are useful for water allocation in systems with multiple stakeholders, giving a base for
 2 1 General Introduction

comprehension and performance in water planning (Valdés-Pineda et al., 2014; Casadevall,

2016).

The relationship between hydrologic supply (surface water and groundwater) and economic
analysis of water resources, produce new approaches in administration, management and
planning for water resources, through the use of hydro-economic models that allow optimizing
profits and reduce operating costs of water management (De O. Torres et al., 2016;
Lopez-Nicolas et al., 2018; Ward, 2009). For instance, hydrological modelling in a regional
scale provides valuable information for adequate management of water resources, even more
in a high-pressure anthropic context (Neverre & Dumas, 2015; Barthel & Banzhaf, 2016).
The main reason to consider the regional scale is to explore the social well being of a region
(resource allocation between stakeholders) from the perspective of political accountability
(Brouwer & Hofkes, 2008; Hanemann, 2006). Nowadays, authors are adopting regional-scale
models, mainly to guarantee the quality of information that they use reducing the amount
of collected data (Reynaud & Leenhardt, 2008). The main limitation in regional models is a
high level of aggregation, because of resource homogeneity not being a solid assumption in
real-world environments, particularly when water is considered (Meyer et al., 2018). Thus,
regional model construction is important, since it seeks to analyze patterns and trends
of hydrological, hydrogeological and economic variables, and understanding interactions
amount them (Fu et al., 2017; Mukundan et al., 2019).

Additionally, when the value of water is included in hydrological modeling, economic

focalization in water resources management is completed (Rosenberg et al., 2008). Thereby,
the integration of economic aspects is built as a tool to ease the construction of
a hydro-economic model. Álvarez Mendiola (2010) identifies two approaches for the
model construction: (i) a modular approach that establishes a connection between the
hydrogeological and economical model, and (ii) a holistic approach which contains in a
single unit both models. The latter focuses on finding a proper technique that allows the
representation of these components in a significant manner. An application of this technique
is the water commercialization strategies that Lund & Israel (1995) presented. Booker
et al. (2012) also introduced applications of market analysis and water banks, enabling
the balance between water offer and demand to decrease the effect of drought. Water
markets are also susceptible to market failure, especially due to the presence of natural
monopoly and public goods that compete with private demand, as showed by Young (1996).
Market failures could be reduced trough the insertion of water rights and the structure
of appropriated incentives, as showed by Spulber (1988). Viability of water management
strategies depends on legal, institutional, environmental and economic implications. An
insufficient organizational context and lack of a solid economical foundation are the main
causes of failure in water resources management projects (Loucks et al., 2005).
 3

Likewise, integration of hydrologic and economical aspects is increasingly more relevant in

the design of policies for management and administration of water resources through the
definition of usages and rates that transmit to the user a sign of the resource real value, and
the cost associated to their supply (Lopez-Nicolas et al., 2018; Pulido-Velazquez et al., 2008;
George et al., 2011). Moreover, this integration will allow local and national authorities to
define strategies to reduce the gap that exists between water demand and the available offer.
These strategies can be applied in all stakeholders (mining, agriculture, livestock, services,
industrial sectors, and new Oil and Gas (O&G) exploration and exploitation projects),
in order to define the cost of associated impacts to water quality, and to determine the
water marginal price depending on its usage (Pedro-Monzonís et al., 2016; Pérez-Blanco &
Gutiérrez-Martín, 2017).

In Colombia, water concession is the right to a limited use of water, granted to develop
a productive activity. This concession must be related to water availability to ensure the
preservation and efficient use of water. The Environmental Regional Authority (ERA) has
the responsibility to apply rates defined by the Ministry of Environment and Sustainable
Development (MESD) in these concessions. However, modifying concessions volume
requires considering supply and water availability, hydrological regime (surface water and
groundwater), use and water allocation, and land use. These concessions are governed by
policies that determine the allocation, rates and water prioritization analyzing the water
supply. However, these policies have no differences between economic activities and conflicts
that are generated by their use. Given the current interest in economic development of
MMV system and the expectation of new O&G exploration and exploitation projects, it
is necessary to evaluate the supply and water allocation between different stakeholders,
due to its use in the economic activities, which can limit supply and water availability
(surface water or groundwater). This limitation could be due to water volumes required
and water quality associated with these economic activities. For instance, in 2012, 55.8
M m3 of water were collected from 217 water sources for O&G projects (Valencia, 2014).
Water management, from its shortage or abundance, may constitute a conflict focus between
stakeholders, demonstrating needs for improvement in the Integrated Water Resources
Management (IWRM), and in the analysis of the territory occupation through an economic
and environmental dynamic that allows to evaluate water resource vulnerability among
different stakeholders.

This situation does not show the current reality regarding IWRM in Colombia, so it requires
tools for decision making by analyzing hydrological regime (surface water and groundwater)
in areas with lacking reliable information on water availability and quality (Agencia
Nacional de Hidrocarburos - ANH, 2012). Consequently, the main goal of this research
is to provide a methodological approximation that enables the integration of hydrological,
hydrogeological, and economic aspects in water allocation between different stakeholders
 4 1 General Introduction

to establish management strategies at a regional scale. Thus, three specific objectives are
proposed: (i) to characterize the water system (surface water and groundwater) to quantify
the water supply and recharge, (ii) to develop an optimization model that permits to define
the water resource allocation based in the usage, cost and quality and (iii) to evaluate the
model performance in the MMV to determine water resources management strategies.

To approach the challenges proposed in the research objectives, three main activities were
developed and implemented in the MMV zone (described in Chapter 2): (i) hydrological
characterization and recharge determination, (ii) geological reinterpretation and hydraulic
conductivity estimation in different geological units and, (iii) establishment of a relationship
among the water availability, quality, and demand in economic terms.

The first activity is developed in Chapter 3, where hydrological zoning is proposed through
the usage of a semi-distributed model. The picking of this model was motivated by the
simplicity in the inclusion of hydrometeorological variables for temporal analysis of an
area. The output used in the model implementation was analyzed with global sensitivity
techniques, starting from statistical moments analysis, local analysis and uncertainty. This
characterization allowed them to determine the water offer and identify the recharge zones,
making a complete evaluation of dynamic interaction between the hydrologic cycle and
climatic conditions. In summary, this activity helped to improve the comprehension of:
(i) the surface flow regimen within the MMV zone, (ii) identification of likely physical limits
for the model parameters and (iii) validation of a new metric for the hydrologic model
calibration in tropical zones. This project has auxiliary data in its development shown in
Arenas-Bautista et al. (2018b, 2017).

The second activity focused on the hydraulic conductivity estimation, which was made using
two approaches: (i) analysis of hydrogeological units as constant zones and (ii) determination
of hydrogeological units that vary spatially and temporarily, analyzed trough the distribution
of pilot-points (PP) in the zone. Thus, it was necessary to collect the geologic structural,
seismic information, and determine the geological units present in the zone to consolidate
the geological model. With this information, it was interpreted the geologic zones and a
three-dimensional model was created, constituted by blocks of similar characteristics. The
lack of observation points in the hydrogeological model allowed to explore the validity of
the PP technique to assure the soil heterogeneity representation in a numeric model with
real information, showing a satisfactory behavior. Chapter 4, details the methodology and
results that highlight an advancement in the knowledge of: (i) streamlines and path lines,
(ii) surface water and groundwater interaction, (iii) determination of biased parameters
and validation of PP usage in real scale models, and (iv) assertive restriction in highly
parametrized models with field data in tropical zones. The results in this chapter have been
presented in Donado et al. (2018) and published in Arenas-Bautista et al. (2020).
 5

Consolidation of these models, allowed to perform the integration of economic aspects.

For that, it was developed a hydro-economic optimization model, as a utility tool in the
decision making processes, to allocate water between different stakeholders. The regional
optimization model is described in Chapter 5. It integrates multiple water offers (surface
water and groundwater) and multiple demands. The model considered the inclusion of a
water quality parameter and an objective function has established penalties in a way that
the model contemplates the possibility that not all the water demand can be fulfilled. The
differed analysis in four scenarios that vary temporally, allowed to widen the knowledge
in: (i) water availability affectation caused by its quality, (ii) the cost-benefit relation
of interaction between multiple users and different requirements, and (iii) management
strategies consolidation that could provide an effective analysis for efficient price policy
determination in tropical basins.

Finally, the management strategies results are consolidated in Chapter 6, providing an

effective analysis for determination of efficient policies in tropical basins. The document ends
with the principal conclusions of the research and discussion of some possible guidelines for
future research work.
2. The Middle Magdalena Valley
(Geological System)

In this section, a general description of the case study is presented. Then, the hydrological
process is described using variables like precipitation, temperature and evapotranspiration.
The hydrogeological processes considered in the model development were introduced from
the general geological description in the area. All this information and the processes it
describes will help to build the hydrological and hydrogeological models, which will be shown
in the next two chapters. This chapter is based on Arenas-Bautista et al. (2020).

The MMV system is located in the central part of the Magdalena-Cauca Basin (MCB).
This is the most important basin in Colombia in demographic, social and economic terms
(Arboleda-Obando, 2018). The approximate MCB area is 257 000 km2 and represents a
complex terrain by its topographic and climatological characteristics. This is mainly due to
the presence of the Andes mountain range and the interaction with the Pacific and Atlantic
Oceans, making it difficult to analyze the environmental conditions (Ideam, 2014). About
80% of the country’s population is concentrated in this area, where 95% of thermoelectric
energy and 75% of national hydroelectric energy is produced. In national terms, the MCB
concentrates 80% of the nation’s economic output (according to Gross Domestic Product
calculations by IDEAM (2019)).
2.1 Study Area 7

The MMV system is defined as an intramontane basin with an extension about 34 000 km2 ,
which separates the Eastern and the Central mountain ranges in northern Colombia. This
area extends longitudinally from South to North between the Ibague fault and the Girardot
belt, which in turn separate it from the Upper Magdalena Valley Basin. The system is
bounded to the South East by Bituima and the La Salina faults system; to the North by
the Espiritu Santo faults system; to the West by the onlap of Neogene sediments against
the base of San Lucas and the Central mountain range, and finally to North East by the
Bucaramanga - the Santa Marta faults system (Ingrain, 2012).

The MMV system is a geographical axis in the middle section of the Magdalena River
in a stretch of 386 km, which supplies different economic sectors making it a central
axis for the country’s development. Several departments (national administrative states)
interact in this region, whose relationship strengthens activities of mining, agriculture,
domestic, livestock, aquaculture, services, industrial, construction and O&G exploration
and exploitation processes. This economic development allows contributions to road
construction, environmental development programs and job creation. However, the most
critical factors in the area are water use in agriculture, hydropower, O&G projects and
domestic supply compared to water offer so that knowledge of spatial and temporal
distribution from its sources is a necessity to achieve an optimal management and planning of
the consumptive use. In this zone, the population is about one million inhabitants distributed
in forty-five municipalities (DANE, 2016). The MMV landscape is a tropical rain forest but
currently, there are few remains of its, due to the strong anthropic intervention that brought
the agricultural and livestock expansion.

2.1. Study Area

The characterization of the study area is presented in this chapter, as a basis for analyzing
the hydrological behavior of the system and understanding the impact that the physical
environment’s transformation generates at the social, cultural and economic level in the
region. The study area is located in the southern part of the MMV system (Figure 2-1) and
represents 17 000 km2 . This area is located geomorphologically along the central part of
the Magdalena River valley, between Eastern and Central Andes mountain ranges (Ingrain,
2012). The Magdalena River extends approximately 170 km across the study area. The
region is abundant in natural resources that includes: quartz, marble, gold, O&G, water,
flora, and fauna (Ingrain, 2012).
 8 2 The Middle Magdalena Valley (Geological System)

Figure 2-1.: Location Study Area in MCB (Red), MMV system (Yellow) and Study Area
(black).

2.2. Hydrology Context

The MMV starts in Honda (Tolima) and extends to El Banco (Magdalena) at is shown in
Figure 2-2, a municipality located at an altitude of 33 m.a.s.l. In this system, Magdalena
river presents a length of 542 km with an average slope of 0.35 m km−1 and a drainage
area of 105 850 km2 . In this area, a large number of wetlands begin to form due to fluvial
dynamics and the flat area geoforms. These wetlands exert a regulatory effect and behave
as affluents or effluents, depending on the river water level.
2.2 Hydrology Context 9

Figure 2-2.: Water Flow Diagram of the Magdalena River in the study area.

In the study area, most abundant sources of water include the Magdalena River and its major
affluents (Cimitarra, Sogamoso, Opon, Carare, Nare and San Bartolome Rivers). The area is
composed of two morphological parts: (i) the alluvial, and (i) the mountains. In the former
zone there exists a system of wetlands connected to the rivers, creating a wetland ecosystem.
Rainfall in this area is distributed across two wet periods: march-june and october-december;
the remaining periods being typically dry. The average rainfall amount is about 2 000 mm
yr−1 and the Magdalena River is characterized by an annual average flow of 2 361 m3 s−1
(with high and low flows of Q5% 4 298 m3 s−1 and Q95% 1 578 m3 s−1 ; Q5% correspond
to a discharge exceed the 5% of time, and so, characterizes high flows, meanwhile Q95%
correspond to a discharge exceed the 95% of time, and so, represents low flows). The annual
average temperature is above 24◦ C throughout the territory and elevations are between 50
and 3 700 m.a.s.l.

The Magdalena River in the study area is a semi-meandering river with curves controlled by
a rock outcrop, it means the location of the curves depends on the geological controls. The
geological control sites are stable points and in the curves where competent rock does not
appear there are points of instability.
 10 2 The Middle Magdalena Valley (Geological System)

This research used the regionalization of monthly precipitation using criteria at the
macroclimate level (Inter-tropical Convergence Zone - ITCZ). The amount of monthly
precipitation at a local level is determined by the cloud systems associated with the
local circulation of each slope and, in turn, is conditioned by the altitude, the mountain
ranges location and the convective activity of each locality. Daily rainfall information
in the MMV area basin has been acquired from the Institute of Hydrology, Meteorology
and Environmental Studies (IDEAM). A total of thirty-seven monitoring stations were
considered, as depicted in Figure 2-3. The data covers a period of time between 2000 and
2012. Missing data was completed via linear interpolation from the three closest stations.
The average precipitation in the area was estimated by the Thiessen method (Ruelland et al.,
2008; Wagner et al., 2012).

Figure 2-3.: Data used in hydrological model development. Left- DEM. Right- Precipitation
-P and Temperature -T stations and Thiessen polygons.

Discharge data was also obtained from the IDEAM database, through the selection of three
limnimetric stations. The selection of stations considered two at the basin entrance and one
2.2 Hydrology Context 11

at the basin exit, in order to obtain the flow in the area. The input and output stations were
the Puerto Berrio and the San Pablo, respectively (Figure 2-4). These stations were selected
because they were used as a reference in the National Water Study (NWS) by Ideam (2014).

Figure 2-4.: Limnimetric stations used in the hydrological model calibration.

Daily temperature information in the study area has been acquired from IDEAM (Figure
2-3). A total of sixteen monitoring stations were considered. The data covers a period
of time between 2000 and 2012. Missing data was completed via linear interpolation from
the three closest stations. The average temperature in the area was estimated by Thiessen
method, correcting the polygon value by its elevation (Zhang et al., 2016; Ayantobo et al.,
2017).

Evapotranspiration was evaluated through the Hargreaves method (Hargreaves & Samani,
1982; Hargreaves & Allen, 2003). This parameter is based on the estimate of solar radiation
(Rs ) via assessing the daily air temperature range (TR = Tmax − Tmin ) and Ra to estimate
Rs , as Eq. 2-1

Rs = KRS Ra T R0.5 , (2-1)

 12 2 The Middle Magdalena Valley (Geological System)

here, Tmax and Tmin being the daily maximum and minimum air temperatures (◦ C),
respectively; KRS is an empirical coefficient fitted to Rs /Ra versus T R data with a value
of 0.16; and Ra is extraterrestrial radiation (mm day −1 ) and was calculated according to
(Allen et al., 1998). Rs is the solar radiation in equivalent water evaporation (given in mm
day −1 since 1 mm day −1 = 2.45 M J m−2 day −1 due to the relationship between latent heat
of vaporization, water evaporated and the energy received by water) (Allen et al., 1998).
Evaluation of ETo is then performed through Eq. 2-2

ETo = CRa (T mean + 17.8)T R0.5 , (2-2)

here, ETo is the potential evapotranspiration (mm d−1 ), C is the original empirical constant
proposed by H. Hargreaves & A. Samani (1985), with value of 0.0023 and Tmean is the
observed daily mean air temperature (◦ C). Mean daily air temperature of day i, measured
at the gauge j, is computed as the arithmetic mean of measured values from a dry-bulb
thermometer at 7, 13, and 19 (or 18) hours (local time, GMT-5).

P, Q and ETo data between years 2000 to 2008 were employed for model calibration.
Meanwhile, for the validation process, observations between years 2009 to 2012 (Figure
2-5).

Figure 2-5.: Time series used in the sensibility analysis and the calibration-validation
processes in MMV. (A)- Daily ETo obtained from Hargreaves equation using
temperature data. (B)- Daily P obtained from rainfall gauges. (C)- Daily river
flows obtained from gauging stations.
2.3 Geology Context 13

2.3. Geology Context

Traditionally, surface water has been the main source of supply in Colombia. However, in the
last decade, groundwater has become more important because it is an alternative source due
to its better quality, the relatively low cost of handling and its influence on the industrial
and economic development of the Country (Ideam, 2014). Colombia has possibilities of
using groundwater in 74% of its extension. However, 56% of this area corresponds to
geographic regions with high surface water yields and low percentage of the settled population
(Vargas Martínez et al., 2013). The MMV has 106 131 km2 of surface area with resources
and groundwater reserves. These reserves are equivalent to 12.5% of the total area covered
by hydrogeological basins in the country. This potential has led to the hydrogeological study
of the area, through the development of conceptual and mathematical models to understand
its operation (IDEAM, 2019).

In Colombia, sixteen hydrogeological provinces have been identified. The hydrogeological

provinces correspond to larger units referred to smaller scales (1:1 000 000 to 1:500 000),
defined on the basis of tectonostratigraphic units separated from each other by regional
structural features. In the MMV, the hydrogeological province has a main aquifer system
(Figure 2-6). The MMV Aquifer System is composed of six aquifers: (i) Aquifers of
Magdalena River terrace, (ii) Aquifers of the alluvial deposit of the Magdalena River, (iii)
Mesa Formation, (iv) Real group, (v) Luna Formation, and (vi) Tablazo and Rosablanca
aquifers.

These features coincide with the boundaries of larger geological basins and, from the
hydrogeological point of view, they correspond to impervious barriers represented by regional
and structural faults. Additionally, they are characterized by their geomorphological
homogeneity (Vargas, 2006). Within each province, hydrogeological units are identified by
characteristics of porosity and permeability, which have different storativity conditions and
allow the groundwater flow.
14 2 The Middle Magdalena Valley (Geological System)

Figure 2-6.: Hydrogeological Provinces of the Magdalena Cauca Basin and the MMV.
3. Hydrological Model

This section presents the consolidation of the hydrological model. This model allows
us to make a comprehensive assessment of the interaction between the dynamics of the
hydrological cycle and climate conditions in the study area. Throughout this chapter, you
will find the hydrological simulation and the application of the sensitivity and uncertainty
analyses to assess the influence of the main parameters in estimating surface runoff. This
chapter is based on Arenas-Bautista et al. (2018a,b, 2017).

TopModel (TOPography based hydrological MODEL) as a tool for the hydrological

simulation in the south of MMV is presented. Additionally, it shows the application of
the sensitivity and uncertainty analyses to assess the influence of the main parameters
associated with TopModel in estimating surface runoff. The model is focused on indices
for variance-based analysis. This methodology is conducive to a Global Sensitivity Analysis
(GSA), according to the statistical moment’s characterization of the output simulations. It
does so upon relying on (a) the Sobol indices, associated with a classical decomposition
of variance and (b) the recently developed indices estimating the delivery of each model
parameter to the mean, variance, skewness, and kurtosis of the outputs. The analysis uses a
collection of 150 000 model simulations, each spanning over a 12-years temporal. These are
constructed by assuming those model parameters are random and associated with a uniform
distribution within a range, whose values are supported by literature studies and preliminary
model calibration against available data.
 16 3 Hydrological Model

3.1. Introduction

In the last decade, mathematical models are getting an important role for solving problems
in water resources management (Reynaud & Leenhardt, 2008; Dunne, 1983; Reynaud &
Leenhardt, 2008; Bossa et al., 2012), giving rise to discussions about their use and application
to evaluate and analyze complex hydrological systems (Bellin2016).

Hydrological models have been developed to understand different processes, which must
be evaluated for different environmental conditions (Jenson, 1991; Loosvelt et al., 2014;
Hollanda et al., 2015). However, they are generally focused on fulfilling two main objectives:
(i) to improve the understanding of the hydrological phenomena in the basins and how the
changes generated in them affect the hydrological phenomena and, (ii) the generation of
synthetic sequences of hydrological data for the design of infrastructure or for its use in
forecasting (Refsgaard, 1997; Kauffeldt et al., 2016; Ibarra-Zavaleta et al., 2017).

These models are classified into three types (Sieber et al., 2005; Hughes, 2016): (i) Empirical:
where the solution is based on empirical parameters, calculated by identifying statistically
significant relationships between certain variables; (ii) Theoretical: that are described by
differential equations and follow the laws of physical and chemical processes; and (iii)
Conceptual: which are simplified representations of physical processes, in mathematical
terms to simulate complex processes based on key parameters that describe the important
functions of the system of interest.

In this research, it was implemented a rainfall-runoff model known as TopModel (Lamb et al.,
1998; Beven & Freer, 2001b,a). TopModel is usually applied to assess the management of
water resources at the regional scale, using conceptual models for detailed assessment of
surface flow (Beven & Freer, 2001b; Mockler et al., 2016; Teng et al., 2017). Likewise,
distributed and semi-distributed models (which do not simulate the basin as a group, but
as a set of divided parts) are necessary for the simulation of spatial patterns of hydrological
response within a basin (Mazzoleni et al., 2015; Ibarra-Zavaleta et al., 2017). Moreover,
hydrological models also provide valuable information to study changes in land use or climate
(Karlsson et al., 2016). Thus, changes in land use are directly related to water supply (and
therefore to hydrological modeling), mainly related to human consumption, food production,
and power generation. These activities have become a global priority in the economic and
social sphere due to the growth of the population and the need to establish economic activities
for the communities’ empowerment (Buytaert, 2011; Crespo et al., 2012; Harou et al., 2009).
Through models, it has been possible to represent dominant hydrological processes in the
hydrological cycle of a particular ecosystem, mainly by calculating the water balances which
allow exploring the validity of the representation, interactions and various levels of model
behavior (Buytaert, 2011).
3.2 Methodology 17

The more hydrological information have the model (precipitation, temperature, evaporation,
evapotranspiration, and flow), the results will be better. However, the uncertainty due to
scarcity of these data is quite common in several areas of Colombia, due to the complexity
of the topography and to the low gauges density in some areas (Meerveld & Weiler, 2008).
In addition, it is necessary to generate tools for calibration and validation of synthetic data
to the hydrological modeling process (Blanco-Gutiérrez et al., 2013; da Silva et al., 2015).
The calibration and validation of data through modeling, has enabled the verification of the
assumptions that underlie the hydrological model used, which has contributed significantly
to the generation of new knowledge (Ibarra-Zavaleta et al., 2017).

Sensitivity analyses allow the identification of key parameters that affect model performance.
Also, these analyses show the parameters’ importance and its role on model structure,
calibration, optimization and quantification for uncertainty, helping to improve the flow
prediction or simulation processes (Yasari et al., 2013; Song et al., 2015; Li & Zhang,
2017). Depending on the uncertainty analysis perspective, global sensitivity methods can be
independent of statistical moments, or based on the model variance decomposition. The aim
of GSA is to determine the contributions of each individual parameter or the combinations
of their values that generate the variance (Kellner et al., 2015; Ballinas-González et al., 2016;
Borgonovo et al., 2017). To exemplify this scenario, Hou et al. (2015), have used sensitivity
analyses to identify important and unimportant parameters in simulated processes in the
Huaihe River basin, China.

3.2. Methodology

This section describes the field sampling, the statistical and graphical analysis proposed for
the hydrological database.

3.2.1. Hydrological Model: TopModel

TopModel is a semi-distributed model based on similarities of the topography which are

expressed through a Topographic Wetness Index (T W I). It can be considered as a
rainfall-runoff conceptual model based on the landscape characteristics which contains three
dynamic storage regions: (i) a root zone, (ii) an unsaturated zone and (iii) a fully saturated
zone (Hollanda et al., 2015; Metcalfe et al., 2017). Figure 3-1 depicts the workflow upon
which TopModel relies and its main conceptual elements. Main assumptions at the basis of
the modeling suite include:

1. processes in the saturated zone are described through a steady-state scenario,

 18 3 Hydrological Model

2. model assumes that the soil properties and hydrologic parameters are homogeneous in
the catchment model, parameters are uniformly distributed in space,
3. hydraulic gradient driving fully saturated groundwater flow can be approximated by
the topographic gradient (at the scale investigated),
4. hydraulic conductivity decreases exponentially with depth

Figure 3-1.: TopModel model description (Xue et al., 2018). Srmax is the maximum root
zone deficit, Srz is the root zone deficit, Suz is the local water storage in the
unsaturated zone, P is the precipitation per unit width, and qv , qb , and qs are
infiltration, base flow and saturated surface flow respectively. Source: Jeziorska
& Niedzielski (2015)

The model theory assumes that the local hydraulic gradient is equal to the local surface
slope and implies that all points with the same value of the T W I have the same hydraulic
properties (Andersen et al., 2001; Mukherjee et al., 2013; Yi et al., 2017; Jeziorska &
Niedzielski, 2015). Its value is computed from the basin topography using Eq. 3-1.
3.2 Methodology 19

ai
T W I = ln , (3-1)
tan βi

here, ai and tan βi are the upslope contributing area, i.e., the area that can potentially
contribute to discharge to the point of interest, and the average slope of the surface of the
i − th pixel, respectively (Suliman et al., 2016).

The process described herein uses a type of digital terrain analysis (DT A) resulting in a T W I
that quantifies topographic controls of basic hydrological processes (Yi et al., 2017; Jeziorska
& Niedzielski, 2015). T W I is derived through interactions of fine-scale landform coupled to
the up-gradient contributing land surface area according to the following relationship: the
cells with similar hydrological characteristics were generated from the grouping of pixels into
different categories based on the topographic index function (Figure 3-2) (Dai et al., 2017).
For the consolidation of the TWI, it was necessary to export the DEM in ASCII format to
TopModel using the Sp, Raster and Topidx packages (Metcalfe et al., 2015).

Figure 3-2.: Schematic diagram of morphology types used to compile T W I. Adapted from
Zimmerman (2016).

The runoff generation in TopModel is given by the relation between topography and
transmissivity of basin (Dewandel et al., 2017; Jeziorska & Niedzielski, 2015; Xue et al.,
2018). Rain infiltrates the root zone until its maximum storage capacity is reached, and
then it can be reduced to a linear velocity by the actual evapotranspiration of the surface,
described by Eq. 3-2.
 20 3 Hydrological Model

Srz
Ea = ETp (1 − ), (3-2)
Srmax

where, Ea is the real evapotranspiration, ETp is the potential evapotranspiration, Srmax is

the maximum root zone deficit and Srz is the root zone deficit.

After the infiltration process in the root zone ends, the excess water fills the unsaturated zone
and recharges the saturated zone generating a decrease in the water table depth. According
to the first assumption of TopModel, the depth of the local water table is represented by the
local storage deficit (D), which can be calculated for each T W I class using Eq. 3-3.

ai
Di = D + m(α − ln ), (3-3)
tan βi

where, Di is the catchment average water table depth, D is the average T W I and m is a
scaling parameter.

The water table is equal to the ground level when D reaches zero. Therefore, the average
value of the T W I for which Di constitutes the threshold for the maximum storage capacity.
Moreover, each point having higher value of T W I is considered to be in a saturation
condition. Additional rain on saturated surfaces cannot infiltrate the soil, therefore, excess
water is transferred directly to saturated surface runoff.

The water storage deficit is reduced by the recharge water flux from the unsaturated zone
to the groundwater, and its rate can be calculated using Eq. 3-4

Su z
qv = , (3-4)
Dl Td

where, Dl is the local storage deficit, and Td is the mean residence time in Su z.

Therefore, the total recharge rate (qv ) is expressed (Eq. 3-5) as the sum of all values of qv
multiplied by the upslope area ai representing a set of hydrologically homogenous points,
associated with topographic index class of the i − th location (Metcalfe et al., 2015).

X
Qv = qvi ai , (3-5)
i=1

where, Qv is the total flux and (qvi ) is the flux of water entering the water table locally (per
unit area).
3.2 Methodology 21

The base flow (Qbase ) is represented as the subsurface saturated zone flux qs , and can be
defined by Eq. 3-6

−D
Qbase = qs = Qo ( m
)
, (3-6)

where, Qo is the hydrological flux for the entire catchment area when D = 0. Both surface
flow qs and base flow Qbase account for the total discharge (Van Der Heijden & Haberlandt,
2015).

The transmissivity (T ) in the low zones (Eq. 3-7), according to the fourth assumption
of TopModel, decreases with the depth following the negative exponential law versus the
saturation deficit (D) being m a recession parameter (Ahmed Suliman et al., 2014).

−D
T = To ( m
)
, (3-7)

where, To is the local saturated transmissivity.

3.2.2. Global Sensitivity Analysis

A major goal of a GSA is to increase our understanding of the behavior of the model
considered (Anderton et al., 2002; Borgonovo et al., 2017; Gupta & Razavi, 2018). In this
context, the sensitivity of a desired model output is diagnosed to given model input(s) across
the entire parameter space, i.e., globally (Peña-Haro et al., 2009; Pianosi et al., 2016). As
such, a GSA approach enables us to (i) naturally account for model input uncertainties that
are typically encountered in hydrological models and (ii) identify sets of parameters that have
the largest influence on the output of the model (Mishra & Lilhare, 2016; Khorashadi Zadeh
et al., 2017). Our study relies on two approaches to GSA: (i) the variance-based Sobol
Indices (Sobol, 2001); and (ii) the Moment-based AMA indices proposed by Dell’Oca et al.
(2017). Sobol Indices are tied to one of the most widespread GSA approaches and rely on
the variance as a key descriptor of uncertainty. The AMA Indices allow characterizing global
sensitivity in terms of various features of the probability density (pdf ) of the model output,
as rendered by its main statistical moments.

Sobol indices for variance-based GSA

Variance-based GSA approaches (Sobol, 1993, 2001; Sudret, 2008; Fajraoui et al., 2011;
Sochala & Le Maître, 2013; Wang et al., 2015) consider variance as the metric to quantify
the contribution of each uncertain parameter to the uncertainty of a target model [Link]
 22 3 Hydrological Model

can be shown (Sobol, 1993) that if the model response f (p) (p representing a vector of
N model parameters) belongs to the space of square integrable functions, then the total
variance, V [ f ], of f (p) can be expressed according to the decompositon in Eq. 3-8

N
X X
V [f] = V pi + Vpi + ... + Vp1 ...pN , (3-8)
i=1 1<i<j<=N

here, Vpi is the contribution to V [f ] due solely to the effect of parameter pi , and Vp1 ...PN is
its counterpart due to interaction of model parameters belonging to the subset p1 ...pS . The
Sobol’ indices, Spi and Sp1 ...pS are defined in Eq. 3-9

Vpi Vpi...pS
Spi = ; Sp1 ...pS = (3-9)
V [f] V [f]

quantifying the contribution of only pi and the joint effect of p1 ...pS on V [f ], respectively.
The total contribution of pi to V [f ] is quantified by the total Sobol’ index (Eq. 3-10)

Spi i = Spi +
X X
Spi ,...,pj + Spi ,pj ,pk + ... + Spi ,...,pN , (3-10)
j j,k

In this sense, the total Sobol’ index represents the relative expected (average) reduction of
process variance due to knowledge of (or conditioning on) a model parameter.

AMA Indices for moment-based GSA

The AMA indices (Dell’Oca et al., 2017) allow quantifying the expected variation of a given
statistical moment M [f ] of the pdf of f (p) due to conditioning on parameter values. These
are defined in Eq. 3-11 and Eq. 3-12

 
1 R

M [ f ] Γpi
|M [ f ] − M [ f |pi ] ρΓpi dpi if M [F ] 6= 0 
AM AMpi =  R (3-11)
Γ pi
|M [ f ] − M [ f |pi ] ρΓpi dpi if M [ F ] = 0 

 
1 R

M [ f ] Γp1 ...pS
|M [ f ] − M [ f | p1 ...pS ] ρΓp1 ...pS dp1 ...dpS if M [ F ] 6= 0 
AM AMp1 ...pS =
|M [ f ]| p1 ...pS ] ρΓp1 ...pS
R

Γpi ...pS dp1 ...dpS if M [ F ] = 0 
(3-12)

here, AM AMpi (3-11) and AM AMp1 ...pS (3-12) are the AMA indices associated with a given
statistical moment M and related to variations of only pi or considering the joint variation
3.2 Methodology 23

of p1 ...pS , respectively; ρΓpi is the marginal pdf of pi , ρΓp1 ...pS being the joint pdf of p1 ...pS ;
and M [ f |p1 ...pS ] indicates conditioning of the (statistical) moment M on known values of
parameters p1 ...pS . Note that AM AVpi , i.e., the AMA index related to the variance (M = V )
of f (p), coincides with the total Sobol’ index Spi only if the conditional variance, V [ f |pi ]
is smaller than (or equal to) its unconditional counterpart V [ f ] for all values of pi . If V [ f ]
can undertake values that are smaller or larger than V [ f |pi ] while varying pi , than AM AVpi
≥ Spi . Note also that, in this latter case, AM AVpi can be either smaller or larger than SpTi ,
depending on the relative impact of the interaction terms (Dell’Oca et al., 2017).

Calibration and Validation Processes

Model calibration relied upon the MCAT library (Pianosi et al., 2015). This is grounded
on the generalized likelihood uncertainty estimation (GLUE) approach (Kellner et al.,
2015; Ballinas-González et al., 2016; Simmons et al., 2017), from which it was evaluated
a probability density of model parameters conditional to available observations. Inputs to
the MCAT come from a Monte Carlo sampling in the parameter space, using a classical
uniform random sampling, where all parameter are simple from an uniform distribution,
without consideration of parameter interaction.

In this model, it was considered the Nash Sutcliffe Efficiency (N SE) as a calibration metric
to analyze the model performed (Eq. 3-13) (Jeong & Adamowski, 2016; Wu et al., 2017;
Dakhlaoui et al., 2017).

Pn
(Qobs − Qsim )2
N SE = 1 − Pi=1
n
, (3-13)
i=1 (Qobs − Q̃obs )
2

where, Qobs represents observed flows, Qsim shows the simulated flows and Q̃obs represent
the average of observed flows.

In addition, the percentage bias (P BIAS) is used as a complement in the analysis of the
performance of the model to measure the average tendency of the simulated values to be
larger or smaller than those observed (Wiant & Harner, 1979). The optimal value of the
P BIAS is 0, the positive values indicate an overestimation bias and the negative values
indicate a bias of underestimation of the model. The equation used for its calculation is
shown in Eq. 3-14.

Qsim − Qobs
P
P BIAS = 100 P . (3-14)
(Qobs )
 24 3 Hydrological Model

3.3. Results and Discussion

The TWI for the study area is presented in Figure 3-3. This map shows a correspondence
between the highest values of the index and the drainage network. Areas with TWI
values between 15 and 22.5 correspond to the location where the drainage network is
generated. According to this, the high values of the topographic index are related to areas
topographically convergent or smooth slopes which generate discharge flows (Beven & Freer,
2001a). These areas are characterized by low transmisivities, thus, the level of water table
reaching the surface (Tian et al., 2016).

Figure 3-3.: Topographic Wetness Index (TWI ) in the study area. Source: Arenas-Bautista
et al. (2018b).
3.3 Results and Discussion 25

3.3.1. Parameter estimation

TopModel uses, in addition to the T W I, ten parameters that represent the characteristics
of the basin through hydrology, soils, and location of the study area (Table 3-1). The
values of each of these parameters or tune using prealable information that is available for
the study area. In this work, the values of the model parameters were initially established
from the reported literature for MMV area by the hydrological and geological institutions of
Colombia (IDEAM and Colombian Geological Service - SGC) and they are shown in Table
3-1 (Ingrain, 2012; Ideam, 2014; Servicios Integrales Hidrogeológicos, 2015).

Table 3-1.: Value of TopModel parameters reported literature for MMV area.
Parameters Description Unit Value
qso Initial subsurface flow per unit area L 0-0.00005
2 −1
LnT e Log of the areal average of T L T -7 - 6
m Model parameter controlling the rate of decline of T in the soil profile [-] 0-3
Sr0 Initial root zone storage deficit L 0-3
Srmax Maximum root zone storage deficit L 0-3
Td Unsaturated zone delay time per unit storage deficit T L−1 20-250
V ch Channel flow outside the catchment T L−1 1200-10800
Vr Channel flow inside the catchment T L−1 50-2500
K0 Surface Conductivity T L−1 0-20
Cd Capilary drive L 0-5

In this context, the interval of variability of model parameters is expected to be greater. The
key purpose of the GSA in this setting is to (i) improve our understanding of the functioning
of the model, in terms of the relevance of each model parameter on the target model output,
and (ii) to identify the relative importance of model parameters to the desired output model
(Liu et al., 2016; Hutcheson & McAdams, 2010).

The drawbacks generated by the model in the calibration process are associated with the
uncertainty of the parameters because the different sets of possible parameter values can have
similar performance values (Loosvelt et al., 2014; Mazzoleni et al., 2015; Ballinas-González
et al., 2016). For this reason, a Monte Carlo analysis (He et al., 2012; Simmons et al., 2017)
has been used to estimate the best sets of parameters that generate a better performance.
For this purpose, a data sampling with uniform distribution based on the interval reported
in the Table 3-1 was used. The number of simulations varied between 10 000 and 35 000,
since making new iterations did not improve the model performance.

This analysis showed that the parameters m and Srmax are the most significant (Figure 3-4),
so the change in their values influences the performance of the model and the shape of the
 26 3 Hydrological Model

simulated hydrograph. The n parameter represents the change in the saturated hydraulic
conductivity with respect to depth. Small values of m imply quick flow and insignificant
subsurface runoff, while large values indicate that more rainfall can infiltrate the soil, thus
less water reaches the outlet via surface route (Sigdel et al., 2011). This parameter is related
to subsurface flow control and the deficit of local storage, which is important in the case of
percolation and recharge of an aquifer (Buytaert & Beven, 2011). Therefore, this parameter
has a significant effect on the calculation of the local storage deficit, contributing areas and
the shape of the curve in the hydrograph recession.

Figure 3-4.: Cumulative distribution function plot of the model parameters according to the
simulations performed for the 12 years of analysis. Source: Arenas-Bautista
et al. (2018b)

The value of the Srmax indicates the influence of evapotranspiration on the hydrological
3.3 Results and Discussion 27

behavior of the catchment. A low Srmax value allows less water to be stored in the root zone
and, hence, available for evapotranspiration which can lead to the increased runoff (Hollanda
et al., 2015).

3.3.2. Global Sensitivity Analysis

Sensitivity analysis was carried out by two methods: Sobol and AMA indices. In the Sobol
method, the calculation of the total and first-order indices presented a high computational
cost, due to a large number of operations performed by decomposing the variance of the
model for each of the parameters. AMA indices had a high computational cost too, due to
computation of dependant statistical moments across the parameters ranges. The analyses
were performed for 150 000 random parameter sets with a uniform distribution for the study
period (2000 - 2012).

Sobol results (Figure 3-5) showed that the parameters of lower sensitivity are those which
tend to zero during most of the modeling time and that correspond to the partial results
of M CAT . The parameter driving the transmissivity recession curve (associated with an
exponential decrease of saturated hydraulic conductivity with depth) -m, the maximum root
zone storage deficit -Srmax , and the initial subsurface flow- qso can be considered as the
most sensitive ones. For Srmax and m the analysis of first-order indices was not as critical as
when the total sensitivity index was analyzed. This suggested that the parameters interacted
strongly within each other.

When considering the statistical moments and the AMA indices, other features can be
assessed. The first observation reveal that the sensitivity of Q with respect to qso, m
and Srmax depends on the statistical moment of interest. It was analyzed the temporal
evaluation of the first four unconditional statistical moments of Q (Figure 3-6).

Figures 3-7-a and 3-8-a show that, at beginning of time period (year 2000), the mean and
the variance of Q, respectively, are more sensitive to the qso parameter. This parameter is
directly related to the initial conditions of the calibration period: if it is a wet period, the
value of qso should be lower, but it would increase if the dry period. However, the range
of values is associated with the physical conditions of the basin, mild slopes for our case
study. These areas in the basin generally correspond to valleys, which present a saturation
condition that causes a decrease in the amount of subsurface flow.
 28

Figure 3-5.: Time evolution of the Sobol index. The parameters vch, ko, Sro, td, vr and cd tend to zero and
may not be clearly seen in the Figure.
3 Hydrological Model
3.3 Results and Discussion 29

Figure 3-6.: Temporal evolution of the first four moments of Q. (a) expected value, (b)
variance, (c) skewness, and (d) kurtosis
.

The Figures 3-7-b and 3-8-b show the mean and variance of Q respectively. It is found
that in 2003, the model is more susceptible to Srmax and LnT e parameters. This indicates
the influence of evapotranspiration on the hydrological behavior of the catchment. The
sensitivity of these parameters can be justified by analyzing the variations in the values
of the Srmax parameter, because it represents a greater or lesser runoff in response to the
model. Now, if it is associated with a variation of the values of Srmax parameter, changes
could be observed in the saturated transmissivity, which represents delays in the runoff and
system variation response.

For the years 2007 (Figure 3-7-c and Figure 3-8-c) and 2012 (Figure 3-7-d and Figure
3-8-d), figures indicate that Q is also sensitive to m. This sensitivity is presented by the
change in the saturated hydraulic conductivity with respect to depth. That is, variations in
subsurface flow conditions and local storage deficit modify the response of the model in the
calculation of water recharge.

These analyzes are consistent with the results of the AM AE and AM AV indices (Figure
3-9-a and b, respectively), where for the year 2000 the AM AE and AM AV indices for the
qso parameter are greater than for the others parameters. For the other years, the AM AE
and AM AV indices for the qso parameter are less important than for others parameters.
The AM AV index is related to Sobol index, as evidenced by the behavior shown in the
Figure 3-9-b.
 30

Figure 3-7.: First moment (mean) of Q conditional to values of parameters: qso (blue lines), m (violet lines), Srmax (gold
lines) and LnT e (green lines) for the years: 2000 (a), 2003 (b), 2007 (c) and 2012 (d). The corresponding
unconditional moments (black curves) are also depicted. Intervals of variation of parameters has been rescaled
between zero and one for graphical representations.
3 Hydrological Model
 3.3 Results and Discussion

Figure 3-8.: Second moment (Variance) of Q conditional to values of parameters: qso (blue lines), m (violet lines),
Srmax (gold lines) and LnT e (green lines) for the years: 2000 (a), 2003 (b), 2007 (c) and 2012 (d). The
corresponding unconditional moments (black curves) are also depicted. Intervals of variation of parameters
has been rescaled between zero and one for graphical representation.
31
 32 3 Hydrological Model

Skewness and kurtosis present a similar behavior to that shown by qso, m, LnT e and Srmax
parameters in median and variance. On 2003, the conditioning to the Srmax parameter
makes the probability distribution function of flow to become more symmetric and tailed
than the unconditional. For years 2007 and 2012 the conditioning is given by the Srmax and
m parameters, and corresponds to the behavior shown in the AM Aγ and AM Ak indices in
Figures 3-9-c and d.

Figure 3-9.: Time evolution of the global sensitivity index. (a) AM AE, (b) AM AV , (c)
AM Aγ (d) AM Ak of log(qso), log(LnT e), log(m) and log(Srmax ).

3.3.3. Model Performance

The model was implemented for the period 2000 - 2012, with a time step of 24 hours.
The data between 2000 and 2008 were used for the calibration process. In this step, the
characteristics of the basin were adjusted and the assumptions of the model were defined,
consolidating the set of initial parameters that adequately describe the hydrological behavior
of the study area. The data from 2009 to 2012 was used for the validation of the calibrated
model.

The results from calibration (in San Pablo Station - Figure 2-4) with GSA (Figure 3-10),
generated a set of parameters with efficiency of 0.74 for the objective function and 2.6
3.3 Results and Discussion 33

for P BIAS. The differences in the efficiencies of the best parameter sets were low, so the
overall performance of all the simulations can be considered similar. This result is associated
with the sensitivity analysis of the parameters and their impact on the representation of
hydrological processes in the basin.

Figure 3-10.: Observed Flows (gray), simulated flows (red) and Precipitation -P (blue) used
in the calibration process in the San Pablo Station.

This set of parameters (Table 3-2) was used to calculate the simulated flows by TopModel
during the period studied, presenting an under estimation of the maximum flows. For the
process of the model validation (in San Pablo Station - Figure 2-4), results generated from
the calibrated parameters with higher N SE values gave efficiency values near to 0.73 and its
simulated flows are shown in the Figure 3-11. The N SE values found are close to the values
obtained for a proven models in Ecuador, Colombia, Poland, Nepal and China with efficiency
values near to 0.7 (Sigdel et al., 2011; Padrón et al., 2015; Gil Morales & Tobón Marín, 2016;
Jeziorska & Niedzielski, 2015).
 34 3 Hydrological Model

Table 3-2.: Value of TopModel parameters used for the validation process.
Parameters Description Unit Value
N SE [-] 0.74
P BIAS [-] 2.6
qso Initial subsurface flow per unit area L 0.000123
LnT e Log of the areal average of T L2 T −1 1.255
m Model parameter controlling the rate of decline of T in the soil profile [-] 0.217
Sr0 Initial root zone storage deficit L 0.086
Srmax Maximum root zone storage deficit L 0.002
Td Unsaturated zone delay time per unit storage deficit T L−1 94.7
V ch Channel flow outside the catchment T L−1 1740
Vr Channel flow inside the catchment T L−1 1044
K0 Surface Conductivity T L−1 8.39
Cd Capilary drive L 4.22

Figure 3-11.: Observed flows (gray), simulated flows (red) and Precipitation -P (blue) used
in the validation process in the San Pablo Station.
3.3 Results and Discussion 35

However, it is possible that longer periods of time could help ensure better simulations
and adjustments, because when more data is added to model, the variability increases,
and so the parameter values are adjusted in concordance. Additionally, it is widely known
that TopModel has problems to accurately represent low flows during droughts (Hollanda
et al., 2015). For periods where precipitation exceeds evapotranspiration, the wide range
of parameters provide acceptable simulations for basin discharge, although base flow is less
accurately simulated, as it happen in other sites (Zhang et al., 2016).

The efficiencies were classified according to the methodology reported by Boskidis et al.
(2012). The N SE values found are close to the values obtained for a proven model in
Ecuador and Colombia with efficiency values near to 0.67 and 0.7 respectively (Padrón
et al., 2015; Gil Morales & Tobón Marín, 2016). For periods where precipitation exceeds
evapotranspiration, the wide range of parameters provide acceptable simulations for basin
discharge, although base flow is less accurately simulated, as it happen in other sites (Zhang
et al., 2016).

The initial determination of the dominant and sensitive parameters was performed with
M CAT through the cumulative distribution curves which were created from the division of
the Monte Carlo sampling results into a group of equal size, showing that the parameters
LnT e, m, qso and Srmax present a greater slope of the distribution curve (Figure 3-4), which
supports the importance premise of these four parameters in the hydrological response of
the basin (Bulot et al., 2016).

AMA sensitivity analysis outlines the importance of the parameters considered by M CAT
and Sobol, but Sobol index and AMA indices remark the need to concentrate the calibration
effort on some parameters, according to the modelling’s final goals. For example, if the main
goal is forecast, or retrospective time-series reconstruction, Srmax and m are the parameter
to take into account in the calibration process, but if the modelling is centered on lowering
the output uncertainty, LnT e could improve the model performance in some moments of
the time period. Finally, the assessment of extreme events will eventually need the tuning
of Srmax , but might make an insight on LnT e and m on some moments, specially on dry
periods.

Those remarks and guidelines could lead other calibration processes in tropical catchments
with similar behavior, saving computational resources and time in the processes, and allowing
to link the modelling framework to the catchment scale hydrologic processes observed in
nature.

In addition, the model performance has been analyzed through the flow duration curve (FDC
- Figure 3-12) between the observed and simulated data (Badjana et al., 2017).
 36 3 Hydrological Model

Figure 3-12.: Flow duration curve - FDC to analyze the flow variability in the study area.

These curves show a variability of 22% between 10% and 90% of the exceedance time, which
indicates that its variability is low and therefore it is related to groundwater storage processes
that dominate the flow of the stream and maintain a flow more stable in time (Salazar, 2016).
This corroborates the analysis carried out with respect to the variability of the parameters
and allows us to identify a coherent process between the behavior of the basin studied and
the performance of the model. In particular, it seems that the model fits well the high-flow
observations, but presents a slightly higher variance for the mid-flows, as observed in a line
with a higher slope; as for the low-flows, the observations is featuring a step at the end of
the FDC, indicating that the basin is not able to maintain the same conditions for baseflow
as for other flow sources, maybe due to anthropic or natural interferences on the percolation
fluxes, and simulated flows roughly follow these patterns, with a smoother curve, due to its
impossibility to represent these unknown interferences.

3.3.4. Recharge

The recharge (Figure 3-13) shows the spatial behavior of recharge in the study area, where
there are values of percolation ranging from 20 to 400 mm/month. The recharge generated
in the area comes mainly from the rain and the zone has a moderate to low potential to
recharge the aquifer. Note that for the model configuration, in the months with the highest
rainfall (April - May and October - November ) the spatial behavior of percolation is similar.
3.3 Results and Discussion 37

Figure 3-13.: Recharge data used in hydrogeological model. The maps show a monthly
distribution of the recharge value in the study area and are the inputs in the
hydrogeological model. This series was built for an average year, based on
results of hydrological model (Arenas-Bautista et al., 2017).
 38 3 Hydrological Model

3.3.5. Limitations and Conclusions

Models are used to represent real-life conditions. So, the major limitation of models, is that
they are a ’simplification’ of real conditions. Additionally, the model selection process should
be conditioned by the modeling objective; however, the available information is a limitation
that ends up prevailing. So, the available data and their quality are important because a
model is only as good, as the information used to build it. If we use simpler information
or wrong information, it will give us, the wrong results. And finally, the models should be
limited by a physical meaning, but, what happens when the information is not available to
compare? I should skip the physical behavior for the model precision?. In these cases, to
increase the parameter values range, or decrease the system simplifications, would take us
again at previous limitation.

The application of the hydrological model developed in this research contributes to national
efforts, and the availability of results helps the development of comparative studies in
the Middle Magdalena Valley at the basin scale. This work constitutes a sample of the
advantage of applying a widely used semi-distributed model that is freely accessible to the
scientific community, contrary to the limitations of using a model that depends on the
singularity of the study area. TopModel was able to reproduce the main pattern of the
hydrograph with acceptable accuracy for the case-study. A low performance to simulate some
patterns (baseflow) can be attributed to input data error, calibration inaccuracy, parameter
uncertainty and model structure. The most probable cause for those results is linked to the
uncertainty of the data series analyzed. Low accuracy of the model can also be an effect of
the model’s inability to represent distributed rainfall pattern. The results obtained will be
used as input data in the hydrogeological analysis of the area.

The sensitivity analysis allowed to establish the intervals of the parameters that offer a better
response of the model. Additionally, it allowed identifying which of these parameters govern
the behavior of the model. The sensitivity results obtained from the output of the model,
with respect to one parameter, is directly related to the global sensitivity index. This is
because the variability of a parameter modifies the statistical moments of the model.

The preliminary results of M CAT are related to those obtained in the Sobol and AMA index.
Sobol sensitivity analysis is a useful tool for hydrological models evaluation and thus allowed
to identify the most sensitive parameters of the TopModel model in the study area. The
AMA indices take into account the mean, variance, skewness and kurtosis of the probability
density function of an output of the model. From these metrics, they quantify the variation
of the results with respect to the change of one or several input parameters of the model,
allowing to analyze in a simple way the characteristics of the probability distribution function
of the output of the model (i.e. the degree of dispersion, symmetry, and tailed).
4. Hydrogeological Model

This chapter presents the consolidation of the hydrogeological model. This model allows us
to identify and analyze the behavior of hydraulic conductivity in the study area. Throughout
this chapter, you will find the hydrogeological simulation and the comparison between
PP and CZ to properly identify any biased parameters and the relationship between the
hydraulic properties heterogeneity and the PP numbers. This chapter is based on Donado
et al. (2018).

In this chapter, the analysis of a regional groundwater flow model is presented. This model
implements two types of analysis: i pilot-points (PP) and ii constant zones (CZ ). The
former is analyzed with the comparison of three regularization methods: (i) Tikhonov, which
is proposed as a restricted optimization problem, (ii) singular value decomposition (SVD),
which defines a small group of “super parameters” given by the model that can be appraised
by non-linear regression of data, and (iii) SVD-assisted, which is based in the determination
of super parameters that can be associated as multiplicating eigen-vectors within a weighted
sensitivity matrix to obtain a minimization in the objective function. SVD-assisted technique
allows to properly identify any biased parameters and the relationship between the hydraulic
properties heterogeneity and the PP numbers. Additionally, the possibility to evaluate
the plausibility’s reviewed, analyzing if this can improve the identification of hydraulic
conductivity (K ) heterogeneity and if it would be sensitive to the number of PP. For this
purpose, a numerical variable density model is developed. This tool is limited to reinterpreted
data from real measurements. For the CZ technique, the initial parameters are assigned
according to each layer, that is considered constant for hydraulic parameter values. In
contrast to PP technique, the initial parameters are assigned according to interpolations
using in-situ point measurements. The established model shows a low correlation with the
observed state variable (hydraulic head), proving the importance of spatial heterogeneity.
The model is calibrated in order to establish K (as an anisotropic parameter that varies
spatially), the porosity (η) and the specific storativity capacity (Ss ) in the PP and CZ,
reducing a mean square error of state variable dependable on the observation points.
 40 4 Hydrogeological Model

4.1. Introduction

Two-dimensional and three-dimensional representations in hydrogeological modeling allow

us to exemplify static and dynamic conditions of hydrogeological systems in natural or
hypothetical scenarios and their relationships with surface water bodies inputs (Alberti
et al., 2018; Gogu et al., 2001). This type of modeling can be used to analyze the
effects of groundwater extraction, to evaluate irrigation strategies in order to establish
an appropriate correspondence with aquifers, or to simulate different water management
scenarios (White, 2018). In this context, a calibrated model is a significant tool for water
resources management.

The calibration process must be used to acquire reliable modeling results (Wu et al., 2017). A
good model calibration minimizes the residual data between the observed and simulated data
and reduces the uncertainty of the hydrodynamic parameters in the area. This procedure was
developed using the inverse method. The inverse methods can solve problems: (i) linear and
(ii) non-linear. The method application with non-linear parameter estimation techniques
(Carrera et al., 2010; Wang et al., 2017b; White et al., 2016; Usman et al., 2018) is common,
given the swiftness that this method shows to determine the best adjustment parameters,
applying low subjectivity in the calibration process (Sun et al., 2015; Usman et al., 2018;
Zhou et al., 2014). Despite this advantage, the method is limited to the satisfactory sampling
of field data (Llopis-Albert et al., 2016; Pool et al., 2015; Zhang et al., 2016). Additionally,
hydrogeological models have a high spatial-temporal variability rate that causes non-linearity
which also increases the correspondence within different model input parameters (Gaganis
& Smith, 2006; Marchant & Bloomfield, 2018). An undesirable effect in this situation is the
appearance of unrealistic parameter distributions that causes the calibration process to fall to
a local minimum without exploring other posibilities (Hu & Chan, 2015; Klaas et al., 2017).
Bearing in mind that the calibration process does not guarantee the total model reliability
and that the results obtained are as real as the veracity in the assumptions used in the
conceptual model (Betancur et al., 2009; Kpegli et al., 2018; Linde et al., 2015), then, it is
appropriate to analyze the sensitivity and uncertainty associated with the parameters. The
uncertainty in the hydrogeological models is related to: (i) inaccurate measurement of input
data, (ii) inadequate simplification of the represented system and (iii) calibration process
(Carrera et al., 1993; Chen et al., 2018; Meeks et al., 2017). In particular, the calibration
process of any hydrogeological model is based on some method of spatial characterization or
zone focused parameters (Dewandel et al., 2012; Benoit et al., 2018).

The model was divided into different hydrogeological units based mainly on geological
properties (Custodio et al., 2016; Islam et al., 2017; Wang et al., 2017a). This subdivision
leads to suppose hydraulic uniformity properties in each zone (Cherry et al., 2004). This
simplification means that the hydraulic parameters values at any location are weighted
4.1 Introduction 41

averages of the actual values measured at different points within an area (Hassane &
Ackerer, 2017; Irsa & Zhang, 2012; Zhang et al., 2014). The sensitivity analysis allows
to address uncertainties identifying the most important model input parameters (Karay &
Hajnal, 2015). Additionally, this analysis provides an overall system understanding, reduces
uncertainties and improves the calibration and validation processes (Carrera et al., 1993;
Ehtiat et al., 2015; Sanchez-León et al., 2016).

The use of these CZ for allocation the hydraulic parameters’ values, lead to unnecessary
uncertainties in the modeling and, therefore, produce erroneous parameter values and high
heterogeneity in the geological units (Linde et al., 2017; Tiedeman & Green, 2013; Woodrow
et al., 2016). Having this into consideration, the PP technique has been proposed to improve
the spatial variability interpretation of parameters, in cases of groundwater flow modeling
that cannot be obtained by working with CZ (Christensen & Doherty, 2008; Jiménez et al.,
2016; Le Ravalec-Dupin, 2010; Usman et al., 2018). This technique interpolates the spatially
correlated hydraulic properties values of a set of points distributed along the model domain
(Ma & Jafarpour, 2017); generating an uniform distribution of parameter values with less
uncertainty. However, the use of a larger number of PP or the area subdivision in large
spaces could result in extensive modeling times (Christensen & Doherty, 2008; White et al.,
2016).

The application of this technique has been widely used throughout many related issues
as shown (Alcolea et al., 2006a, 2008; Christensen & Doherty, 2008; Hernandez et al.,
2003; Jung et al., 2011; Le Ravalec-Dupin & Roggero, 2012; Le Ravalec-Dupin, 2010; Ma &
Jafarpour, 2018; Panzeri et al., 2012). Nonetheless, the use of this technique can generate
over parameterization, situation that would lead to the optimization problem instability
(Amini et al., 2009). The instability can imply infinite or very large ranges where the
variations of the parameters are established, large variations in the parameter’s value,
non-existent correlations due to the small amount of data, and large second derivatives
of the hydraulic properties (Riva et al., 2011; Tóth et al., 2016; Yeh, 2015). The use of
variograms to correlate the initial value of hydraulic properties and limit their search ranges,
can help to improve their stability (Friedel & Iwashita, 2013; Jardani et al., 2012; Kashyap &
Vakkalagadda, 2009; Sheikholeslami et al., 2017). Although the biggest issue with application
of PP, is associated with the number of parameters to be calibrated, which is actually the
definition of PP parameters itself, where each of those parameters is another constraint.
Determination of PP number within the model can be defined in a fixed way associating
a spatial location or a random variation during the optimization process (Christensen &
Doherty, 2008; Jiménez et al., 2016; Sun et al., 2011). The usage of a reduced PP value
can improve model stability, although this approach could generate a larger homogeneity
of hydraulic properties, approaching it to CZ technique and making the problem sensitive
to the point locations (Ma & Jafarpour, 2017). However, Alcolea et al. (2006a) established
 42 4 Hydrogeological Model

that when implementing PP, and regularization processes, it is recommended to implement

as many number of PPs as computational viability is allowed. This implementation permits
to obtain a conductivity field that is as close as possible to the true field.

4.2. Methodology

4.2.1. Governing Equations and Modeling

Mathematical equations describing groundwater flow through a porous medium are based
mainly on Darcy’s law (Jiao & Zhang, 2016) (Eq. 4-1) and the continuity equation.
According to Darcy’s law, the average flow velocity is proportional to the hydraulic gradient
(Freeze and Cherry 1979) and the effective porosity.

V = −K∇h, (4-1)

where, V is the Darcy velocity which is defined as the flow per unit cross-sectional area of
the porous medium, K is the hydraulic conductivity, and h is the hydraulic head.

Substituting the Darcy equation in the Continuity equation is obtained flow equation
(Eq.4-2).

∂h
∇.(K∇h) ± W = Ss , (4-2)
∂t

where, Ss is the specific storativity, W represents the source/sink term, and t is the time.

In the mass balance, the rate of change in storage of volume overtime must be equal to the
input rate minus the mass output rate. Under steady-state conditions there is no change in
the hydraulic head, so time is variable dependent. The steady-state flow can be described
by the Eq. 4-3.

∇.(K∇h) ± W = 0, (4-3)

In a homogeneous and isotropic medium (Eq. 4-3) can be expressed by the Eq. 4-4

∇2 h ± W = 0, (4-4)
4.2 Methodology 43

The solution h(x, y, z, t) describes the value of the hydraulic head at any point in the system
at time t. The flow in 3D transient state in a porous, heterogeneous and anisotropic medium
is described by this equation combined with the initial and boundary conditions. Thus, the
three-dimensional aquifer is conceptualized in layers, which are discretized in nodes within
the model (Anderson and Woessner, 1992). The flow between nodes is defined by storativity
and transmissivity through T = Kb, where, b is the saturated thickness of that aquifer.

Given this, the equation that governs the flow of groundwater in the each layer is expressed
by Eq. 4-5.

! !
∂h ∂h ∂h ∂h ∂h
Tx + Ty =S + W, (4-5)
∂x ∂x ∂y ∂y ∂t

The z dimension is not considered in this layer, and the storativity term is expressed as
the S = Ss × b. When the aquifer is confined, the saturated thickness and transmissivity
remain constant. Additionally, the saturated thickness may vary over time. To calculate this
thickness it is necessary to know the elevation of the bottom of the aquifer and the hydraulic
head (Eq. 4-6).

! !
∂h ∂h ∂h ∂h ∂h
Kx + Ky = Sy + W, (4-6)
∂x ∂x ∂y ∂y ∂t

here, the specific yield (Sy ) that represents the unconfined equivalent of the hydrogeologic
system.

Finally, when the groundwater system has been characterized, the numerical model is
developed. The methodological development used in this model is presented in Figure 4-1.

4.2.2. Boundary Conditions

The conditions of the hydrogeological limits can be explained through the following
mathematical conditions (Sahoo & Jha, 2017; Tóth et al., 2016).
• 1st type: Dirichlet Condition. The main assumption for this condition is that,
regardless of the flow of water within the domain, there is no influence on the potential
of the outside water body, so that this potential remains constant (Eq. 4-7).

h(xi , t) = hR
l (t), Γl ∈ t(0, ∝). (4-7)

Here, hR
l is the prescribed limit values of the hydraulic head, and Γ is the given limit.
 44 4 Hydrogeological Model

Figure 4-1.: Hydrogeological Modeling Methodology. The thick blue arrow indicates inputs.
The thin blue arrow indicates an iterative process for the model’s optimization.
The red arrow indicates the methodology steps.

• 2nd type: Neuman Condition. This condition states that regardless of the state and
the flow of water within the domain and at the limit, the normal flow is set by external
conditions and is still determined by the state of the limit (Carrera & Neuman, 1986;
Tóth et al., 2016) (Eq. 4-8).

∂h
qnh (xi , t) = qhR (t) = −Ki,j ni , Γ2 ∈ t(0, ∝). (4-8)
∂xj

Here, qnh represents the Darcy flux of the fluid, qhR is the normal boundary fluid flux,
and ni is the normal unit vector.
• 3rd type: Cauchy Condition. In this condition, the flow through the limit depends on
the magnitude of the difference in the head through the limit, with the head on one
side of the limit entered into the model and the head on the other side calculated by
the model (Eq. 4-9).

qnh (xi , t) = Φh (hR

2 − h), Γ3 ∈ t(0, ∝). (4-9)

Here, the Φh is the fluid transfer coefficients.

• 4th type: Pumping/Injection Condition. In this condition it is established that the
4.2 Methodology 45

aquifer is subjected to extraction or injection processes, which in turn cause changes

in the hydraulic head, and can be represented mathematically as shown in Eq. 4-10.

Qw Qw {δ(xi − xm ∀(xi , xm
X Y
p (xi , t) = m i )}, f or i ) ∈ Ω. (4-10)
m i

Here, Qw w
p represents the well function, Qm is the pumping or injection rate of well m,
and xmi is the location of well m.

4.2.3. Regularization and Parametrization

In order to understand the regularization methods proposed in this chapter, an introduction

to the conditioned probability computation over a parameter space whose probability
distribution function (pdf ) is unknown is presented. This approach focuses on finding an
optimum function from an eligible set of pdf ’s that represents the best approximation of the
data set. The goodness-of-fit of the approach is measured through the M SE, which implies
the approximation being a correlation as accurate as possible between the input and output
data, represented by the interpolated function.

However, in this research, parameters as the hydraulic conductivity were not well represented
when the spatial analysis was carried out due to the few field measurements. This
misrepresentation is due to field data errors during the measurement process, to clustering
on the field data location, and to an important data scarcity across the area, with zones
presenting a few measurements, and others with no information. Given this, the use of
minimization functions such as M SE does not eliminate the noise generated by measurement
errors. As for the location clustering and small or null amount of information, a function
which is robust enough and permits generalization is an advantage that allows to predict
information across the area, when no data is available.

The regularization process in hydrogeological models applies some constraints in the function
to reduce the error. The soft constraint is the most used to regularize a solution due to is
limited with the same parameter values to obtain models with fewer failures. Another
possibility is the use of sparsity constraints, which forces the function to simplify the number
of parameter combinations around a set of representative data in the model’s domain.

The methodology presented in this chapter for parameterization of a regional groundwater

flow model consists in the comparison of three regularizations.
1. Tikhonov: guides the solution of the inverse problem towards parameter estimates
that approximate the minimum error variance. In this research, the inverse problem
 46 4 Hydrogeological Model

was formulated as a restricted optimization problem in which an adequate parameter

value is sought, defined by the field measurements of hydraulic parameters and which is
subject to restrictions of the objective function. When looking for these values, without
exceeding the model constraint, it is sought to minimize the objective function for the
parameter and reduce the predictive error. In this modeling a soft constraint was used,
assuming that the PPs that are close to each other must have similar values. The soil
heterogeneity level was controlled by the corresponding semi-variogram. However,
due to the lack of information in some areas, it was not possible to correlate the
hydraulic parameters in areas with ranges greater than 2000 m, so the soft constraint
was combined with the sparsity constraint. In this constraint, the a-priori information
relates the PPs to their initial value, which leads the model to find value closed to the
initial value of the assigned PP.
2. Singular Descomposition Value - SVD: solves the inverse problem by removing
parameters’ combinations. This method is more flexible than Tikhonov since is
not necessary to estimate combinations of parameters which are not relevant for the
calibration process. The method works by subdividing the parameter space into two
orthogonal subspaces. One space is encompassed by combinations of parameters that
correspond to the solution and the other is the null calibration space, in which the
parameters of the inverse problem that cannot be estimated are omitted.
3. SVD-assisted: subspace methods provide numerical stability in the modeling process
by removing parameters or combining them in the calibration process (Aster et al.,
2005). As a result of the removal, the calibration process is no longer necessary
to estimate individual parameters or combinations of correlated parameters that
are indeterminate in the parameters’ space. These combinations are determined
automatically by the weighted Jacobian matrix which is calculated only once. Then,
the parameter space is broken down into the combinations of estimated parameters
(super-parameters) and the combinations of indeterminate parameters. The calibration
problem is reformulated so that it only solves in these super-parameters (Christensen
& Doherty, 2008; Tavakoli & Reynolds, 2009; Khaninezhad et al., 2012; Khaninezhad
& Jafarpour, 2018).
In general, the parametrization processes: (i) achieves uniqueness of a fundamentally
nonunique calibration problem (nonuniqueness is always the case for groundwater models),
and (ii) achieves a level of fit that stops overfitting. Overfitting a model describes the random
error in the data instead of the relationships between the variables. That is, overfitting
models have too many variables for observations number. Thus, the regression coefficients
represent noise rather than genuine relationships between the variables.

Each sample has its unique peculiarities, thus, a model that adapts to the random
peculiarities of one sample is unlikely to fit the random peculiarities of another sample.
4.2 Methodology 47

Therefore, overfitting a regression model reduces its generalization outside the data set.
In other words, an overfitting model describes the noise and is not applicable outside the
sample, which is not useful.

An important consideration is that the sample size limits the quantity and quality of the
conclusions that can be drawn about a population. The more you need to learn, the bigger
the sample should be. That is, as the number of observations per estimate decreases,
the estimates become more erratic. In addition, a new sample is unlikely to reproduce
inconsistent estimates produced by smaller sample sizes. Overfitting often occurs when a
high R2 is sought. When choosing a regularization model, the goal is to approximate the
true model for the entire population. If it achieves this goal, the model should fit most of
the random samples taken in an area.

The general parametrization for a hydraulic property Y (x, t) of a groundwater flux system
is defined by: (i) a deviation YD (x, t) representing the real values from n observation in Y ,
and (ii) a residual error (ε), corresponding to a model’s parameters lineal combination (Pj ).
The deviation YD (x, t) is represented through a model’s discretization using Kriging. This
discretization allows to consider field measurements of the property Y (y ∗ ) and geophysical
data (g ∗ ). For the lineal estimation, deviation is written in Eq. 4-11.

Dim
XZ
Y (x, t) = βiz (x)Z(xi , t), (4-11)
i=i

where, x is the deviation location, t and xi are time and measurements’ locations, and βiz
are the weights assigned in the discretization for every measurment, arranged in the vector
Z = (y ∗ , g ∗ ). The residual error’s paramterization express the required modification on
the deviation to preserve the dependent variable measurements. The residual error can be
written as the lineal combination of hydraulic properties’ values in the PP (see Eq. 4-12).

NPP

(µzj )Pj ,
X
ε= (4-12)
j=i

here, NP P is the number of PP, µzj are the weights assigned in the discretization of Pj .

4.2.4. Estimation and Objective Function

A priori estimation of PP values and the covariance matrix is done through conditional
estimation of measurements in vector Z. PP variance located near the measurement points
will be lesser. When using the SVD method, the values for super parameters are estimated
 48 4 Hydrogeological Model

iteratively, minimizing the total error used to quantify the deviation between observations
and simulations. In the development of this model, the identity matrix has been used as
the weighting matrix using the Monte Carlo method described in Alberti et al. (2018). The
regularization process helps to reduce the non-singularity in the estimation of parameters
(Mahmoud & Saleem, 1993; Alcolea et al., 2006a; Jafarpour, 2013; Carniato et al., 2015).
Tikhonov regularization involves a series of information based data equations to set the initial
parameters values (Christensen & Doherty, 2008). The calibration process with Tikhonov
is proposed as a restricted minimization process that decreases ϕ for a specific target set
by the user. To ensure that Tikhonov’s regularization process provides numerical stability
to the model, the limits for the model must be between sub-spaces S1 and S2 so that the
sensitivity matrix values are high. If this condition is not accomplished, ϕ is minimized
without taking consideration of the conditions specified by the user, adjusting the weights
to a-priori information.

The Objective Function (ϕ) is defined in order to optimize iteratively model parameters and
fit model output and observed values. The goodness-of-fit uses the ϕ written in Eq. 4-13.

ϕ = ϕr + γϕm , (4-13)

here, ϕr is the regularization of ϕ, ϕm is the minimization of the ϕ and γ is the Lagrange

multiplier.

To solve this problem, ϕr is minimized while ϕm is maintained at an upper limit selected by

the user. All the results generated below these limits allow us to consider that the model is
calibrated.
Formal definition of the ϕ (Medina et al., 2001; Carrera et al., 2005) in this research is
written in Eq 4-14.

PP X
n OP X
n
(Pti − Pti∗ )Vp−1 (Pti − Pti∗ ) + wh (hti − h∗ti )Vh−1 (hti − h∗ti ),
X X
ϕ = µj
i=1 t=1 i=1 t=1 (4-14)
Constraint Pjl ≤ Pj ≤ Pju ,

where, i is the number of observation points (OP), t is the observed time interval, h∗ is
the hydraulic head vector at the OP, h is the simulated hydraulic head vector, µ is the
associated weight, P is the parameter to estimate, p∗ is the prior estimate of PP values, V
is the covariance matrix, wh is the observations weights which has a value of one in our case,
l and u are the lower and upper parameters limits and j is a node of PP.
4.2 Methodology 49

In the Minimization of ϕ, was used GLMA algorithm (Gauss-Levenberg-Marquardt). This

search algorithm linearize state variables dependencies in model parameters, imposing at the
same time a change on pk in every iteration (Eq. 4-15) (Finsterle & Zhang, 2011; Yao &
Guo, 2014).

(H k + δ k I) 4 P k = −g k , (4-15)

here, H k represents an approximation of Hessian matrix of ϕ. g k is the gradient in the

model parameter’s vector in k iteration, I is the identity matrix and δ k is the Marquardt
parameter.

After each iteration, the model parameter’s vector is updated. Before consolidating this
step, it is checked if the vector parameters components by iteration meet the assigned upper
limit (Doherty, 2004). This process is iterative until: 1.) if i ≤ 0.0001, where i is the ϕ value
at the end of the optimization iteration i-th, 2.) when the number of iteration reaches 30, it
is set as the maximum number of optimization iterations, 3.) if ϕi −ϕ ϕi
min
≤ 0.1, where ϕmin
it is the lowest ϕ reached so far.

Although the only guarantee of minimizing the objective function is to obtain a value of cero,
it is also valid to analyze the convergence of the parameters to their optimum values. Thus,
number 1 has been selected as the convergence control for this model. This convergence
control was selected because allow to analyze the objective function itself behavior, showing
whether there is a reduction in several successive iterations. That is, if the iterations are
producing changes below the set value, there is probably little to gain by continuing with
the optimizer execution.

In models with lack of information, sensitivity analyzes based on a plausible range of values
or observations for calibration are the only way to restrict a parameter. It is proposed in
this model to adopt for the statistical analysis the methodology by Alcolea et al. (2006a)
that assess maximum likelihood equation’s minimization (Eq. 4-16) using different weight
values, in order to evaluate the model’s results.

ϕ X X
S = N + Ln(H) + N Ln( )− ni Ln(wi ) − ki Ln(µi ), (4-16)
N j j

here, N is the total number of data, ni and ki are the number of state variable and H is the
first order approximation of Hessian matrix of ϕ at the optimization process end.
 50 4 Hydrogeological Model

4.3. Model Development

4.3.1. Geological Model

MMV basin geology has been described by Gómez et al. (2003); Morales (1958);
Sarmiento-Rojas (2011). This geological basin represents a depression that contains
sedimentary formations deposited from the Cretaceous, emplaced in a Jurassic igneous
and metamorphic basement. The tectonic evolution of the basin has been proposed by
Cooper et al. (1995). The Jurassic crystalline basement is composed of intrusive igneous and
associated metamorphic rocks consisting for the most part of Segovia Batolite.

Cretaceous sedimentary rocks were formed in a rift system during the Mesozoic in
consequence to back-arc extension and marine transgression, in order to that in the
basin were deposited a sequence of mostly shales, limestones, and chert beds (Luna and
Umir Formations). Late Cretaceous–early Paleocene tectonic inversion from extensional to
compressive regime marks a significant change in depositional environments from marine to
continental in the incipient foreland basin (Lisama Formation).

Marine transgression occurred during the Oligocene epoch, during which the MMV area
was flooded. The sediments laid down in these waters are known as La Paz, Esmeraldas,
Mugrosa, and Colorado Formations, but except for their connate water, they are essentially of
continental facies (Morales, 1958). Changes in plate tectonic motion documented in the late
Oligocene to early Miocene causes by the reactivation of the middle Eocene structures created
an upper Miocene unconformity. During this deformation phase, the Eastern Cordillera was
uplifted and eroded, erosional deposits in MMV are lithic fluvial sands derived from the
Central Cordillera dominate (Real Group) (Gómez et al., 2005). Major deformation of the
Eastern Cordillera and Llanos Foothills began at approximately 10.5 millions of years (Late
Miocene) and resulted from Panama’s collision with Nazca plate (Cooper et al., 1995). The
Pliocene Mesa Group rests conformably upon the Real Group and is reportedly composed
of massive conglomerates, cross-bedded lithic sandstones, and mudstone layers; this unit is
575 m thick (Morales, 1958).

Result of this geological background, the MMV area is tilted towards the East, having a
monoclinal structure, disturbed by some folds and faults (Servicio Geológico Colombiano,
2014). At this basin, clastic sediments of alluvial type, and sedimentary rocks of Quaternary
and Tertiary age developed. The main characteristic of this material is the low consolidation
and sediments predominance such as sand and gravel, interspersed with fine-grained
materials such as clays and silt. In the MMV basin, groundwater is extracted from units
that function as an aquifer. These units are recent alluvial and terrace deposits that emerge
4.3 Model Development 51

in the Magdalena River proximity with an average productive thickness of 150 m (Gallego
et al., 2015), whose origin is associated with Magdalena river, the Mesa formation and Real
group, and unconsolidated sediments (sandstones and conglomerates).

The underlying rock represented at the East limit is used as a vertical and lateral no-slip
boundary condition for every aquifer formation, which is not continuous through the whole
domain. The following aspects were taken into account for the conceptual model integration:
(i) the relationship between each formation material, (ii) homogeneity of hydraulic properties
for regional scales, (iii) a resistivity analysis for each formation material, and (iv) common
permeability values. At a regional scale, the geological reinterpretation of the study area led
to a simpler model of hydrogeological basin. This model included depth of sedimentary
formations, the Salina-Buitima fault, and every main fold. In addition, interpolation
algorithms were used including secondary information from seismic and magnetotellurics
studies, superficial geological interpretations, and stratigraphic from drilling well. In the end,
seven layers resulted from this exercise, each one separated from the other by lithological
contacts or erosive surfaces. Near-surface formations are considered the most important
given the current water resource exploitation from those aquifers.

The model had been constructed by geological interpretation and analysis of spatial
distributions of the formations and include major fault Buitama-Salina fault zone. Owing
to confidentiality of information by oil companies, this work only uses public information.
Base information has been collected by previous studies such as surface geology, geological
maps scale 1:100 000 (Figure 4-2) that have been development by the Colombian Geological
Service (SGC), model area are cover by eight maps (Beltrán & Quintero, 2008; Beltrán
et al., 2014; Gavidia et al., 2008; Ward et al., 1977). Each of these maps is accompanied by
an interpreted geological cross-section (Figure 4-3). Balanced geological cross-sections have
been proposed by Moreno et al. (2013) in order to analyze the geometry of the Nuevo Mundo
Synclinal and The Opon oilfield. The Nuevo Mundo Syncline is located at north-east part
of study area. La Salina Fault bounding the western margin of the syncline and separates
the Cenozoic strata from Cretaceous outcrop of the Eastern cordillera. In this work, three
balanced sections of Moreno et al. (2013) are included in order to have subsurface information
on the Guayabito fault, the Guayabito syncline, and the Armas syncline.

Magnetotelluric study has been carried by (INGENIERIA GEOTEC, 2016), as a result

they proposed six geological-geophysics profile that were collected in the north part of the
study area, these interpreted profiles show the interaction between Dos Hermanos fault and
distribution of Quaternary deposits, Mesa formation and Real Group.
52 4 Hydrogeological Model

Figure 4-2.: Geological map, scale 1:500 000. Source: SGC & MINMINAS (2016)
 4.3 Model Development

Figure 4-3.: Simplified geological map and cross section available: 18 seismic lines, 17 interpreted and 6 magnetotelluric
profiles and example of data interpreted cross-section from geology map N. 119.
53
 54 4 Hydrogeological Model

Seismic surveys and stratigraphic well data were obtained by the Seismic Atlas of Colombia
(Cediel et al., 1998) and from the seismic interpretation on oilfields Lisama, Tesoro, Nutria
and Peroles (Ortiz et al., 2009). As a result 19 seismic lines were collected, in order to
integrate the seismic lines with cross-section and geological map, the time-depth conversion
was carried out using the velocity plot corresponding to the sonic log of 6 soundings in the
area (Romero et al., 2015). To estimate depth in seismic lines it was used the quadratic
function described in Eq. 4-17.

y = 0.00096v 2 + 1.9v + 41, (4-17)

where, y is the depth in meters and v is the velocity (m/s).

Seismic lines are useful for estimating geometrical, dips, major structures, discontinuities,
erosional surfaces and faults zones (Figure 4-4). In order to simplify the model, units were
combined that can represent homogeneous properties on a regional scale, these units are
divided by major surfaces such as unconformities and lithology. This division from younger to
older: Mesa formation, Real group, Chuspas Group, Lisama formation and Late Cretaceous,
Early Cretaceous, and Jurassic Basement. Seven layers were created, and the division of
rocks units correspond by lithological contacts or erosional surfaces, considering that shallow
formations have more hydrogeological importance due to its current exploitation. With the
intention of simplified model, faults were represented by folds using the dip separation and
true displacement of the strata.

4.3.2. Hydrogeological Conceptual Model

Figure 4-5 represents the conceptual application model area. The zone is limited to the East
and West by the limits of VMM aquifer system that acts as an impermeable boundary due
to its igneous-metamorphic nature (Ideam, 2014). The South and North limits of model act
as an inflow and outflow, respectively, in the different zone sections. The groundwater flow
was estimated by Darcy’s law (Karay & Hajnal, 2015; Şen, 2015) using piezometric data.
The elevation in the focus zone descends smoothly in south-north direction, conserving the
properties of an alluvial valley and, therefore, provoking a regional movement of groundwater
throughout this depression (Ingrain, 2012). Additionally, there is a register of low scale
domestic and industrial use of groundwater collection obtained with pumping equipment
(Gonzalez et al., 2010). Most recharge of groundwater occurs at the mountain ranges, from
rainfall infiltration, and at the low-lands, from channel drainage losses.
 4.3 Model Development

Figure 4-4.: Geological Model. The model had been constructed by geological interpretation and analysis of spatial distributions
of formations. Source: Arenas-Bautista et al. (2018c).
55
 56 4 Hydrogeological Model

The region presents high temperatures, especially during the dry season, which causes a loss
of water in the form of evapotranspiration (Asociacion Colombiana del Petroleo, 2008).

Figure 4-5.: Hydrogeological Conceptual Model. P is the precipitation, Q represents the

flow, and ETo is the evapotranspiration.

The geostatistical model was defined by the transmissivity variogram in the model domain.
The soil texture details for different depth profiles were available from well-logs. This
data was used to establish the initial values of different hydraulic parameters in each
hydrogeological zone, (Freeze & Cherry, 1979). The K values obtained vary between 10−5 to
35 m day −1 . It was defined 167 observation points based on piezometric levels (Figure 4-6)
because of values recorded in piezometers were not continuous in twelve years of analysis,
an average data level was used as a reference in the numerical model in steady-state. For
transient state, a measurement grouping by month was made. This grouping assumes that
levels in the same months of different years, exhibit a similar behavior due to results of the
4.3 Model Development 57

hydrological analysis. Finally, a series of monthly data was consolidated for an average year.
The extraction data comes from 78 independent pumping tests in the wells where information
was found within the model domain (Figure 4-6). Pumping rates range between 10−2 m3 /day
and 250 m3 /day (Annex 1).

Figure 4-6.: Locations of piezometric levels (red points), Limnimetric stations (blue
triangles), and Wells (green stars).
 58 4 Hydrogeological Model

The hydraulic properties of the aquifer system make possible to identify areas with primary
and secondary porosity and characterize them as units able to store and transmit water with
relative ease. The horizontal (x axis) permeability is an order of magnitude higher than the
vertical (Han & Cao, 2018; Xie et al., 2006). The system total porosity ranges between 25%
and 3%, with an average specific yield of around 14%. The test holes pumping results and
the lithological analysis allow us to conclude that the hydraulic conductivity varies from 20
m day −1 to 0.1 m day −1 .

4.3.3. Numerical Implementation

The objective of this chapter is to identify the variability of K in a real case regional
flow model with respect to the number of PP and analyze whether plausibility contributes
to the identification of heterogeneity in the model. The results are explored within the
conception of a regional flow model that presents high soil hydraulic and temporal-space
parameter variability, and hydrometeorological characteristics conditioned to the presence
of the Inter-Tropical Confluence Zone (ITCZ).

The information was consolidated on a spatial scale of 6 km x 6 km resulting from

hydrological modeling. The recharge is assigned as a material associated property, which
becomes variable in time for transient model. The loading boundary conditions (Figure 4-7)
were assigned as a first type variable Dirichlet condition in south and north edges from the
model domain, according to water level historical records. For the seven consolidated rivers
in the model domain a Cauchy boundary condition was applied. This requires the hydraulic
conductance of the riverbed material and its geometry procurement, which were obtained
establishing the initial conductivity from the pumping tests and the cross section geometry,
supplied by the limnmetric stations in the area. The East and West model limits were
conditioned by the application of Neumann boundary conditions; given they are considered
as a watershed. The inlet and outlet flow sections and the model domain flow were identified
using piezometric water level contours, and applying Darcy’s law. Finally, pumping wells
were assigned as a well edge condition where a defined extraction was applied to a node.
4.3 Model Development 59

Figure 4-7.: Boundary conditions used in the hydrogeological Conceptual Model. Source:
Donado et al. (2018)

For CZ application a pattern on materials disposition according to the hydrogeological units

was defined (Figure 4-8-A). In the PP technique, it was observed that it was not possible
to obtain a defined pattern for the same materials, so the initial values for PP were define
assuming that the hydraulic conductivity value resulting in the pump tests represents the
expected. To correlate these values, the semivariogram was used with an exponential model,
a range of 2000 m and a variance of 2 without nugget effect. The K values obtained vary from
10−5 to 35 m day −1 . Then, the information was spatialized via Kriging (Tsai & Yeh, 2011).
This technique was used given its additional use for regularization processes, which are based
on the same variogram used for regionalization (Tsai, 2006; Kashyap & Vakkalagadda, 2009;
Janetti et al., 2010; Riva et al., 2009). For the development of this model, each layer was
treated as a single zone and tested with a variable number of PP to evaluate the response
model. The principle of parsimony dictates that the numbers of parameters involved in an
estimation process must be kept to a minimum (Merritt et al., 2005; Tonkin et al., 2007;
Golmohammadi et al., 2015). However, when PP is used in conjunction with regularization,
this number tends to increase (Panzeri et al., 2012). In the regularization process, the
assignment by preferred value was used to restrict the variation of parameter value depending
 60 4 Hydrogeological Model

on the limits established by pump tests in the geological units. This method was combined
with the use of the covariance matrix to find a base that represented the data optimally.

Figure 4-8.: A-. Definition of geological units for CZ. B-. location of the PP according to
the number of points.

In the model development, 40 PP, 100 PP, and 200 PP were randomly generated with a
uniform distribution throughout the model domain (figure 4-8-B). The fundamental principle
of this constraint is to prefer homogenous distributions of parameters whose values show as
small a deviation as possible from those expected. In a geological context, a PP should
preferably have a similar value to its neighbors within a certain distance. This distance
and the strength of correlation are defined through a semi-variogram. In this research, due
to the study area does not have sufficient and reliable data throughout the domain, and
the available data are pooled in small areas of the study area, was decided to generate the
PPs with uniform distribution. The model ran 50 times, and in each run the PPs were
distributed so that their locations were different (x, y, z), but the space between them was
constant. It means that a rectangular pattern was used for every run, but the PP locations
changed after every run, always keeping the same pattern. The random location of PPs
with a uniform distribution allowed to represent without bias the locations that had no
field measurements in the calibration process and the post-calibration analysis identified the
parameter contributions to the uncertainty. To corroborate the calibration process results,
an uncertainty analysis consisting of two phases was performed. The first phase consisted
of a basic analysis where the model error was estimated computing the RMSE, the Pearson
coefficient and MAE. In the second phase, the goal was to analyze how the plausibility would
improve the identification of heterogeneity of K and, if it would be sensitive to the number
of PP, plausibility was evaluated by computing the S factor, obtained from µ values between
10−3 and 102 .

Additionally, nine experiments were performed, responding to regularization method changes

4.4 Results and Discussions 61

and the PPs numbers used in the initial model proposed. Then, to the best response to each
variation, six adjustments were made for changing the ponderation factor µ in order to
explore the plausibility term influence. The µ values oscillated between 10−3 and 102 , taking
into account that the optimal value for µ must be equal to one on an error-free geostatistical
model.

4.4. Results and Discussions

4.4.1. CZ and PP

The two models calibration (CZ and PP) is compared briefly in their value parameters, model
fit and predictions. This comparison focuses on model ability to simulate the real system and
to show conceptual differences. The model residuals respect to the observation points allows
us to infer that model calibration allows to adequately represent the assumptions presented
in the conceptual model, both in the CZ and in the PP technique (Figure 4-9-A).

The average residual differences in PP results is lower than 1 m. the CZ technique exhibits
oddment points between 1 and 3 m. The possible reason for these mayor differences, would
be associated with the large number of wells in this region, which pumps extensively the
groundwater. Therefore, the possibility of an occasional partial inflow across the study area
boundaries, especially on the eastern and western boundaries, could not be well represented
by the input/output flow limits conditions. The error obtained in the observation points is
summarized in Table 4-1.
 62

Figure 4-9.: The conceptual model results from the residual values of the numerical model. A-. Shows the model residuals
respect to observation points in the CZ and PP techniques. The red lines represents the mean. B-. Shows the
result of a graphical fitting of observed and simulated heads at the observation points with R2 = 0.93.
4 Hydrogeological Model
4.4 Results and Discussions 63

Table 4-1.: Description of the experiments evaluated and results of the statistical metrics.
Experiment Description ϕm ϕr φ M AE RM SE R2
CZ geological units 145.4 178.6 323.9 0.25 0.93 1
PP 40 PP 114.3 159.4 273.6 0.25 0.81 1

These results include R2 , M AE and RM SE. The value of R2 for the calibration model
is 0.99; likewise, M AE, which is an efficiency of calibration indicator, is 0.25. Figure
4-9-B allows to evidence optimal simulation results, as the observation values are adequately
represented under the numerical model assumptions, and the Figure 4-10 presents the results
of transient-state calibration. The hydrographs of simulated versus observed heads over time
show a relatively good match between observed and simulated heads, even if the model tends
to overestimate the piezometric levels. The general behavior of PP technique offers average
values of 142 m in the OF. However, the results obtained in CZ show a variation of 6 m
above the results with respect to values obtained from PP. These results are in agreement
with the findings of Alcolea et al. (2006a) where it is concluded that the PP technique offers
better results for the analysis of heterogeneity.

The hydraulic conductivity analysis (Figure 4-11) in the model shows that in the Quaternary,
the average Kx values in CZ is different from the average PP values. This occurs because
of the ignorance of hydraulic properties of outcrop geological units in the zone. The most
information and monitoring compiled for this model has an 80 m depth (for mesa formation
and real group); because of this, the model representation fits with higher certainty to reality.
In the seven model units remaining, the Kx values assignment does not represent a direct link
either with CZ nor PP. This is due to the ignorance of depths of hydraulic properties higher
than 600 m. Additionally, the scarce available information is focused on specific influence
areas, which gives another struggle for its representation. Finally, results show that in lower
model units there can be generated equifinality issues when the same model performance
with a different parameter for the same node is generated.
 64

Figure 4-10.: Result of observed and simulated levels in the model for transient state. The results are shown for four wells
(green stars) because these are the most representative in the area and have consolidated information over time.
These wells are: Felicidad, Coyote, SAM1.1 and M-5237.
4 Hydrogeological Model
4.4 Results and Discussions 65

Figure 4-11.: Result of K in the proposed hydrogeological model. A-. The model
development with PP. B-. The model development with CZ.

Additionally, is evident a strong separation of the Quaternary aquifer and Mesa formation
with values of 10 m day −1 and 2.5 m day −1 respectively. The PP calibration process presents
a K three times higher than the CZ model. In general, the K values for all hydrogeological
units are within the reported range in literature. Comparing the two methods evaluated
through the CZ and PP, they show similar model characteristics: (i) The east and west
regions of the model, corresponding to Jurassic and Early Cretaceous units, have a uniformly
low conductivity, which corroborates the conceptual model assumption establishing those
zones as water divisions, (ii) The model presents small areas of high conductivity in the
north and central part and, (iii) The highest conductivity is found in Quaternary and Mesa
formation. An interesting fact is that the estimated field begins to reflect some large known
scale geological structures also perceived on the CZ model. The storage term does not
significantly influence the hydraulic head during the one-day simulation period. Therefore,
it can be concluded that K is the only definitive parameter in hydraulic head modeling.

In summary, compared to the previous CZ technique, PP approach is not only able to provide
satisfactory matches for all measured heads and reproduce large-scale and local features of
the measured groundwater level, but it can also extract more information of heterogeneity at
several scales using the same amount of measured information. All the results shown above
indicate that PP approach provides a more realistic model to simulate river-groundwater
interactions based on parameters. Takin into account the parameters calibrated in both
techniques the ratio between the horizontal and vertical hydraulic conductivity is at the
center of its range with a value of around 0.1. Therefore, the vertical flow in the aquifer is
ten times slower than the horizontal flow that coincides with the regional dips (sub-horizontal
layers).
 66 4 Hydrogeological Model

4.4.2. Regularization Methods

The ϕm behavior was evaluated for simulations in a transient state (Figure 4-12) for the PP
set in each applied regularization. This plot shows that, in general, 200 PP technic offers a
better ϕm value with respect to 40 PP. The SVD method does not show improvement in ϕm
value reduction since stabilized after the third interaction. SVD general behavior in all PP
experiments is notably alike, showing average ϕm values of 115 m. The difference between
Tikhonov and SVD-assist method application is focused on the number of interactions for
ϕm stabilization. The Tikhonov method allows to make evident a lower ϕm value when 200
PP are used. These results are in accordance with the findings by Alcolea et al. (2006b),
where it is concluded that within a bigger number of PP better ϕm results will be obtained.
SVD-assist method analysis also reports a better model performance using 200 PP.

The “lever” for how well SVD does regularization is the number of singular values used
(equal to the number of “super parameters”). This is chosen implicitly using the FEPEST
EIGTHRESH variable. Therefore EIGTHRESH was probably set to 5x10−7 by FEPEST
(Doherty, 2003). This is suggested as the lowest possible value to stop over-fitting where
noise is strictly numerical (i.e. measurement noise is not taken into account). The model
does SVD only once and then defines super parameters on the basis of this. This is not
optimal because model nonlinearity may alter the optimal definition of super parameters as
parameter change; however, it is convenient. The error obtained in the observation points is
summarized in Table 4-2.

Table 4-2.: Description of the experiments evaluated and results of the statistical metrics.
Experiment Method ϕm ϕr φ S M AE RM SE
SVD 114.3 159.4 273.7 28.4 0.251 0.827
40 PP Tikhonov 109.9 104.9 214.8 27.52 0.189 0.658
SVD-assist 109.6 18.97 128.6 22.04 0.136 0.457
SVD 72.31 273.5 345.8 28.82 0.26 0.81
100 PP Tikhonov 62.95 234.7 297.6 28.27 0.209 0.613
SVD-assist 57.61 61.41 119 23.73 0.141 0.412
SVD 34.93 206.4 241.3 27.52 0.267 0.811
200 PP Tikhonov 28.35 150 178.4 26.39 0.196 0.587
SVD-assist 20.27 41.49 61.76 23.44 0.121 0.356
 4.4 Results and Discussions

Figure 4-12.: Model performance for regularization used under transient state conditions. This figure compares the number
of iterations needed in each regularization to achieve the best value of ϕr
67
 68 4 Hydrogeological Model

A further results review can be achieved through the boxplot analysis for each geological
unit (Figure 4-13). In the Quaternary, among PP models the range values are really
close whatsoever. The most information available for the present model has an 80 m
average depth (for Mesa formation and Real group), thus, the model representation fits with
higher certainty to reality in this depth. However, the influence of observed data spatial
configuration is unknown, and it could lead to different inverse problem results and model
performance, but this subject is out of the scope of this research. In the seven remaining
model units, the Ky values assignment does not represent a direct link with different PP
quantities. This is due to the ignorance of hydraulic properties at depths greater than 600
m. Additionally, the scarce available information is focused on specific influence areas, which
results in another struggle for its representation. Finally, the results show that in the deep
unit model equifinality problems can be generated when the same model performance is
generated with a different value of a parameter in the same node.

Following those reasons, it is not enough to evaluate ϕm to conclude which is the best
method. As an additional tool of evaluation, ϕr and S were also calculated (Table 4-3) for
the PP experiments. The lower value of ϕr for all the experiments is got by the SVD-assited
method, and the calculation of S corroborates this result. It is important to remark that
all the experiments perform similarly on the likelihood function minimization, which calls
into question the relationship between execution time and model’s performance in the model
with the least PP.

These results had already been reported in (Donado et al., 2018) when the Impact of the
PP number was evaluated (Figure 4-14). These characteristics correspond to the idea of
maintaining a low hydraulic conductivity in the geological units that were characterized as
a watershed. The model presents a limitation by not defining the limits of the geological
units where surface aquifers are present. However, the model represents high conductivities
in the central part of the study area. Therefore, it can be concluded from the representation
of K that this parameter is important in the modeling of hydraulic heads.
 4.4 Results and Discussions

Figure 4-13.: Box-plot for the geological units analysis in the model. The red color shows the distribution of
data in different geological units using a normal distribution with 40 PP. The blue color shows the
distribution of data in different geological units using a normal distribution with 100 PP and the
green color shows the distribution of data in different geological units using a normal distribution
with 200 PP. Parameters values correspond to the nodes parameters values.
69
 70

Figure 4-14.: Conductivity values estimates as of hydrogeological model applying SVD-assisted regularization methods.
4 Hydrogeological Model
4.4 Results and Discussions 71

4.4.3. Plausibility Effect

A sensitivity analysis of the PP model was performed with the variation of factor µ. Table
4-3 summarizes the total results concerning the identification of heterogeneity. The data
analysis shows the ponderation factor effect µ in the consolidation of ϕ and the likelihood
minimization. The results show that the use of lower values of µ generates constantly low
values of ϕ. Likewise, high values of µ provoke a high error in K estimation, i.e., the biggest
errors are generated for values of µ = 102 for any amount of PP, except for the 100 PP model,
where the largest error is generated by µ = 10−3 . Nevertheless, there were no restrictions
in the plausibility term, and thus the estimated values in PP locations vary extensively,
generating thicker solutions. Although this result does not constitute the optimum reply
awaited, authors as (Alcolea et al., 2006b; Zhang & Burbey, 2016) obtain similar estimations
and propose as solving methodology to vary the PP locations. Also, as expected, the use of
higher values of µ generates higher values of ϕ (Table 4-3). In this case, the results tends
to bring about a smooth representation, because it is biased and contains little information
about the real variability of the true field.

Table 4-3.: Results of the sensitivity analysis to the weighting factor µ and to the number
of pilot points.
PP µ ϕm ϕr φ S M AEr RM SEr M AEh RM SEh
0.001 149.2 88.8 149.3 21.3 0.57 1.51 0.24 0.945
0.01 149.5 18.3 149.6 17.8 0.61 1.69 0.24 0.946
0.1 116.5 55.4 122.1 21.2 0.614 1.71 0.27 0.835
40
1 109.6 18.9 128.6 22.1 0.133 0.697 0.268 0.811
10 151.3 10.1 252.27 24.17 0.04 0.17 0.295 0.952
100 217.5 285.9 28804 37.24 1.226 2.7 0.409 1.141
0.001 160.6 141.2 160.7 21.6 0.236 1.042 0.216 0.981
0.01 160.6 72.84 161.3 17.2 0.125 0.749 0.216 0.981
0.1 160.5 52.08 165.7 19.4 0.216 0.98 0.216 0.98
100
1 57.6 61.41 119.0 23.7 0.026 0.687 0.197 0.587
10 241.9 22.92 471.1 22.7 0.017 0.42 0.393 1.203
100 341.1 57.67 6108 35.5 0.088 0.66 0.365 1.429
0.001 146.3 243.7 146.5 15.42 0.276 1.082 0.174 0.936
0.01 63.5 102.9 16.5 0.087 0.553 0.084 0.783 0.783
0.1 85.1 117.0 21.9 0.034 0.64 0.102 0.806 0.806
200
1 41.5 61.76 23.4 0.053 0.447 0.121 0.357 0.357
10 93.9 1080.3 26.7 0.041 0.672 0.287 0.92 0.92
100 172.6 17426 40.1 0.196 0.911 0.35 1.008 1.008

The optimum results generated from the S criteria analysis, is obtained when µ = 10−2
(Table 4-3). The analysis show that the difference of S for values between 100 PP and 200
PP is minimal, and so could be related to the random generation of the PP. In addition, this
 72 4 Hydrogeological Model

results can also be due to the field transmissivity measurements location which were used
to build the initial interpolation. Singh et al. (2008); Pool et al. (2015); Khaninezhad et al.
(2018) demonstrated that a reduction on µ from its theoretical optimum leads to a robust
solution regarding the geostatistical model.

If µ is not adequately weighted in correspondence to the real system conditions, the

heterogeneity identification of the hydraulic properties fails, even reaching worse values than
estimations from numerical model interpolation. However, minimization of the maximum
likelihood function allows to identify if a weighted is adequate and to exclude the bias from
the number of PP used.

Regarding the optimum number of PP, the error decreases when PP increase. Figure 4-15-A
shows error and variance decreasing in hydraulic head measurements, corroborating that the
more PPs, the better the system representation. Figure 4-15-B shows error and variance of
measurements on the PPs of each model and it is evident there is no discernible differences
between the 100 PP and 200 PP models. Heterogeneity identification performance increases
with the PP number until 100 PP are reached but variations between 100 PP and 200 PP
are depreciable (Figure 4-16). Results from synthetic models (Alcolea et al., 2006b; Riva
et al., 2010; Yoon & McKenna, 2012; Carniato et al., 2015; Jiménez et al., 2016; Klaas et al.,
2017; Bao et al., 2018) also suggest that in the development of synthetic models a reduction
of the objective function is evidenced as the number of PP increases, leading to a better
relationship with the regularization data and heterogeneity identification. However, in this
model and under the real conditions of the application area, although there is a significant
change in the identification of heterogeneity in the use of 40 and 100 PP, the reduction of
the objective function is not significant between 100 and 200 PP, and has required more time
to generate the model solution.
4.4 Results and Discussions 73

Figure 4-15.: K estimation errors versus µ. (Left) mean absolute error. (Right) root mean
square error. A-. Shows the analysis for hydraulic head and, B-. Shows the
analysis for K values assigned to each PP in the model.
 74 4 Hydrogeological Model

Figure 4-16.: Qualitative comparison of spatial heterogeneity of K based on the plausibility

analysis. Each row shows the log − K for different PP numbers and each
column shows the result when varying µ. The middle column is field generated
from the field measurements and Kriging.
4.5 Conclusions 75

4.5. Conclusions

The characterization of complex groundwater systems tends to be interpreted with multiple

models, but they are affected by data scarcity and limited knowledge, adding uncertainty
to the model results. Therefore, the model uncertainty significantly affects the prediction
of the groundwater system behavior. Given this complexity and the direct link between
surface and groundwater flow, the real impact of groundwater extraction only can be
accurately measured using a regional flow model. An adequate geological characterization
and the proper extraction point representation and water injection, are essential for model
calibration, and to provide tools for water resources administration in the study area.

The use of a highly parameterized inverse model offers a guide to estimate heterogeneity
in the hydraulic conductivity as an anisotropic variable in non-confined aquifers. The
application of real data as appropriated aquifer initial parameters and the inverse model
restriction with parameters upper and lower limits reduces the number of necessary
interactions to minimize ϕ, improving potential estimations and estimating possible
convergence.

The results of inverse model calibration prove a better response in real system representation.
The error in the hydraulic head is reduced by 3% compared with the developed conceptual
model. Even though the state variable minimum square error is minimized in the observation
points, the model validation evidence coherent results in the validation points. This supports
the generated model solution, the model calibration through PP is more robust and flexible
in contrast con CZ parameterization given its lower subjectivity, and lower representation
of hydraulic properties heterogeneity.

Although the PP method reduces the homogeneity produced by CZ, it has some limitations,
e.i., to overlook other sources of model error and result in an over-parameterization or
equifinality. An important uncertainty of the PP method is the choice of the number
and location of the PPs, as well as the effect it has on reducing the variability of the
parameters in the model during the investment problem. Although there is no defined
guideline for generating these points, the results of this article indicate that better results
can be obtained using a uniform pattern, reducing the spaces between them which is useful
to correlate the hydraulic properties as a guide for the separation distance. In the same
way, the interpolated fields of K resulting from the application of the PP method function
properly for the estimation of the flow at the regional scale; however, it can lead to significant
limitations for other applications, such as solute transport modeling.

The use of a focused methodology in the parameters estimation through PP technique in

accordance with a highly parameterized inverse model offers a guide to estimate heterogeneity
 76 4 Hydrogeological Model

in hydraulic anisotrophic conductivity in non-confined aquifers. The application of real data

as appropriated aquifer initial parameters and the inverse model restriction with parameters
upper and lower limits reduces the number of necessary iterations to minimize ϕ.

The sensitivity analysis shows that K is the most influent hydraulic parameter in comparison
with η and Ss . Nevertheless, this sensitivity directly depends on location and the model
layers’ configuration: the sensitivity of different observation points highlight Quaternary
and Mesa unites as the ones with particular prevalence.

A modification has been proposed in the evaluation of the PP technique, which includes
a plausibility term in the optimization process. The suggested method was tested on a
real basin with in-situ data, and three elements regarding heterogeneity identification were
explores: (i) the plausibility term influence, (ii) sensibility to the number of pilot points
and, (iii) the effect that reducing the parameters variability had on the model during the
inversion problem.

The standard PP method does not includes the plausibility term, which leads to a better
adjustment of the reduction data but also to an unstable identification of model parameters as
well. This instability means higher parameter variations and can be qualitatively described as
rough field’s estimations. On the contrary, too much prevalence to the plausibility term bias
the solution toward an unreal system. If the geostatistical model contains less information
of the real variability patters, like this case, the estimated field will also fail on identifying
heterogeneities.
5. Integrated Water Resources
Optimization Model

This chapter presents the hydro-economic model consolidation. This model allows us to
maximize the value of water consumption in the study area. Throughout this chapter, you
will find the optimization model in a current mean condition, at thirty years, fifty years
and a hundred years.

Water resources management must be in all cases effective, efficient and sustainable,
especially when considering the effects of climate change and variability. Achieving this is a
challenge that is tackled in this chapter with a hydro-economic optimization model, which can
be used as a decision-making tool for water assignment between several users. The model
has been developed by integrating hydrological aspects (surface and groundwater) in an
economical optimization framework for water allocation and water quality management. The
model also has the purpose of maximizing the value for water consumption that integrates
multiple water supplies (surface and groundwater), and water demands. For hydro-economic
analysis, the model contemplates four main study scenarios: (i) current mean condition
(ESC1), (ii) at thirty years (ESC2), (iii) at fifty years (ESC3) and, (iv) at 100 years
(ESC4). These scenarios show fluctuation in water demand, and water supply based on
the population increase. Water resources management is often limited to studying water
quantity, although both quantity and quality of water have to be considered in order to
improve decision making. The incorporation in the model of a water quality parameter in
hydro-economic optimization increases its complexity and uncertainty.
 78 5 Integrated Water Resources Optimization Model

5.1. Introduction

The constant increase of water demand for energy, agriculture, industrial development, and
environmental sustainability, has constraint water resources throughout the world (Tortajada
et al., 2019; Davis, 2007). This rise is strongly attached to quality worsening, according to
the requirements necessary for each use, given the relationship between quantity and quality
in water supply (Davidsen et al., 2015; Martinsen et al., 2019; Walker et al., 2015). Water
quantity stored in hydrographic basins depends on the hydrological process and the water
quality depends on the interaction with punctual or diffuse contamination sources, produced
in the industrial development processes.

Water quality spoilage is related to the increase in supply costs (Cánovas et al., 2017;
Davidsen et al., 2015; Maliva, 2014). This increases summed to a demand increment for
water in specific areas have shown the necessity to apply more efficient processes in the
water allocation methods (Assaf & Saadeh, 2008; Metcalfe et al., 2017; Pulido-Velazquez
et al., 2008). According to, Integrated Water Resources Management (IWRM ) concept is
used to express the relation between water offer and demand based on availability; integrating
quantitative and qualitative conditions (Al-Jawad et al., 2019; Ferreira et al., 2019; Wongsa,
2015; Rogers et al., 2003).

In recent years, one of the IWRM biggest challenge is to provide an efficient water allocation,
due to the up-growth that agricultural, aquaculture, livestock, O&G, industrial, mining and
domestic demand has turned towards the water availability (Dessu et al., 2019; Pérez-Blanco
& Gutiérrez-Martín, 2017; Pérez-Uresti et al., 2019). IWRM uses optimization models that
can integrate the hydrological modelation (surface water and groundwater), the economic
evaluation of water, and mathematical optimization. This focus provides a cheap analysis
tool to assess water allocation within the study area, which could help in decision-making
and in the understanding of the system behavior (Al-Jawad et al., 2019; Li et al., 2019;
Pulido-Velazquez et al., 2008). The water quality incorporation in the water resources
optimization process allows to restrict the water allocation according to its use (Davidsen
et al., 2015; Fu et al., 2017; Martinsen et al., 2019). Nevertheless, literature reviews showed
few studies related to quality and quantity water optimization.

Hydro-economic models combine the economic system of regions and the engineering
concepts for water resources modeling in order to analyze water allocation inside a
cost-benefits appraisal under hydrology parameters (Booker et al., 2012; Harou et al., 2009;
Reynaud & Leenhardt, 2008). Optimization models that incorporate economic objectives are
widely used to evaluate decisions in water resources planning and management (Al-Jawad
et al., 2019; Fu et al., 2017; Mohtar et al., 2019). The applications of hydro-economic
5.2 Methodology 79

optimization models are numerous. Authors as Harou et al. (2009) provide a complete
description of hydro-economic modeling concepts and Labadie (1975) gives a revision of the
mathematical optimization process for water resource planning. Punctual water sources
analysis is useful to determine and analyze the system operation, except for a regional scale
water resources planning process, where optimization of various water sources is required.

Hydro-economic optimization models that integrate water quality, focus in administration

strategies that minimize the pollution or improve water allocation management to limit the
pollutant loads. Authors as Ferreira et al. (2019); Gunawardena et al. (2018); Morales-Marín
et al. (2017); Peña-Haro et al. (2011) present studies that emulate water quality as a result of
different management strategies. Those strategies seek to optimize the optimization model
outcomes under imposed quality restrictions. Water quality standards may also be fulfilled
using allocations and its impacts in water quality.

Modelling effects over water allocations regimes that involve several users and sources
is complex and interdisciplinary. Various studies establish models related to economy,
hydrology, hydrogeology, and ecology, and even though all studies aim to integrate different
disciplines, each of them focuses in a particular subject. For example authors like Velázquez
et al. (2003) focuses in determination of values to evaluate water shortage; Switzman
et al. (2015) centers in availability and salinity of water for irrigation purposes; Jia et al.
(2018) centers in create optimal strategies for water allocation between users; Lopez-Nicolas
et al. (2018) determines prices and institutional limitations for water allocation in irrigation
purposes; Blanco-Gutiérrez et al. (2013) focus their studies in integration of economy and
ecology and Maliva (2014) shows the relationship between economic activities and water
quality.

5.2. Methodology

IWRM is interpreted as the maximization of net profits in a regional model (García-González

et al., 2007), seeking to achieve an economic efficiency (optimal) water allocation for all users.
In this research, an integrated water management approach was performed in order to: (i)
obtain an economic solution for allocation and optimization of hydro-economic models, (ii)
to evaluate maximum profits of users in the region, and (iii) to analyze the impacts of water
shortage base on the quality parameters. In this sense, the model explains optimal water
allocation within economic sectors, based mainly in the population up growth. Moreover,
when including physical and economic restrictions, the model provides a cost estimation
for water allocation with different availability levels. Thereupon, are explained the main
elements of the model: water use profits in each sector, relevant restrictions, and the relations
that represent surface and groundwater flow.
 80 5 Integrated Water Resources Optimization Model

5.2.1. Parameters Associated with Water Usage

Optimization model parameters determination has been performed in a municipality level

scale, to evaluate benefits and costs related to water management and usage in regional
level. The model results show optimal water levels, profits, and costs of the system for water
allocation in the region. Nonetheless, most of the parameters correspond to 2018 which is
considered the reference year of this work. Some prices are adjusted to reflect projection
values at thirty, fifty and a hundred years, with a 3.33% increment based in consumer price
index analysis (PIA) reported by DANE (2019).

As in most cases, the rates vary according to water use type (commercial or domestic), in this
research it was used the ordinance N. 1155 by the environment and sustainable development
Ministry issued in 2017 (Ministerio de Ambiente y Desarrollo Sostenible, 2017), to calculate
the water rates in each sector. The estimated cost for service (VP) is formed by water use
price rates (T U – M U SD/M m3 ) and volume of water collected (V – M m3 ). This volume
is equivalent to demand satisfied in each economic sector, in a period of time t, corrected by
the cost of opportunity factor (Fop) (Eq. 5-1)

V p = (T U )(V )(F op), (5-1)

where, Fop factor (Eq. 5-2) takes in whether the users make adequate usage of the water or
not. For this model, it was analyzed a non-returning water basin scenario.

Vc − Vv
F opreturtning = ; F opnon−returtning = 1, (5-2)
Vc

Additionally, T U (Eq. 5-3) is expressed by multiplication between minimum rate – T M

(0.0032 U SD/m3 ) and the regional factor (F R) (Eq. 5-4), that compiles the water
availability factors, basin investment requirements, and social and economic community
condition.

T U = (T M )(F R), (5-3)

F R = [1 + (Ck + Ce )Cs ]Cu , (5-4)

where, Ck is investment coefficient (Eq. 5-5) and represents the total cost of basin
management (Basin Management Plan, in Spanish POMCA - order N. 1279 of 2002). Cpmc
is the annual total costs for the management plan of previous year, and CT M shows the
5.2 Methodology 81

annual billing of water use. In absence of POMCA, the investment coefficient value will be
zero.

Cpmc − CT M
Ck = ; 0 < Ck < 1, (5-5)
Cpmc

Ce is the shortage coefficient (Eq. 5-6) and varies depending the source (surface or
groundwater). This coefficient is considered from the water usage index, which reflects
the relation between offer and demand reported by each municipality, for the area affected
by the POMCA.

Cesurf ace =
0 → Ies < 0.1
5/6 (5-6)
→ 0.1 < Ies > 0.5
1 − (5/3)Ies
5 → Ies > 0.5

Cegroundwater =
0 → Ieg < 0.1
40
→ 0.1 < Ieg > 0.5
49 − 90Ieg
10 → Ieg > 0.5

Cs is a social and economic conditions coefficient (Eq. 5-7) and is differentiating the domestic
supplying and the other kinds of water usage trough the unsatisfied basic needs index (DANE,
2017).

100 − N BI
Cs = ; 0 < Cs < 1, (5-7)
100

And finally, Cu is usage coefficient, that varies according to final water use in each sector.
For domestic, agricultural, livestock, aquaculture and energy generation, Cu = 0.0775, for
the other usages Cu = 0.2 (Ministerio de Ambiente y Desarrollo Sostenible, 2017).

5.2.2. Water Demand and Quality

The domestic sector demand was based on the allocation of water provision for human
consumption of the Technical Regulation of the Drinking Water and Basic Sanitation
 82 5 Integrated Water Resources Optimization Model

Sector - RAS 2000 (Order 2320 of 2009 - Ministry of Environment, Housing and Territorial
Development) and the National Census of Population and Housing, 2018. The projected
domestic sector demand was performed trough the construction of a population up growth
scenario from reference levels in 2018 reported for DANE (2019) and following the Eq. 5-8.

GP = Pbase (1 + g)t , (5-8)

where, Pbase refers to population in 2018. This information is spatially distributed throughout
the model area. g is the projected growth rate in the period. This information is
geo-referenced by the municipality and spatially distributed in a 6 km by 6 km grid in
the model area by Kriging method.

The estimation of the agricultural and aquaculture demand was carried out following the
methodology of (IDEAM, 2019) (Eq. 5-9). This methodology consists in the assignment of
a water footprint to the products sown and produced in the area. The green water footprint
is used exclusively in the agricultural sector and is based on the natural use of water from
the soil moisture. The blue water footprint refers to water that is extracted from rivers or
lakes and that does not return to the source (IDEAM, 2019).

Dtotal = %P (HHblue + HHgreen )%M A, (5-9)

where, Dtotal is the total water demand in the agricultural or aquaculture sector. P
represents the farming production in the municipality, HHblue is the blue water footprint and
HHgreen is the green water footprint, and M A is the municipality area. This information is
geo-referenced by the municipality and spatially distributed in a 6 km by 6 km grid in the
model area by Kriging method.

To calculate the livestock sector demand, water consumption was classified by type of animal.
In the case of beef cattle, Table 5-1 shows the water amount considered, which includes in
a global way the water used in breeding, raising, fattening, benefit and slaughter. Table 5-2
shows the water consumption for other species. This consumption must be multiplied by
the percentage of the municipality area.

Table 5-1.: Daily water demand for beef cattle (Ideam, 2014).
Age (years) Demand (l/day)
<1 30
1-2 45
2-3 110
>3 115
5.2 Methodology 83

Table 5-2.: Daily water demand for animals for other animals (Ideam, 2014)
Species Demand (l/day)
Swine 30
buffalo 82.1
Goats 10
Sheep 5
Equine 50
Birds 2.4

Construction, mining, industrial, and services sectors monthly demand were obtained from
data reported by DANE (2019) and compared with data addressed by the National Water
Study (IDEAM, 2019). O&G sector demand was compiled from data registered in extraction
wells reported in the study area (Annex 4).

Water quality for the domestic sector was classified by the water quality index (WQI )
(Eq. 5-10). This index represents the numerical value that qualifies the water quality of
a surface current based on the measurements obtained for a set of five variables recorded in
a monitoring station j at time t.

n
X
W QInjt = (Wi )(Iijt ), (5-10)
i=1

where, Wi is the weighting or relative weight assigned to the quality variable i; Iijt is the
calculated value of variable i (obtained by applying the corresponding functional curve or
equation), at the monitoring station j, recorded during the measurement made in the period
of time t;and n is the number of quality variables involved in the calculation of the indicator;
for this research, n is equal to 5.

The variables used in the computation of the W QI were: dissolved oxygen (DO), total
suspended solids (TSS), chemical oxygen demand (COD), electrical conductivity (EC) and
pH. The weighting of each variable was 0.2. The optional values that the index can take
have been classified into categories, according to these values, the quality of the surface
water is rated between 0 (non-viable by sanitary standard) and 1 (Free risk) (Table 5-3)
(Ministerio de Ambiente, 2007).
 84 5 Integrated Water Resources Optimization Model

Table 5-3.: Classification of the health risk level according to W QI.

Classification Risk Level Alert signal
0 - 0.25 Sanitarily Inviable Red
0.25 - 0.5 High Orange
0.5 - 0.7 Medium Yellow
0.7 - 0.9 Low Green
0.9 - 1 Risk free Blue

This information was obtained from 380 samples in the influence area for the year 2018
(Instituto Nacional de Salud, 2018). 54% of this data corresponds to samples in the urban
area (Figure 5-1). Water quality in the other sectors (O&G, mining and industry) was
analyzed through the HCO3 , SO4 , Cl, Na, Ca, Mg and K. This information was obtained
from samples in the surface and groundwater spots in the area and the quality data
registered in the environmental fulfillment reports delivered to the ANLA as reported by
Malagón-Navarro (2017).

Figure 5-1.: Water quality behavior in the study area, based on the WQI.
5.2 Methodology 85

According to the results obtained by Malagón-Navarro (2017), the average pH behavior

in the study area (Figure 5-2-pH) is slightly acidic (pH = 6). These results indicate to
dissolution processes are facilitated in the area and the main pH decrease factor is the
industrial activities. The liquid waste discharge that contains nitrates and sulfates affects the
natural weathering rates, and reduces the soils and rocks attenuation capacity, inducing a rise
in the acidity of shallow groundwater, especially in areas with carbonated minerals deficiency
such as the Quaternary units. Additionally, these results showed low values of the electrical
conductivity (EC) (Figure 5-2-EC), common of the aquifers with shallow groundwater levels
and short residence times. However, the study area showed areas with high EC values,
located in the influence area of the O&G fields.

Figure 5-2.: Groundwater quality behavior in the study area base on Electrical Conductivity
and pH. Adapted from Malagón-Navarro (2017).

The hydrochemical classification results were analyzed by Malagón-Navarro (2017) based

on the Piper diagram (Figure 5-3). These results determined that the calcium bicarbonate
hydrogeochemical facie is the water type dominant in the MMV area. Additionally, these
results relate to the behavior of the Quaternary deposits present in the study area in a
manner consistent with shallow groundwater behavior and short residence [Link] water
type represents a recent infiltration and its chemical signature is not different from the
rainfall water. The sodium bicarbonate water is also important in the area, and are located
in the Magdalena River floodplain deposits, in which silicate dissolution processes occur.
 86 5 Integrated Water Resources Optimization Model

Figure 5-3.: Piper diagram according to the six groups established. The samples are found in
the calcium bicarbonate water domains and in a smaller proportion for sodium
bicarbonate water type and calcium-sodium chloride. Group 1: Shallow samples
with short residence times. Group 2, 3 and 4: Samples with intermediate depths
and EC. Group 5: Deep wells. Mineralized samples and extracted at greater
depth. Group 6: Piezometers. Shallow samples with a higher mineralization
degree. Source: Malagón-Navarro (2017)

Based on the behavior analyzed above, a water quality variation was proposed to thirty,
fifty, and a hundred years scenarios, that allowed to propose a water price change. These
scenarios constituted a possibility to analyze the basin behavior, under a deficit situation.
This change is described in Table 5-4.
5.3 Baseline and Optimization Model 87

5.3. Baseline and Optimization Model

In this stage, it was created a simple code in the R software to solve the optimization model.
This model seeks to find the values of the decision variables that maximize the profit in water
allocation. The baseline model (ESC1) maximizes the water usage benefit in the social and
economic current conditions (Bwater) conditioned to the aforementioned restrictions. The
baseline optimization model structure is based on the subtraction between, total income and
the total cost (Eq. 5-11). The total income, is in function of the water productivity and
deficit. The total cost, is in function of water price and a penalty for not meeting the water
demand.

M ax(Bwater )
(5-11)
Bwater = Itotal − Ctotal

In areas with water shortage, the total demand exceeds the availability thus, the requirements
from every user cannot be fulfilled. This situation is not evenly distributed, doing necessary
to use regulations that allow prioritizing and organizing the assignment. For this, the model
propose has contemplated an economical penalty related to the deficit.

The Itotal (Eq. 5-12) obtained by water usage is based on water productivity by each user
(ui ) and used water amount (Ru ). The water used is the difference between demand water
amount (W Du ) and the water deficit in each sector. Economic value for productivity in the
domestic sector is cero. These data were obtained following the methodology used by DANE
(2017).

XX X
Itotal = Ru,t (W Du,t − Def icitu,t ),
t u s
XX XXX (5-12)
Itotal = Ru,t W Du,t − (W Du,t − W Su,t ),
t u t u s

where, t represents the time, s is the water extraction source (surface or groundwater) and
W S represents water availability by each source.

Likewise, the Ctotal obtained, refers to water shortage costs (Eq. 5-13).

XXX
Ctotal = P ru,t,S W St,S − P E(W Du,t + W St,S ), (5-13)
t u s
 88 5 Integrated Water Resources Optimization Model

where, P r is water price that every user must pay, and this will depend on the water source.
P E is the penalty that takes into account the cost of not supplying the water demand by
the users.

The general model restrictions focus on four axes: (i) the environmental flow, (ii) the water
availability in a period, (iii) the minimal domestic provisioning and (iv) the price variation
with respect to water quality. The environmental flow was determined as 3% of water supply.
This flow is used as a necessary ecological flow for environment conservation (MAVDT, 2013;
Parra & Carvajal, 2012). That each source can be sent to different demands’ destinations,
but it can not exceed the total source availability (Eq. 5-14).

X
W A ≤ W S, (5-14)
t,S

where, W A represents the water allocation for each source (s) in a period of time t, and W S
is the water availability by source.

Minimum domestic water supply restriction is defined according to the current normative.
This normative establishes the minimum water quantity must be provided for each person per
day is 50 liters (in the study area conditions ) (Ministerio de Vivienda Ciudad y Territorio,
2018). Thus, water quantity that comes to every domestic demand unit must be equal higher
that legal requirement. Finally, quality restriction considered a 5% rising in the price when
WQI is less than 0.7.

The hydro-economic model was solved through a lineal optimization process, that links all
available water resources and all water demands (Annex 4), under the limitations of demand
rising, water quality variance and offer decrease. The objective function maximizes the
profit in the MMV basin during a planning period of a year. The water demands may be
supplied with surface or groundwater. Nonetheless, given its high variability for each usage,
is necessary to evaluate the water source quality. The system considers the monthly water
demand from each user and a penalty for no satisfy it.

Each scenario described under-here, provide variations in the basic results in order to evaluate
availability or water demand changes. The ESC2, ESC3 y ESC4, to a 30, 50 and 100-year
projection respectively and they consider: (i) rise in water demand based on the population
growth (2% annual), (ii) changes in the water resource availability based on the results
obtained by Arboleda-Obando (2018), where water availability was evaluated by changes in
the climatic patrons, and change in the land use (1% for every 10 years) and (iii) changes
in price estimation as result of the relationship among the availability, demand, quality.
5.4 Modeling Results and Discussions 89

The generation of these scenarios is an internal and consistent vision of the modeler on
how the demand for water could be, based on the climate change analyzes shown by
Arboleda-Obando (2018). This model implementation is based on recreated data displayed
in Table 5-4.

Table 5-4.: Scenarios proposed in the integration of the economic aspects.

Experiment Demand Availability Price
ESC2: to 30 years +60% +3% +5%
ESC3: to 50 years +100% +5% +7%
ESC4: to 100 years +200% +10% +10%

5.4. Modeling Results and Discussions

In this section, it shows the results of the water allocations (surface and groundwater) and
profits, highlighting: water deficit in the study area, supply and demand, water availability,
and benefit maximization at the basin. The four scenarios analyzed shown in the optimal
solution the profit spatial distribution in the MMV Basin (Figure 5-4). This figure shows
that the ESC1 (Figure 5-4- 1A), presents positive yields in the analysis period as long as
there is no deficit in any productive sector. Nonetheless, is evident that some areas from
the oriental zone, have high values profit mainly due to water quality (Figure 5-1). The
maximum average profit is of 52 000 USD/year, and in a monthly basis, the highest profit
is presented in march (Figure 5-4- 2A). This allows ascertaining that the maximum profit
associated with water use is obtained when the 53.8% surface water and 57.6% groundwater
are assigned to the agricultural sector.

All scenarios present a deficit (Figure 5-5) in some economic sector, which is translated in
a penalty in the system leading to months with negative yields. In ESC1 and ESC2, the
water deficit (Figure 5-5-A and B) is slanted towards the east limit of the model. This
tendency is maintained over the years (ESC3 - figure 5-5-C and ESC4 - figure 5-5-D), and
the deficit tends to cover the center of the area. The month with more profit is February,
the lowest is March, and the annual average profits in each scenario are 52, 1 457, 1 112, and
-117 thousand USD, respectively (Figure 5-4). This data shows that in ESC4 the system is
not sustainable given the demand increase by the population variation and the water quality
affectation.

In ESC2, the projected rise in water demand has a rate of 2%. In this scenario, the analyzed
basin presents a 2% deficit to supply the user’s requirements (Figure 5-5-B). However, is
 90 5 Integrated Water Resources Optimization Model

evident that water quality worsening (Figure 5-1), limiting the surface and groundwater
allocation.

In this scenario, it can be seen that the economic impact associated with water quality
worsening is crucial in the election of the supply source. The maximum obtained profit in
this scenario with respect to the baseline scenario (ESC1) show a significant difference of
1.4 MUSD generated by the demand variation and water quality. Arboleda-Obando (2018)
suggests in his hydrological analysis of this area, that water availability variation and its
relation with demand, provides in the MMV basin a change in the land-use regimes that can
be settled to obtain a better water quality. The analysis also shows that variation in water
quality is presented close to the industrial, mining and O&G extraction and exploitation
developing points, located mainly in the central and oriental area at the model.
 5.4 Modeling Results and Discussions

Figure 5-4.: Results of economic aspects integration in the hydrological model. The numeral one, shows the profit in the
optimal solution and the numeral 2 shows the behavior over time of demand, deficit, and profit in the study
area.
91
 92

Figure 5-5.: Average annual deficit per scenario. A- shows deficit in current situation. B- shows deficit in projection to
30-years. C- shows deficit in projection to 50-years. D- shows deficit in projection to 100-years.
5 Integrated Water Resources Optimization Model
5.4 Modeling Results and Discussions 93

Figure 5-4-2B shows that the water deficit is higher in March with 2.5 M m3 . This allows
us to conclude that in March the penalty fee in the system is higher and thus the profit
is lower. Additionally, in the aquaculture and livestock sectors are not a deficit due to the
water quality and amount for each sector. The ESC2 results show that the higher deficit is
in the agricultural sector (1.8 M m3 ) due it is the sector that needs more water. The O&G
sector (0.48 M m3 ), industry (1.15 M m3 ) and mining (0.33 M m3 ) have a low deficit due to
lower demand and the cost associated to obtain a higher water supply. In ESC3, the basin
presents a 3% water deficit to supply the users needs (Figure 5-5-C). The comparison among
the ESC3, ESC1 and the projection to 30 years (ESC2) scenarios, shows that the deficit
is approximately double.

This analysis permits to show that the economic impact associated with water quality
spoilage is important in the supply source choice. Results comparison in this scenario with
respect to the previous ones shows a 300 000 USD/year difference approximately in profit
obtained for demand variation and water quality. Further analysis of this variation will lead
to defining the importance of the land-use changes and the economic impact in the region.
This scenario shows that water quality variation behaves similarly to the results showed
in ESC2, where the higher spoilage is produced around the industrial, mining, and O&G
sectors.

Figure 5-4-2C shows that higher deficit is in March with 4.3 M m3 , and supports the results
previously generated in the ESC2. In this scenario, the livestock sector has not water deficit
due to the relation demand/cost that the model evaluates. Sectors with biggest deficits are:
industrial (2 M m3 ), O&G (0.9 M m3 ) and mining (0.6 M m3 ) due to the model penalizes
with lower cost in these sectors, the non-fulfillment of water needs. This result shows that
the O&G and mining deficits doubled with respect to the ESC2 results. Finally, the ESC4
showed a 6% water deficit to supply the user’s needs (Figure 5-5-D). The comparison of
this scenario with prior ones indicates that spoilage behavior in water quality and supply
deficit in all sectors is similar. Results comparison obtained in this scenario with respect to
previous outcomes, shows a significant difference of -200 000 USD/yr in the profit obtained
for the demand variations and water quality.

Figure 5-4-2D shows that deficit continues to higher in March with 9.5 M m3 , which
represents the double deficit than in ESC3. In this scenario, the aquaculture sector has
not deficit. This trend correlated to relation demand/cost. The sectors with a higher deficit
are: agriculture (9.9 M m3 ), industrial (4.5 M m3 ) and O&G (2 M m3 ). Here, is evident that
the model no longer responds to restriction applied in supply, and although the agriculture
sector must be prioritized for allocation, the required water volumes are no longer viable
given the population water demand increase.
 94 5 Integrated Water Resources Optimization Model

5.5. Further Model Development and Limitations

A paramount concern in water resources management is uncertainty about future water.

Allocation control and water demand fulfillment associated with phenomena as droughts
and floods are often the main objectives of water management. The model results
presented in this research were optimized throughout an annual middle period, that cannot
capture the essence of all-weather variability in the area. Studies made in the same
area (Arboleda-Obando, 2018) have found that the future weather of MMV basin will
be influenced by changes in evapotranspiration, temperature, and some alterations in
precipitation patrons. These changes lead to changes in water usage perspectives and living
conditions adjustments in the communities regarding water quality and cost. The model
results may interfere with some possible future climatic changes, of those who we cannot
be certain of. Also, the model framework assumes a perfect prevision throughout the whole
planning period, which does not actually occur in reality. The water managers and planners
will not have a perfect prevision and will be restricted by the reach of the forecasts data
available in the moment. The main focus of this chapter was to compare different scenarios
of water allocation and not to generate operative decision backing. The perfect prevision
will underestimate real water prices, and the appraisal of this cost must be identified with
higher limits.

Input data and model optimization have nonlinear reliance with results water allocation
scheme. This improves the computational performance of the model but has a risk of
losing sight of general changes effects in water allocation. The model optimization can
not capture the effects in recharge, given the changing conditions in agricultural irrigation,
the water quality, return flows of domestic, and industrial sectors and agricultural nutrients
filtration. This limits the computational resources and uncertainty introduced starting from
parametrization and this interactions modelling. In the water management process, any
important change in the system must be identified, and the model has to be updated with
new information. Adaptation towards more sustainable groundwater allocation will result
in important changes in the system. Therefore, the model results must be evaluated as a
starting point for future analysis of water management in the area for decision making.
5.6 Conclusions 95

5.6. Conclusions

The general analysis of the studied scenarios establishes a series of important ideas about the
global effects of economic activities that must be prioritized in the study area and the required
infrastructure design projects to guarantee water allocation, bases on the hydro-economic
dynamics of the basin.

This part of the research shows that, in the ESC4 drastically alters the distribution of
profits in basin water resources management, affecting significantly domestic consumption,
and agricultural production. Through this analysis and assuming the behavior under a mean
hydrologic year, an 11% reduction in the agricultural, domestic, livestock, and industrial
sectors be would achieve an 8% in the system profits. Additionally, availability decrease
(surface water and groundwater) and water quality, reduce the profits in one million dollars
approximately. In this way, the concerns associated with total demand, allocation, and
fulfillment in a middle and long term basis does not seem excessive.

A remarkable discovery of the presented analysis is that there are significant differences
between the stakeholders allocation regimes. This has been identified between domestic and
agricultural sectors, but not between commercial activities. In conclusion, a general analysis
that would contemplate variable rates for each sector according to its productivity must
be updated. It is important to highlight that the whole system was optimized under an
equitable distribution in allocation and costs, and thus, the resulting profits would improve
results to satisfy all economic sectors.
6. Water Resources Management and
Planning

This chapter presents the management strategies of the optimization model. These
strategies allow us to characterize the water system with respect to allocation, availability,
supply, and demand (surface and groundwater) from an intensity levels assessment in the
study area. Throughout this chapter, you will find the indices presented to determine the
possible strategies of management, planning, and adaptation in water systems, and the
management strategies proposed for the allocation process.

A Integrated Water Resources Management IWRM tool, offers: (i) hydrological

consolidation (surface water and groundwater), (ii) management strategies design, planning,
and (iii) adaptation provide a solution in decision making that leads to efficient water
allocation, and to protect and preserve the ecosystems at a regional scale. In this
chapter, it describes a methodology appropriation to determine the management strategies
for joint use of water resources at a regional scale, from problem identification. For
that purpose, it describes four indices that allow characterizing the water system with
respect to allocation, availability, supply, and demand (surface and groundwater) from an
intensity levels assessment. The analyzed scenarios are based on the results obtained in the
optimization model presented in the previous chapter.
6.1 Introduction 97

6.1. Introduction

There is a growing concern regarding the unawareness of water offer (surface water and
groundwater), its quality, and the necessity to satisfy the future need for water especially
in the regions of high climatic variability (as a tropical countries) (Pérez-Sánchez &
Senent-Aparicio, 2015). The attention to this problematic in water resources management
has emerged in recent years as a key element to guarantee sustainable regional growth and
constitutes a transcendental and dynamic matter in societies (Rogers et al., 2003). This has
been made evident from the moment in which societies had to acquire political compromises
and promote agreements that allow responding to the growing tension and concern about the
actual a future state of water resources and their role in the development for the communities
(Al-Jawad et al., 2019).

The joined actions and efforts of governments are leading to improve the institutional
framework, including key subjects as: (i) implementation of coordinated agreements for
assignment and water usage, (ii) revision of legal an institutional groundwork regarding
the water subject, and (iii) promotion in the concept of economic value of water, thus, the
world is living a new legal development period towards: (a) regional lowland regulation,
(b) IWRM, and (c) sustainability of market and usage, based in an efficient allocation, the
optimization in profiteering, and satisfaction of required demand. These actions, represent
then an opportunity to build relationships between different state institutions for regional
water use, considering the local, regional and national synergies.

Water management systems evolution tend to the adaptation of the natural, political, social
and economic reality of the regions, including functional concepts, mechanisms, models and
strategies to satisfy the need of regions, the existence of competition between users and
economic sectors by water, and characterize the system problems. Thus, the main load and
incidence of resource assignation and determination of its management by the economical
actors, will determine its future, also including the economic activities associated to its use.
For this, hydroeconomic models may lead to analyze and interpret the system behavior to
identify planning, adaptation and water management strategies.

Given the fact that inclusion of these models to lead the orientation towards decision
making and policies design is pretty recent, the literature indicates a few revisions, analysis
and research studies that show the impact that generates these models in planning and
management for water resources. As an example, Payne et al. (2004) analyzed the
implications of different scenarios of climate change using simulation models. Subsequent
studies about water management and its impact on the climatic change also were based
in simulation models (Tzabiras et al., 2016). Another approach to this analysis, is the
 98 6 Water Resources Management and Planning

optimization models like the one used by Xie et al. (2018) to manage water resources in
agriculture and land use management, presented by Pérez-Uresti et al. (2019) to optimize
water resources management in shortage areas or the one showed by Martinsen et al. (2019)
to make a jointed optimization of assignments and water quality to satisfy the final users
demand.

To help the evaluation process of system performance and in the planning and management
process, several authors have proposed different strategies based on the indices results, which
allow to condense the impact of modelling in the water resource system management. For
example, Hashimoto et al. (1982) defined some indices to describe the reliability, resilience
and vulnerability of the water system. More recent research proposes methods to calculate a
series of the index that allows studying the water shortage in water resources systems, using
the results of an optimization model (Martin-Carrasco & Garrote, 2006). Considering that
these methods also have derived in research focused on identifying management, planning,
and adaptation of water systems that allows reducing the impacts as was showed by
Pulido-Velazquez et al. (2011); Du et al. (2018); Pedro-Monzonís et al. (2015).

Some of the previously proposed indices in the mentioned studies useful to analyze the
effects of the up-rise in demand, the variability of water quality and water availability in
the system. In this chapter, is presented a series of management strategies generated from
the incorporation of a methodology based on the index appropriation, those index evaluate
water assignment between different users to define management and planning strategies to a
regional scale, that contributes to the system improvement and the reduction in the impact
with respect to quality and availability.

6.2. Management Strategies

The four indices presented to determine the possible strategies of management, planning,
and adaptation in water systems are proposed by Martin-Carrasco & Garrote (2006);
Pulido-Velazquez et al. (2011). The inclusion of these indexes allows diagnosing the system
behavior after the water allocation to identify the current problems scenario, in thirty, fifty
and a hundred years.
• Demand Satisfaction Index (Is ): is the relation between assigned water quantity (S) to
each user (u) in an analyzed time period, and the total water demand (D) generated
by this user in the same period of time (6-1).

S
Is = , (6-1)
D
6.2 Management Strategies 99

when the water volume delivered is equal to the demand the relation tends to 1 and is
presumed that there is no vulnerability in the water resource.

• Reliability in Demand Index (Ir ): is the relation between water availability to be

assigned to users, with respect to the total demand in time period t (Eq. 6-2).

Sr
Ir = , (6-2)
D

In this index, it is considered that there is no reliability in the water offer when the
water available in the system is lower than 70% of demand. Values close to zero in this
relation indicates that the system cannot satisfy the demands in a reliable manner,
hence the system is prone to water shortage.

With the appropriation of those two indexes, it is possible to diagnose the system
management strategies, but in some cases, it is also necessary to identify the source or cause
of the problem in a more assertive approach, which can be achieved with the incorporation
of the extraction index and the extraction usage index.

• Extraction Index (Iw ): is defined (Eq. 6-3) as the relation between the total extracted
water volume from each of the system sources (Ysup and Ybase ), without considering
stored water in different structures with respect to total water demand.

Ysup + Ybase
Iw = , (6-3)
D

values close to zero in this relation indicate problems with water supply, therefore
strategies must focus in determine and assign complementary resources.

• Extraction Usage Index: this relation is defined to evaluate the percentage of water
resources extracted in the system to satisfy the demand (Eq. 6-4).

S
Iu = , (6-4)
Ysup + Ybase

low values in this index show that more water is extracted from the system that
necessary, leading to a higher system regulation need.
 100 6 Water Resources Management and Planning

6.3. Strategies for Planning and Management of Water

Resources.

Identification of problems evidenced as a response to indices analyzed, allow us to establish

different management strategies, planning, and adaptation. For this, considered three
intensity levels proposed in a qualitative scale: high (70%), intermediate (70% and 40%) and
low (less than 40%). Management problems are described trough four indexes that allow
characterizing the basin problems, giving a series of management, planning and adaptation
strategies (Figure 6-1).

System characterization starting from Is and Ir indices establishes that a problem associated
to resource availability (deficit, water insufficiency, rationing, and vulnerability), generate
in the worst scenario, problems with a shortage in the system (low qualification in Is and
Ir indices). In these cases, useful to use index Iw and Iu to identify the source of the
problems (Figure 6-2). Problems associated with low values of those indexes make reference
to excesses in the extraction flows, generating supply regulation, storage and infrastructure
problems in the system. These indexes (Iw and Iu ) will be qualified as high and low.
 6.3 Strategies for Planning and Management of Water Resources.
101

Figure 6-1.: Management, planning, and adaptation strategies used in the model.
 102 6 Water Resources Management and Planning

Figure 6-2.: System characterization through to problems identification base on the

management, planning, and adaptation strategies used in the model at the
study area. Adapted from Pulido-Velazquez et al. (2011)

6.4. Results

6.4.1. Management strategies proposed for allocation process.

The management model was used to calculate the described indices in the previous section,
in a current scenario, and in the three scenarios (defined with the combination of future
estimated hydrology, and different hypothesis regarding possible demand, and water quality
and availability). Indexes obtained for those scenarios were calculated on a monthly basis,
assuming an average medium behavior of the current condition that was evaluated. Figure
6-3 shows the correlation between Is and Ir indexes. The results from ESC1 (Figure 6-3-A)
6.4 Results 103

do not evidence problems of vulnerability and reliability, given the fact that water availability
and the offer vs demand relation is high. In ESC2 (Figure 6-3-B) correlation values are
presented for the indexes Is and Ir between 80% and 60%, which indicates that there are
changes in water availability, problems of system control, and possible affectations around
its capacity.

Figure 6-3.: Correlation between Is and Ir indices. A- Shown the correlation in the current
situation. B- Shown the correlation at thirty years. C- Shown the correlation
at one hundred years. D- shown the correlation at two hundred years.

The Iw and Iu indexes analysis (Figure 6-4-A), shows that for this scenario the main cause
of the problem is given by the low values of Iw that linked the changes in water availability
with variations in the assignment and possible problems of flooding.

Additionally, this scenario may present system control problems given the excesses in the
 104 6 Water Resources Management and Planning

volumes of water extraction. Given the management, planning, and adaptation strategies
the ESC2 would find itself in the situation proposed in Figure 6-5- Black Line.

The ESC3 and ESC4 present the same tendency of ESC2 and the correlation values in
the indexes Is and Ir can be found between 40% and 60% (Figure 6-3-C and D). In these
scenarios, the indexes Iw and Iu analysis (Figure 6-4-C and D) show that the problems of the
system may be associated to the low delivery efficiency and the water transit in the system,
also to the lack of control of the water requirements between different users. Given the
management, planning and adaptation strategies, the ESC3 and ESC4 will be encountered
in the situation established in Figure 6-5- Red Line.

Figure 6-4.: Correlation between Iw and Iu indices. A- Shown the correlation in the current
situation. B- Shown the correlation at thirty years. C- Shown the correlation
at one hundred years. D- shown the correlation at two hundred years.
6.4 Results 105

Figure 6-5.: Management strategies for scenarios ESC1, ESC2, ESC3 and ESC4 based
on problem identification. The red line represents the expected situation for
scenarios ESC1 and ESC2. The black line represents the system analysis for
ESC3 and ESC4.

The results showed, allow us to infer that high values in domestic water demand (as a result of
the ESC2, ESC3 and ESC4) may produce an increase in the necessity of water extraction,
therefore the system responds with high values of Iu . Additionally, the evaluated system
evidences that the values of Is and Ir indexes will increase in accordance, as the extraction
rates get higher. If the management strategies contemplate the possibility of making more
 106 6 Water Resources Management and Planning

flexible the joined water usage by different users or increase the reuse rates (decreasing the
pressure and restrictions of extraction volumes), the demand in ESC2 and ESC3 will be
supplied with less.

6.4.2. Management Strategies Analysis.

Analysis by scenario of the management, planning and adaptation strategies, based on the
risks and the potential benefits are shown in the Table (6-1). The strategies suggested, do not
necessarily constitute an exhaustive list, or restricts in any way the adaptation and inclusion
of others instead of the proposal here. These strategies do not make up a set of policies
but try to reflect an idea that may be appropriated within different political or institutional
guidelines to edifice the variables of impact in future scenarios. In this analysis is required
that implementation of management planning and adaptation strategies, depend on local,
regional, and national conditions. The regions with social and economic inequality must
contribute to the management policies to be focused on efficient and fair water assignment,
that produce economic and health benefits.

The lastly detailed strategies allow inferring that the water assignment processes, the
interventions, and regulation must consider in the solutions the water supply and the demand
(Navarro-Chaparro et al., 2015; Ki-moon et al., 2014; Harmanny & Malek, 2019). In so far
as to the offer, the planning, adaptation and management strategies involve an increase in
the storage volume or the water extraction. In demand, the strategies focus on how to
increase the efficiency of water allocation to ensure that the economic and social benefit is
maximized throughout the usage in sectors of higher value (Gleick & Palaniappan, 2010;
Lionboui et al., 2018). This allows to stand out that the importance of the water resource in
productive and social scopes require that policies and management strategies are well aware
of the generalized impacts of water.

Considering the scenarios with variable conditions to availability and water quality, it is
evident that the reallocation of water resources is a key action of adaptation to approach
the possible problems of water shortage. In parallel, it could be considered as potential
solutions in the changes and improvements of infrastructure, the land use or the definition
of usage priorities (Iglesias et al., 2018; Iglesias & Garrote, 2015; Alameddine et al., 2018).
Additionally, it is necessary to consider that these processes of allocation and decision making
might lead to conflicts between different users, it is essential to incorporate the interests of
the different parties (Pulido-Velazquez et al., 2018; de Sousa Fragoso & de Almeida Noéme,
2018; Pereira et al., 2019).
 Table 6-1.: Management strategies to different scenarios with risks and opportunities.
6.4 Results

Scenario Index Problem Strategies. Level1 Category2 Time Technical Potential Potential
Scale3 Difficulty4 Cost5 Benefits6
A Innovation: water usage efficiency. Us, Mu Mg M M M H
1
E Increase the system extraction regulation and P, Mu Mg M, L H H H
increase the regulation volume available for water
ESC1 conservation.
C Water reservoirs and water allocation for Us, Mu, P Mg, In L M H H
2
environmental and consumptive uses.
D Better monitoring and early alert. Mu, P Mg M M M H
Increase the system extraction regulation Mu, P Mg M M M H
3 C Improve infrastructure for storage in the system. Mu In M, L H H H
1 A Innovation: water usage efficiency. Us, Mu Mg M M M H
A Increase the system extraction regulation and P, Mu Mg M, L H H H
2 increase the regulation volume available for water
conservation.
ESC2
C Water reservoirs and water allocation for Us, Mu, P Mg, In L M H H
environmental and consumptive uses.
D Better monitoring and early alert. Mu, P Mg M M M H
3 C Increase the system extraction regulation. Mu, P Mg M M M H
A Improve infrastructure for storage in the system. Mu In M, L H H H
4 Industry relocation. Mu Mg L H H H
B
Clear priorities for usage. P Mg M L L H
C Innovation: water usage efficiency. Us, Mu Mg M M M H
1
D Increase the system extraction regulation. P, Mu Mg M, L H H H
A Water reservoirs and water allocation for Us, Mu, P Mg, In L M H H
2 environmental and consumptive uses.
ESC3 and ESC4 B Better monitoring and early alert. Mu, P Mg M M M H
C Increase the regulation volume available for P, Mu Mg M, L H H H
water conservation.
3 B Improve infrastructure for storage in the system. Mu In M, L H H H
A Industry relocation. Mu Mg L H H H
4 Clear priorities for usage. P Mg M L L H
B
Socialize the impact of future works to be P Mg M L L H
performed.
3
Users level (Us), policy level (P), and Municipality Level (Mu); 2 Management (Mg) or infrastructural (In); short time (S), medium time (M) or
long time (L); 4 –6 low (L), medium (M) or high (H).
107
 108 6 Water Resources Management and Planning

Reference processes around the water policies design have been establishing in several
countries. For example, in Europe, the EUWFD promotes public participation in the
allocation processes, decision making and policies around the management and handling
of water. Relly & Sabharwal (2009); Choi (2018); Garrido-Rodríguez et al. (2019) assures
that the users and communities demand more bluntly to be included in the processes used
to assign resources and watch for these to be transparent, based on scientific evidence and
to give results that can be of public interest. For that, a better comprehension of how
the user perceives the water system, its changes and needs, the adaptation policies and the
factors that influence in the support to those management, planning and adaptation policies,
may be a useful tool in the development of these political decisions. In the United States
exists the reference to the design of water policies with an analytic focus. In this case, the
general concern of communities for the state of the environment has increased the interest
in participatory decision making.

The local and regional necessities and capabilities are based on the potential to develop
new systems (Sordo-Ward et al., 2019; Garrote, 2017), implement efficient and adaptable
technologies to the conditions and needs of the system (Fabre et al., 2016), ideas and
technologic development to re-use water (González-Zeas et al., 2019), different alternatives
of water management generation of groundwater (Hernández-Bedolla et al., 2017), water
recollection techniques (López et al., 2019) and capacity to develop water markets (Harmanny
& Malek, 2019).

With this, integration of the community, the environmental analysis of the systems, and
the demands of water in the joint systems, permits the jointed management of surface and
groundwater to overcome the dry periods and to provide sturdiness in the water resources
systems (Pulido-Velazquez et al., 2011). Even though the local needs determine the scenario
for adaptation, planning, and management, the cooperation is a priority that allows enlarging
the knowledge transference generating local and regional resilience in the management
strategies (Hashimoto et al., 1982; Escriva-Bou et al., 2017). Finally, given the costs and the
lack of incentives associated with management planning and adaptation promotion capacity,
is unlikely that water management would be enforced through the introduction of new
independent policies, but rather through the revision and adaptation of existing policies
(Iglesias et al., 2011, 2018).
6.5 Discussion and Conclusions 109

6.5. Discussion and Conclusions

The main limitations of the management, planning and adaptation strategies for the handling
of water systems, are set by the definition of metrics in all the analysis units. The main
difficulty to develop efficiently this definition is ruled by the acting areas of the water resource
within the communities and ecosystems.

In this research, it is evidenced the importance of quantifying the surface and groundwater
offer to guarantee the economic development of the communities. Additionally, the offer
quantification leaves open the necessity to regulate and determine the available water amount
that can be used to cover the demand of communities and the environment. Likewise, it
is evidenced that the water demand is influenced by social and economic factors of the
community and in order to improve the efficient allocation processes it must acknowledge the
bias that the economic value of water may generate to users with low acquisition capacities.

The analyzed scenarios show that the population up growth enlarge the existing demand
patterns, which leads to the need of implementing planning, adaptation and management
strategies that maximize the benefit in the domestic and agricultural sectors and that reduce
the areas that in the future may present high levels of water stress. The analysis presented
in this research does not contemplate projections of technologic change that optimize or
improve the proposed scenarios. The assumption that the technological context and the
development of current water infrastructure will be valid in the future is evidently wrong. The
technologic change and innovation in the management of water resources will determine and
will reorient the adaptation, planning, and management strategies that were considered here.
Additionally, the analysis of the relation of cost-benefit of different strategies in management
requires information about social and economic changes in the communities that are not
contemplated in this research.

The political definition of clear guidelines that express the water rights may lead to the
sustainability of different economic sectors trough research methods (Tian et al., 2016).
Nevertheless, it has to consider that a favorable and equal governability may not be
applied around the water resource management. Even though policies that develop financial
incentives may lead to short term profits, they also may encourage unsuspected behaviors
between users that may generate negative impacts. For this linking, engineering and
economics around a social context are important, considering that the policies that are
based in objectively verifiable indicators, will result to be adequate.

The analysis presented in this chapter gives an evaluation of the main management strategies,
that can respond to risks and opportunities of future scenarios, in high variability climatic
 110 6 Water Resources Management and Planning

zones. The results here consigned show that the most interesting management, planning, and
adaptation strategies in terms of effort–benefit ratio are focused to: (i) improve the systems
of groundwater monitoring, (ii) establish clear priorities of usage among different users and,
(iii) regulate the water allocation between users to ensure a minimal conservation for the
environment. Additionally, the results may also evidence that the beneficial management
strategies in a regional scale are: (i) generation of a water marketplace, (ii) the improvement
of infrastructure for water storage and, (iii) the negotiation of agreements for water
assignment between users in an inclusive manner.
7. Conclusions and Outlook

7.1. Conclusions

The importance of characterize the water systems (surface water and groundwater) is based
in the implicit need of communities to survive and develop. Water as a vital subject for
sustainable development is determinant for social and economic development, water also is
a main concern when climate changing are faced. Given this situation, the water resource is
constituted as the main link between society and the environment, and thus, it generates the
need for conciliate the water demands. This impose the necessity for works and methods that
allow to understand the behavior of water systems, and also to foresee several explanations
to their complex behavior. Within those systems, groundwater flow in shallow aquifers are
part of the hydrological cycle, and they are affected by variability in recharge processes and
by human intervention. In the last years, water (mainly groundwater) has emerged as an
important topic given the increase in activities that requires provisioning from a water source.
Water supply to communities, agricultural development, mining, industry, and exploration
and production of O&G have led to seek different techniques that provide information in
specific scales. These techniques have analyzed the water systems based in simulations,
transformations, mathematical expressions, numeric models, and geo-statistical correlations,
without focusing in obtaining equivalent models and parameters that can be totally reliable
to represent different conditions.

This document is an attempt to characterize a water system (surface water and groundwater)
in an economical framework to determine strategies of management in a regional scale. This
objective involved to perform different tests and analysis that allowed to integrate those
aspects. To evaluate the efficiency of the proposed model and techniques, it was used the
MMV as a real scenario. First the system was characterized and recharge was defined,
taking in consideration that hydrological modelling is the base for integral water resource
management, and also the fact that in basins of the valley there is a complex interaction
of physical and chemical processes. For this, it was analyzed the interaction between the
hydrological cycle and the climatic conditions, observing that the used modelling patterns in
this research, even on other research made in the area, do not reproduce in an exact manner
the singularities of analyzed hydrograms. These differences in hydrograms representation
 112 7 Conclusions and Outlook

were applied mainly in base flow simulation and this behavior was attributed to the ignorance
of ground characteristics and the inability of the model to represent in detail deep percolation
processes. To establish temporal and spatial relations for this parameters it was necessary to
identify the range in which the parameters were adjusted to observed data and face them with
a posterior simulation period to assure this correlation. Given this, the completed sensitivity
analysis allowed to conclude that parameters that are involved in the model behavior and
the interaction of surface water and groundwater are the transmissivity recession curve, the
storage deficit in root zones, and the initial flow characteristics. These results were useful
to understand the hydrological dynamic in the valley area, the initial flow is correlated with
physical conditions of the basin, and that this flow vary depending in the analyzed period;
although this conclusion is obvious given the fact that the hydrological cycle is dynamic,
it was useful to establish similarities between different time periods in specified basin areas
that provide ideas in how the drainage networks and the superficial runoff velocity are going
to behave. The condition of saturation in the low end of the model caused a decrease in
surface flow quantity, due to gradient soil moisture decrease, that in consequence, increased
the superficial runoff. This allow to conclude that the influence of evapotranspiration in
hydrological behavior in the area through soil moisture analysis.

The hydrological characterization results, added to the geologic description of the area,
allowed to propose a conceptual hydrogeological model for MMV. In this model established
the correlation of superficial conditions with the hydraulic conductivity. The conceptual
model was validated trough a numeric model simulate groundwater flows and the flow
directions, there, it was seen that the representation of hydraulic conductivity and the
thickness of modelled aquifer variations were variables in time and space. This information
led to evaluate geo-statistical techniques that allowed to represent in a more appropriate way,
the zone characteristics. Although the MMV is a high ecologic, social and economic area
of importance for the country, it does not show an adequate relation between groundwater
information points, affecting flow behavior pattern determination. Additionally, groundwater
system characterization allowed to conclude that geostatistical methods reduce the errors
in determination of hydraulic properties. The sensitivity analysis performed, allowed to
identify which parameters govern the model behavior and quantify the correlation with
defined geological units. The hydrogeological modelling results allowed to conclude that
hydraulic conductivity, the porosity and the specific storage directly rule the flows and flow
directions in the aquifers. The conclusions about the hydraulic conductivity establishment,
allowed to identify a clear correlation of superficial layers parameter that represent the
Quaternary aquifer. The uncertainty in the identification of characteristics in higher depths
than 600 m, biased the initial interpretation results established in the conceptual model.
The results in this geological units allowed to identify equifinality problems in some areas
specified in the model. The conclusion in the stage allowed to establish that flow directions
are mainly direction from South to North. Additionally, some areas in the model add flows
7.1 Conclusions 113

of West and East limits. The Jurassic and Cretaceous units are identified as water dividers.
This behavior is concluded from the low values of conductivity, an interesting fact is this
model, is that the estimated field reflected defined geological structures in a large scale,
which were perceive in the conceptual model trough the constant zone definition.

Finally, the integration of these aspects in the economic optimization model, allow to
determine management strategies from an allocation process based in usage, amount, and
water quality. These strategies were proposed in the development of four scenarios. The
general analysis of these scenarios allowed to consolidate a series of important ideas about
global effects of different economical activities that can be prioritized. The results of this
integration showed that the demographic increase and the water need changes in economic
developments raise the demand patterns. The general conclusions of this investigation are
focused in provide strategies that allow to establish a balance between water resource use,
and ecosystem protection. Thus, it is concluded that integral water management is restricted
by: (i) the limitations in the allocation processes, (ii) information identification that allow to
correlate problems to evaluate the management capacity and (iii) the difficulty to quantify
the real offer and demand of water. The uncertainty in hydrological modelling (surface and
groundwater) and long term planning are added to the complexity of management strategies.
Because of this, the knowledge transference, and technology to define policies and rules, are
essential to allow an adaptive action of water needs and usages. The need to strengthen the
knowledge base towards hydrodynamic processes through investigation, will not be enough
to modify and boost the management strategies in water resource planning.

With respect to specific impacts, it was concluded that there are significant variations
between allocation regimes of users. Because of that, is evident that the need of involve
the interested parties, trough the development of proper methodologies to evaluate the
impacts, vulnerabilities and planning as a requirement for a profitable management. The
direct influences in the water users’ behavior, as changes in the demand, allocation and rate
regulation, must be considerate in conjunction with variations of water offer. This allow to
conclude that even though the strategies must be sustainable and participative, they also
must be appropriated and relevant in relation with specific factors of each economic sector.

The outlook presented in this investigation allow to conclude in a general manner that the
demographical increase raises the existing demand patrons. This leads to the necessity
of implement planning strategies, adaptation and management that maximize the social
benefits in the domestic and agricultural sectors. The analysis of the major strategies
of management proposals in this investigation, respond to problems of different economic
sectors in the area. These results allowed to conclude that the main management, planning,
and adapting strategies in terms of effort-benefit ratio are centered in: (i) improve the
monitoring systems for groundwater. Priorization of monitoring systems that allow to
 114 7 Conclusions and Outlook

understand and characterize the best existing conditions, the current and potential problems
and establish priorities, policies and viable strategies for water resource management. (ii)
establish clear priorities in usage between different users. The process of water management
and planning must be dynamic and iterative, and must allow the interaction, learning and
feedback between different users. This strategy will allow the simultaneous implementation
of practices and activities according the needs of users without losing sight of the general
context. And finally (iii) regulate the water allocation between users to guarantee a minimal
conservation for the environment. To balance the increasing water demand between the
aquatic and land ecosystems and the base flow in the upstream sections, will allow to analyze
the bidirectional relations between the macroeconomic policies, the social and environmental
objectives, the development, and the management and use of surface and groundwater. A
last consideration about the implementation cost of management strategies allow to conclude
that they are from low to medium cost and technically manageable in regional planning for
the basin. Nevertheless, is also evident that there are strategies that require a large scale
effort, either in the distribution network management or in the infrastructure development.

Finally, the novelty in this work is the combination and coupling of several tool of modelling
and analysis, to a real case study in a tropical basin. The technics and metrics of validation
used in this investigation, must be enlarged to guarantee their representativeness and to
improve the hydrological and hydrogeological characterization, and the identification of
allocation and usage problems. This way, there will be establish more reliable parameters,
and more more accurate results.

7.2. Outlook

Characterization of MMV water system presented in this document allows to suggest future
research focused on the following subjects.

• Hydrological models and coupled hydrodynamics and physically based that allow
to identify hydraulic parameters in different scales. Evaluate the correlation of
these parameters and establish mathematical models that represent the general
characteristics of the systems must be addressed in new researches.

• The variation in water quality, that may interfere in surface water quality or worsen
the groundwater quality. To analyze the interactions in the water systems and analyzes
the decrease on flow velocity that raise the presence of dispersion phenomena. A study
about the fluids re injection in aquifers must be approached to add planning tools, and
decision making management.
7.2 Outlook 115

• Generalized application of geo-statistical methods to hydro geologically characterize

water systems, will allow to consider correlations of parameters and analyze the
mathematical model behavior that allow to escalate the hydraulic conductivity. These
studies will allow to reduce the uncertainty in areas where there is not a detailed
information monitoring.
• Dynamic price policies could be established trough multi scale studies of integral water
management, that include strategical environmental evaluations in different economic
sectors
A. Annex: Initial Data

Hydrology Data

Name Code Number Category Latitude Longitude Z Missing Data (%)

Nus 23085080 CP 6.49 -74.84 835 5.25
Virginias 23090020 PM 6.39 -74.68 639 5.29
El Tigre 23100030 PM 6.84 -74.79 994 2.65
Apto OTU 23175020 SS 7.01 -74.72 718 7.03
El Bosque 27030090 PM 7.32 -74.84 220 0.74
Palma de Coco 27030140 PM 7.24 -74.81 100 3.98
La Sierra 24025030 CP 5.97 -73.16 2700 1.72
Pto Berrio 23095010 CP 6.47 -74.41 150 4.50
La Verde 23120010 PM 6.41 -73.90 305 1.22
Pto Araujo 23120200 PM 6.53 -74.08 159 8.34
Carare 23125050 CO 6.65 -74.06 168 6.50
Chucuri 23130010 PM 6.88 -74.03 100 1.80
Apto Yariguies 23155030 SP 7.03 -73.81 126 8.70
Yondó 23160010 PM 7.09 -73.94 140 5.15
Sabana Torres 23180070 PM 7.39 -73.49 176 6.39
Eloy Valenzuela 23180080 PM 7.49 -73.68 132 1.84
Villa Leiva 23185010 CP 7.46 -73.54 328 9.72
El Playon 23190140 PM 7.46 -73.20 500 7.55
El Picacho 23190300 PM 7.11 -72.97 3310 1.50
Matajira 23190340 PM 7.21 -73.06 996 0.02
Portachuelo 23190360 PM 7.33 -73.17 800 8.54
La Galvicia 23190400 PM 7.12 -73.06 1779 3.91
El Naranjo 23190440 PM 7.21 -73.30 825 4.29
El Pantano 23190600 PM 7.00 -73.23 1280 1.34
Piedecuesta 23190700 PG 6.99 -73.07 1000 1.95
Vivero Surata 23195090 CO 7.37 -72.99 1725 4.36
Llano Grande 23195110 CO 7.03 -73.17 777 6.44
Palonegro 23195130 SP 7.12 -73.18 1189 2.09
Cachiri 23195200 CO 7.47 -72.99 1850 2.84
La Comoda 24010110 PM 6.08 -73.47 1242 4.66
Confines 24010230 PM 6.36 -73.24 1523 5.81
Bolivar 24010640 PM 5.99 -73.77 2260 8.85
Olival 24010650 PM 6.15 -73.34 1502 1.49
Simacota 24010660 PM 6.44 -73.33 1050 1.80
Suaita 24010760 PM 6.10 -73.44 1617 6.36
Continued on the next page
 117

Name Code Number Category Latitude Longitude Z Missing Data (%)

Guavata 24010820 PM 5.96 -73.70 2018 5.52
Chima 24015260 CO 6.36 -73.37 1090 4.08
Coromoro 24020120 PM 6.30 -73.04 1520 2.30
El Cucharo 24025020 CP 6.53 -73.20 975 9.55
Mogotes 24025040 CP 6.47 -72.97 1673 3.24
Cepita 24030300 PM 6.75 -72.98 600 3.55
El Tope 24030330 PM 6.94 -72.93 2050 3.21
Sta Isabel 24040050 PM 6.64 -73.21 1300 0.31
La Putana 24050070 PM 7.13 -73.52 150 1.46
Albania 24050110 PM 6.91 -73.63 300 0.94
Zapatoca 24055030 CO 6.79 -73.28 1810 4.49
Aguasclaras 24060040 PM 7.26 -73.55 188 0.65
La Mesa 24060050 PM 6.76 -73.09 1460 0.94
La Paz 24060060 PM 7.11 -73.42 180 1.22
La Parroquia 24060070 PG 7.08 -73.33 267 8.92
Hda Brisas 24065010 CO 7.25 -73.79 138 12.81
Campo Capote 23125040 CO 6.62 -73.92 180 13.23
Landazuri 23125130 CO 6.22 -73.81 1085 2.92
Pte Ferrocarril 23140040 PM 6.77 -73.94 90 3.74
Pto Wilches 23180020 PM 7.35 -73.89 128 6.11
El Porvenir 23180040 PM 7.45 -73.48 154 1.04
La Coquera 23180120 PM 7.22 -73.92 170 2.29
Tona 23190130 PM 7.20 -72.97 1910 2.01
Laguna 23190260 PM 7.08 -73.21 1050 8.58
Palo Gordo 23190280 PM 6.97 -73.13 950 0.33
Llano de Palmas 23190350 PM 7.24 -73.20 778 0.97
Palmas 23190380 PM 7.21 -73.22 855 3.06
Cooperativo 23205020 CO 7.48 -73.93 165 5.37
Oiba 24010240 PM 6.26 -73.30 1400 2.18
La Laja 24015250 CO 6.24 -73.42 1400 3.26
Velez Granja 24015270 CO 6.00 -73.67 2170 2.35
Encino 24020040 PG 6.14 -73.10 1814 8.64
San José 24020080 PM 6.44 -73.14 1300 0.40
Curiti 2 24020130 PM 6.60 -73.06 1626 1.07
Charala 24025050 CO 6.27 -73.15 1350 8.37
Remolino 24040060 PM 6.61 -73.28 630 0.82
San vicente 24050060 PM 6.87 -73.41 721 0.62
La Fuente 24050100 PM 6.71 -73.28 815 3.36
Pto Boyacá 23115010 CO 5.98 -74.57 350 8.68
Pto Berrio-Aut 23097030 LG 6.49 -74.40 108 0.67
San Pablo-Aut 23207040 LG 7.48 -73.92 64 19.62
 118 A Annex: Initial Data

Points Inventory - Observation Points in Steady State

x y Q (m3 ) Level Depth Radius Z top bottom

975497 1184466 9.07 6.0 38 0.0762 146.2 140.2 108.2
978367 1196624 146.88 18.3 38 0.0762 150.0 131.7 112.0
991395 1198951 80.64 17.1 25 0.0508 160.8 143.7 135.8
988823 1207855 103.68 7.9 20 0.0508 122.9 115.0 102.9
989394 1209989 80.64 18.3 38 0.0508 135.6 117.3 97.6
1010571 1259484 15.26 2.0 8.91 0.1651 73.1 71.1 64.2
1010191 1259529 9.50 0.8 7.15 0.0127 77.8 77.0 70.7
1012465 1260133 4.78 1.3 4.5 0.0127 76.9 75.7 72.4
1012153 1260146 53.28 4.4 7 0.0127 91.4 87.0 84.4
1012800 1260184 1.44 0.5 1.54 0.01524 92.6 92.1 91.1
1013313 1260320 39.74 3.4 4.2 0.0127 76.5 73.1 72.3
1009463 1260467 20.16 3.8 0.12 0.3175 94.9 91.1 94.8
1013196 1260494 26.21 3.4 30 0.0127 86.5 83.1 56.5
1013224 1260503 11.81 3.0 4 0.0127 87.6 84.6 83.6
1013284 1260513 12.10 3.4 90 0.002032 90.6 87.2 0.6
1013451 1260656 15.84 102.0 1 0.01905 84.9 -17.1 83.9
1011176 1260660 10.37 0.5 1 0.01778 77.5 77.0 76.5
1011832 1260737 9.79 2.6 2 0.0127 84.1 81.6 82.1
1011534 1261027 12.38 3.1 24 0.001524 81.5 78.4 57.5
1012010 1261836 79.78 1.3 31 0.001524 86.0 84.7 55.0
1014643 1261973 6.34 0.9 2 0.00762 78.9 78.0 76.9
1016614 1263818 3.74 2.5 358 0.003175 86.3 83.8 -271.7
1015233 1264063 15.84 4.4 50 0.000762 98.6 94.2 48.6
1015537 1264162 2.88 2.2 388 0.003175 92.5 90.3 -295.5
1016819 1264484 11.81 1.3 32 0.001524 87.3 86.0 55.3
1015625 1265058 9.50 10.5 1.6 0.023495 67.7 57.2 66.1
1018309 1276957 7.32 1.8 7.5 0.001143 71.6 69.8 64.1
1018293 1277517 6.91 3.2 7 0.001143 72.7 69.5 65.7
1018692 1277546 4.90 2.4 9 0.001143 74.0 71.7 65.0
1018809 1277605 16.99 0.9 11 0.001524 73.6 72.7 62.6
1018691 1277656 11.81 2.7 11 0.000762 72.8 70.1 61.8
1018432 1277676 10.08 1.1 16 0.000762 76.5 75.4 60.5
1017014 1277931 12.10 4.4 3 0.0127 71.3 66.9 68.3
1018736 1278595 23.90 2.2 8 0.0127 66.3 64.2 58.3
1018700 1278846 10.66 1.9 6.5 0.001524 67.0 65.1 60.5
1024130 1278908 15.84 5.5 8 0.01397 71.8 66.4 63.8
1023910 1278981 7.78 3.5 6 0.0127 75.3 71.8 69.3
1018438 1279046 8.93 2.1 12 0.001524 62.7 60.6 50.7
1023907 1279056 288.00 8.8 43 0.00129 77.5 68.7 34.5
1018719 1279125 15.84 2.5 9 0.001143 73.0 70.5 64.0
1023853 1279268 57.60 9.9 11 0.015875 81.0 71.1 70.0
1024572 1279277 23.90 2.7 6 0.01397 75.6 72.9 69.6
1024147 1279443 19.01 9.6 13.5 0.01397 81.5 71.9 68.0
1018773 1279485 0.04 2.3 9 0.002286 68.8 66.5 59.8
Continued on the next page
 119

x y Q (m3 ) Level Depth Radius Z top bottom

1024023 1279585 10.08 6.9 10 0.01397 82.1 75.3 72.1
1017263 1279610 47.81 3.8 3.7 0.0127 72.9 69.1 69.2
1023687 1279663 11.52 6.5 9 0.01397 77.1 70.7 68.1
1018696 1279706 3.74 2.0 9 0.001524 69.8 67.8 60.8
1018396 1279806 8.35 3.5 7 0.001524 71.2 67.7 64.2
1024075 1279831 3.17 10.5 13 0.01397 83.1 72.6 70.1
1018425 1279906 8.35 3.3 12 0.001524 70.8 67.5 58.8
1023999 1279961 23.90 10.0 12 0.01397 81.0 71.0 69.0
1024223 1280182 7.78 1.7 2.66 0.0127 69.2 67.6 66.6
1024228 1280378 23.90 2.1 5 0.0127 70.8 68.7 65.8
1018685 1280536 7.78 2.1 11 0.000762 71.5 69.5 60.5
1023829 1280821 23.90 4.0 6 0.01397 74.0 70.1 68.0
1018388 1280847 24.48 1.0 7.5 0.001143 73.8 72.8 66.3
1025423 1281125 31.68 10.5 14 0.01397 97.8 87.3 83.8
1018326 1281567 8.35 2.0 4.3 0.0127 61.1 59.1 56.8
1018189 1281607 37.15 2.8 21 0.001524 74.7 71.9 53.7
1025800 1282203 11.81 6.5 8 0.01397 94.4 87.9 86.4
1026374 1282333 11.81 8.8 12 0.01397 97.1 88.3 85.1
1026294 1282446 23.90 2.7 5 0.01143 95.4 92.6 90.4
1026247 1282467 15.84 2.7 4 0.01397 94.4 91.7 90.4
1025774 1282649 47.81 6.0 8.25 0.01397 90.6 84.7 82.4
1026332 1282673 31.68 1.4 2.5 0.01397 89.2 87.8 86.7
1025030 1282724 47.81 0.7 3.2 0.01397 76.1 75.4 72.9
1025518 1282726 39.74 5.6 7.38 0.01397 83.0 77.4 75.6
1025431 1282797 7.78 0.6 2.77 0.01397 80.1 79.6 77.4
1026334 1282797 15.84 1.0 3.5 0.01397 87.5 86.5 84.0
1026570 1282817 23.90 1.3 4.38 0.01397 76.1 74.8 71.7
1026448 1282820 10.66 1.4 2.5 0.0254 87.0 85.6 84.5
1026080 1282929 7.20 3.9 9.4 0.01397 90.1 86.2 80.7
1025924 1283042 0.75 3.5 10 0.01397 87.3 83.8 77.3
1026378 1283164 46.08 1.2 3 0.0127 88.4 87.2 85.4
1026123 1283184 15.84 2.2 3.2 0.0127 85.2 83.0 82.0
1026829 1283243 46.08 2.3 3.5 0.01524 72.2 69.9 68.7
1025747 1283578 11.81 2.8 9 0.01397 81.3 78.5 72.3
1025819 1283824 11.81 1.9 4.1 0.01397 76.7 74.8 72.6
1019033 1283854 8.35 1.2 5 0.0508 80.1 78.9 75.1
1016905 1283971 9.22 3.2 30 0.000254 69.7 66.6 39.7
1025115 1283994 31.68 5.6 7 0.01397 76.5 70.9 69.5
1025562 1284008 6.34 3.7 6 0.01397 77.2 73.5 71.2
1027796 1284192 31.68 1.5 7.15 0.01397 80.2 78.6 73.0
1027647 1284332 15.84 0.7 7 0.01397 85.1 84.4 78.1
1027652 1284505 24.77 2.7 5.3 0.01397 87.0 84.3 81.7
1027815 1284615 7.78 1.0 4.3 0.01397 89.6 88.6 85.3
1027278 1284674 7.78 0.9 4 0.01397 77.5 76.6 73.5
1026142 1284714 63.36 3.3 7 0.01397 75.4 72.1 68.4
1027560 1284816 11.81 0.9 3.12 0.01397 85.2 84.3 82.1
Continued on the next page
120 A Annex: Initial Data

x y Q (m3 ) Level Depth Radius Z top bottom

1018518 1284976 63.36 0.6 3.32 0.0127 69.1 68.5 65.8
1026385 1284993 23.90 3.0 5 0.01397 79.6 76.6 74.6
1028172 1285067 95.04 4.8 6 0.01397 77.4 72.5 71.4
1026659 1285089 15.84 2.4 7 0.01397 93.5 91.1 86.5
1027555 1285211 23.90 1.7 4.73 0.01397 83.2 81.5 78.5
1027890 1285432 23.90 1.1 5 0.01397 79.2 78.1 74.2
1015844 1285504 15.26 2.0 0.8 0.02667 62.0 60.0 61.2
1028024 1286164 95.04 8.0 14 0.0127 94.5 86.5 80.5
1028755 1286383 15.84 1.2 3 0.01143 78.9 77.7 75.9
1027562 1286404 17.86 1.0 4 0.01397 72.9 71.9 68.9
1029190 1286436 7.78 1.1 2.8 0.0127 81.5 80.4 78.7
1028632 1286462 15.84 0.9 4 0.0127 77.0 76.1 73.0
1029095 1286547 46.08 1.8 6 0.0127 80.1 78.2 74.1
1012601 1287242 8.06 1.9 2.5 0.0127 73.9 71.9 71.4
1014458 1287378 10.08 0.5 24 0.001524 67.2 66.7 43.2
1029971 1287509 66.24 5.0 9 0.02032 82.7 77.7 73.7
1022068 1288246 0.99 8.0 9 0.0762 68.3 60.3 59.3
1022343 1288503 11.81 3.0 7 0.000806 68.6 65.6 61.6
1022616 1288667 8.06 6.0 12 0.000484 77.0 71.0 65.0
1022672 1288693 7.20 7.0 12 0.000484 74.4 67.4 62.4
1023159 1288901 4.03 6.0 11 0.000484 68.5 62.5 57.5
1023014 1288948 28.80 6.0 10 0.000484 76.9 70.9 66.9
1023270 1288957 0.79 1.7 3 0.00635 71.0 69.3 68.0
1023353 1289024 15.84 4.5 10 0.000484 69.5 65.0 59.5
1023835 1289314 6.05 4.0 10 0.000484 73.3 69.3 63.3
1024918 1289330 11.52 1.2 7 0.000484 72.8 71.6 65.8
1024178 1289612 3.74 4.0 10 0.000484 72.4 68.4 62.4
1025800 1290658 12.67 2.4 8 0.01397 74.7 72.3 66.7
1029543 1290781 2.43 1.5 3.5 0.009779 75.6 74.1 72.1
1015894 1290804 5.76 0.4 32 0.001524 68.4 67.9 36.4
1029520 1291054 2.38 2.2 4.5 0.00635 70.0 67.8 65.5
1027087 1291217 17.28 4.0 7 0.0127 79.9 75.9 72.9
1034557 1291223 31.68 2.5 7 0.01397 80.8 78.3 73.8
1027160 1291230 13.25 3.9 10 0.01397 75.5 71.6 65.5
1029215 1291943 9.50 7.9 14 0.0127 81.4 73.5 67.4
1028821 1292016 23.90 0.9 7.15 0.01143 85.0 84.1 77.9
1029521 1292138 0.39 3.5 8.5 0.017399 85.4 81.9 76.9
1029592 1292270 0.59 4.0 6 0.01524 89.8 85.8 83.8
1031457 1292857 0.99 6.0 15 0.0127 94.7 88.7 79.7
1032120 1292890 15.84 8.5 10 0.01524 84.2 75.7 74.2
1031697 1293002 31.68 6.5 13 0.01524 92.1 85.6 79.1
1032695 1293017 6.30 5.0 8.57 0.019812 84.0 79.0 75.4
1032100 1293097 15.84 7.0 14 0.0127 83.9 76.9 69.9
1032413 1293354 23.90 2.5 5 0.0127 84.0 81.5 79.0
1032200 1293914 2.38 1.4 4.28 0.01524 92.7 91.3 88.4
1032230 1294105 0.49 2.9 6.25 0.02032 94.9 92.1 88.7
Continued on the next page
 121

x y Q (m3 ) Level Depth Radius Z top bottom

1016469 1294132 44.93 5.0 4.6 0.0127 62.0 57.0 57.4
1032272 1294384 2.16 2.4 5.69 0.02032 97.4 95.0 91.7
1032071 1294490 5.76 4.7 6 0.01397 96.0 91.3 90.0
1032375 1294755 28.80 2.3 10 0.01397 92.4 90.1 82.4
1032463 1294910 27.65 1.9 10 0.01397 92.2 90.4 82.2
1016684 1295258 4.61 2.7 237 0.003175 61.6 59.0 -175.4
1016452 1295362 26.50 0.7 13 0.001524 64.5 63.8 51.5
1016443 1297743 10.66 2.7 18 0.000762 67.8 65.1 49.8
1016505 1299592 18.72 1.5 10.8 0.001524 66.4 64.9 55.6
1013014 1303611 23.33 1.9 2.6 0.0127 59.7 57.8 57.1
1016392 1304903 9.50 0.9 2.5 0.0127 61.8 60.9 59.3
1019212 1313972 9.22 1.6 3.05 0.0127 57.8 56.2 54.7
1019151 1315474 5.47 1.2 3 0.0127 60.9 59.7 57.9
1016738 1316342 7.20 7.9 27 0.001524 61.3 53.4 34.3
1018227 1316397 6.91 1.5 33 0.001143 58.6 57.1 25.6
1017665 1316929 18.43 3.5 27 0.000762 59.5 56.0 32.5
1017142 1318401 19.87 4.6 6.49 0.0127 71.6 67.1 65.1
 122 A Annex: Initial Data

Observation Points in Transient State

X Y Jan Febr Mar Apr May Jun Jul Aug Sept Oct Nov Dec
996087 1144836 157 154 177 219 232 191 162 151 165 215 238 201
1003087 1144836 146 143 164 203 217 181 154 145 158 203 222 187
1010087 1144836 139 136 154 192 206 176 151 144 156 194 209 176
1017087 1144836 151 148 164 200 215 190 168 160 172 204 214 184
1024087 1144836 167 165 179 213 228 208 188 179 190 217 225 196
989087 1151836 164 162 186 230 244 201 169 157 172 226 251 212
996087 1151836 154 152 175 217 231 191 161 150 165 215 237 199
1003087 1151836 144 142 162 203 217 181 152 143 158 203 222 186
1010087 1151836 135 133 152 191 205 173 146 138 154 193 208 174
1017087 1151836 129 127 144 181 196 168 144 137 151 186 198 165
1024087 1151836 126 124 140 177 192 167 144 137 151 182 192 161
989087 1158836 161 160 183 229 243 200 167 155 172 225 249 210
996087 1158836 152 150 172 216 230 190 158 147 165 214 236 198
1003087 1158836 143 141 162 204 218 181 151 141 159 205 223 186
1010087 1158836 137 135 154 195 209 175 146 138 156 197 212 177
1017087 1158836 131 129 148 188 202 170 143 136 153 191 204 169
1024087 1158836 126 125 143 183 197 168 141 134 151 187 197 164
1031087 1158836 124 123 141 182 197 167 140 133 150 186 196 162
961087 1165836 184 185 212 266 281 236 203 182 196 257 290 247
968087 1165836 181 181 208 260 274 229 195 176 191 252 283 240
975087 1165836 174 175 201 251 265 219 185 168 185 244 273 231
982087 1165836 167 166 191 240 254 209 174 160 178 235 261 220
989087 1165836 158 157 181 228 242 198 164 152 171 225 248 208
996087 1165836 151 149 172 217 231 189 156 145 166 216 236 197
1003087 1165836 144 143 164 207 221 182 150 141 161 208 226 188
1010087 1165836 139 137 158 201 215 178 146 138 159 202 218 181
1017087 1165836 135 134 153 196 211 175 145 137 157 199 212 176
1024087 1165836 132 131 151 195 210 175 145 137 157 198 210 173
1031087 1165836 131 131 152 197 212 176 146 138 158 200 211 173
1038087 1165836 131 132 154 200 215 179 148 140 159 202 213 174
1045087 1165836 133 135 157 205 220 182 152 143 162 206 216 176
961087 1172836 186 187 215 270 285 240 206 186 201 262 294 249
968087 1172836 182 182 209 263 277 231 197 178 196 256 286 242
975087 1172836 175 176 202 253 268 222 186 170 189 248 276 233
982087 1172836 167 167 192 242 256 211 175 160 181 238 264 221
989087 1172836 159 158 182 230 244 199 164 151 174 228 251 209
996087 1172836 151 150 173 219 233 190 155 144 167 218 238 198
1003087 1172836 145 144 166 211 225 183 149 140 163 212 229 190
1010087 1172836 142 141 162 207 222 181 147 138 162 209 225 186
1017087 1172836 139 139 161 207 222 182 148 139 163 210 224 184
1024087 1172836 139 139 162 209 225 185 151 142 165 212 226 185
1031087 1172836 140 141 164 213 229 188 154 145 168 216 229 186
1038087 1172836 141 143 167 217 234 193 159 150 172 220 232 189
1045087 1172836 144 147 171 223 240 198 164 156 178 226 237 192
Continued on the next page
 123

X Y Jan Febr Mar Apr May Jun Jul Aug Sept Oct Nov Dec
954087 1179836 203 205 234 293 307 261 228 205 222 285 317 269
961087 1179836 198 199 227 283 298 253 219 198 216 276 307 261
968087 1179836 191 192 219 273 288 242 207 189 208 267 296 252
975087 1179836 183 183 210 262 277 231 195 178 199 257 285 241
982087 1179836 174 173 199 249 264 217 181 166 189 246 271 228
989087 1179836 162 162 186 235 249 203 166 153 178 233 256 213
996087 1179836 152 151 174 222 236 190 153 142 169 221 241 200
1003087 1179836 147 147 169 216 230 186 149 139 166 217 235 194
1010087 1179836 146 145 168 218 233 188 150 141 168 220 237 194
1017087 1179836 146 146 170 220 237 192 154 145 171 224 240 195
1024087 1179836 147 148 172 224 241 197 159 149 176 229 243 197
1031087 1179836 149 151 175 228 246 201 164 155 181 233 246 199
1038087 1179836 152 155 180 234 253 207 171 162 187 240 251 203
1045087 1179836 156 160 185 241 260 214 178 169 194 246 256 208
1052087 1179836 160 165 191 247 267 220 185 176 200 253 262 212
954087 1186836 231 231 260 319 334 288 255 234 253 315 345 296
961087 1186836 218 219 247 304 319 274 240 219 238 298 329 281
968087 1186836 209 209 235 290 306 261 226 207 227 285 314 269
975087 1186836 197 197 223 276 292 246 210 193 215 272 299 255
982087 1186836 185 184 209 261 276 230 192 177 202 259 284 239
989087 1186836 170 170 194 245 259 212 173 160 187 243 266 222
996087 1186836 155 154 178 227 240 192 153 142 172 226 246 203
1003087 1186836 152 151 176 227 242 193 152 141 172 229 248 202
1010087 1186836 152 152 177 231 249 199 157 146 177 235 254 205
1017087 1186836 153 154 179 235 253 204 162 152 182 241 257 207
1024087 1186836 155 156 182 238 258 209 168 159 188 246 260 209
1031087 1186836 158 160 186 243 264 215 175 166 194 252 264 213
1038087 1186836 162 166 192 250 271 222 183 174 202 258 269 217
1045087 1186836 167 172 198 257 278 229 191 183 209 265 275 222
1052087 1186836 172 178 204 264 285 236 198 191 217 273 280 227
1059087 1186836 177 184 210 271 292 242 206 198 224 280 286 232
961087 1193836 250 249 276 333 350 306 273 253 271 329 359 313
968087 1193836 236 234 261 316 333 289 255 236 255 312 342 296
975087 1193836 221 220 246 300 317 272 235 218 241 298 325 279
982087 1193836 205 203 229 283 300 253 214 199 224 282 308 262
989087 1193836 188 187 212 266 283 234 192 179 208 267 291 244
996087 1193836 166 165 192 248 265 213 166 155 188 251 274 223
1003087 1193836 159 158 186 245 264 209 160 149 184 251 273 218
1010087 1193836 158 159 187 246 267 213 166 155 189 255 274 218
1017087 1193836 160 161 189 249 271 217 172 162 195 259 275 219
1024087 1193836 162 165 192 252 275 223 179 170 201 263 277 222
1031087 1193836 166 170 197 257 280 229 186 178 208 269 281 225
1038087 1193836 171 175 203 264 287 236 195 187 216 276 286 230
1045087 1193836 177 182 209 271 295 243 204 196 224 283 292 236
1052087 1193836 183 190 217 279 303 251 212 205 233 292 298 242
1059087 1193836 190 197 225 287 311 259 221 215 243 301 306 248
Continued on the next page
124 A Annex: Initial Data

X Y Jan Febr Mar Apr May Jun Jul Aug Sept Oct Nov Dec
968087 1200836 269 266 292 346 366 324 290 271 289 343 375 329
975087 1200836 249 246 273 328 349 304 266 248 270 328 358 310
982087 1200836 228 225 253 310 331 282 240 224 250 312 340 289
989087 1200836 204 202 231 291 312 259 211 197 228 295 322 267
996087 1200836 175 174 205 269 292 233 178 164 202 278 304 242
1003087 1200836 164 164 196 262 286 226 170 157 197 274 297 233
1010087 1200836 164 165 196 260 285 227 175 164 201 274 293 231
1017087 1200836 166 168 198 261 286 230 181 172 207 276 292 231
1024087 1200836 169 172 201 264 290 235 189 180 214 280 293 233
1031087 1200836 173 178 206 269 295 241 197 189 221 285 296 237
1038087 1200836 179 184 212 276 301 248 206 199 230 292 301 242
1045087 1200836 186 192 220 284 310 257 216 209 239 300 307 248
1052087 1200836 193 201 228 293 319 266 226 220 249 310 315 255
1059087 1200836 202 210 237 303 329 276 237 231 260 320 324 263
968087 1207836 299 295 320 374 396 357 323 304 320 373 405 360
975087 1207836 274 271 299 355 378 333 294 277 298 356 387 338
982087 1207836 252 250 280 341 364 313 268 252 278 343 374 319
989087 1207836 223 221 255 321 346 287 234 218 252 328 358 295
996087 1207836 183 182 218 291 319 253 189 174 217 305 334 262
1003087 1207836 170 170 205 277 306 240 178 165 209 295 320 247
1010087 1207836 169 171 203 272 301 239 184 173 213 292 311 242
1017087 1207836 170 174 205 272 300 241 190 181 219 293 308 241
1024087 1207836 174 179 208 275 303 246 198 190 226 295 307 243
1031087 1207836 179 185 213 280 307 252 207 200 233 300 310 247
1038087 1207836 185 192 220 286 314 260 217 210 242 307 314 252
1045087 1207836 193 201 228 295 323 269 227 222 253 316 321 259
1052087 1207836 202 210 238 305 333 279 239 233 264 326 330 267
1059087 1207836 213 221 249 316 345 291 251 246 276 338 341 277
1066087 1207836 225 234 261 330 359 303 264 259 290 352 353 288
968087 1214836 286 282 305 354 377 341 313 298 313 359 385 342
975087 1214836 287 284 313 368 389 345 310 296 314 368 398 349
982087 1214836 280 278 311 373 394 341 299 285 308 371 403 348
989087 1214836 260 259 296 364 388 326 273 259 289 366 400 335
996087 1214836 215 215 255 332 360 289 223 207 250 344 378 299
1003087 1214836 175 177 214 292 324 253 186 172 221 316 342 260
1010087 1214836 172 176 210 282 315 249 191 181 224 309 327 251
1017087 1214836 174 179 210 281 312 251 199 190 230 307 321 249
1024087 1214836 178 184 214 283 313 256 207 200 237 309 319 251
1031087 1214836 183 190 219 288 318 262 216 210 245 313 321 255
1038087 1214836 191 198 226 295 324 270 226 221 254 320 325 261
1045087 1214836 199 208 235 304 333 280 237 233 265 329 333 268
1052087 1214836 209 218 245 314 343 290 249 245 276 339 341 277
1059087 1214836 220 229 256 325 355 301 261 257 288 351 352 286
1066087 1214836 232 242 269 339 370 314 274 271 302 366 365 298
968087 1221836 258 254 274 313 335 307 287 280 293 326 341 303
975087 1221836 301 298 326 374 392 352 326 319 330 372 397 354
Continued on the next page
 125

X Y Jan Febr Mar Apr May Jun Jul Aug Sept Oct Nov Dec
982087 1221836 323 321 356 413 429 378 344 335 349 402 434 384
989087 1221836 323 323 364 428 444 382 338 328 348 416 453 393
996087 1221836 285 289 333 406 424 350 293 281 315 404 442 368
1003087 1221836 180 186 225 304 336 259 192 178 232 337 361 271
1010087 1221836 174 180 214 289 325 257 198 189 235 324 340 258
1017087 1221836 176 182 214 286 321 259 206 199 240 320 330 255
1024087 1221836 180 187 217 288 321 263 215 209 247 320 327 257
1031087 1221836 187 195 223 294 325 270 224 220 255 324 329 261
1038087 1221836 195 203 231 301 332 279 235 232 264 331 334 268
1045087 1221836 203 212 239 309 339 287 245 242 273 338 339 274
1052087 1221836 211 221 247 317 347 295 254 252 282 345 346 281
1059087 1221836 221 231 257 326 357 304 264 262 292 355 354 289
1066087 1221836 233 243 269 338 369 315 277 274 305 368 366 299
975087 1228836 320 318 345 385 397 363 348 348 351 377 398 363
982087 1228836 387 385 423 472 479 431 411 409 408 443 478 438
989087 1228836 421 422 468 526 531 469 440 437 438 487 531 483
996087 1228836 377 384 433 500 507 430 386 377 398 477 522 454
1003087 1228836 208 222 262 339 363 277 213 198 258 375 400 303
1010087 1228836 172 181 214 290 330 261 204 196 244 335 345 259
1017087 1228836 174 183 213 287 326 264 212 207 249 328 335 256
1024087 1228836 180 189 218 290 326 269 221 218 255 328 332 260
1031087 1228836 188 197 225 297 330 276 232 229 263 332 333 265
1038087 1228836 196 206 233 304 336 284 242 240 272 337 337 272
1045087 1228836 203 213 239 309 340 289 249 248 278 341 340 276
1052087 1228836 210 220 246 315 344 294 256 255 284 345 344 281
1059087 1228836 218 228 253 321 351 301 263 262 291 352 349 286
1066087 1228836 227 238 262 329 359 309 272 271 300 360 357 293
968087 1235836 201 197 208 233 254 239 233 237 250 263 259 230
975087 1235836 295 291 314 347 361 333 325 328 332 349 361 330
982087 1235836 414 412 449 491 495 452 442 446 436 456 488 457
989087 1235836 567 567 621 672 662 602 592 597 569 592 647 617
996087 1235836 481 484 536 596 597 529 505 506 499 545 593 544
1003087 1235836 326 334 375 442 465 395 356 353 376 445 472 402
1010087 1235836 187 198 229 302 347 280 230 227 270 353 358 270
1017087 1235836 168 177 205 278 324 265 217 215 256 331 330 249
1024087 1235836 177 187 215 287 326 272 227 226 262 331 330 258
1031087 1235836 187 198 225 297 331 280 238 238 270 335 333 267
1038087 1235836 196 207 232 303 334 286 246 246 276 339 336 273
1045087 1235836 201 212 237 306 335 288 250 250 279 339 335 274
1052087 1235836 205 216 240 307 335 289 253 253 280 338 335 275
1059087 1235836 210 221 244 309 336 291 256 256 283 339 336 276
1066087 1235836 215 226 248 311 339 294 260 260 286 342 338 278
1073087 1235836 220 231 251 313 342 296 262 262 289 345 341 279
954087 1242836 159 154 160 180 202 194 187 191 208 219 211 184
961087 1242836 187 182 191 215 237 224 217 220 235 249 244 216
968087 1242836 224 219 233 262 283 264 255 258 270 287 288 258
Continued on the next page
126 A Annex: Initial Data

X Y Jan Febr Mar Apr May Jun Jul Aug Sept Oct Nov Dec
975087 1242836 272 267 287 321 339 314 303 305 314 333 342 311
982087 1242836 332 327 355 395 409 376 362 363 367 390 409 377
989087 1242836 403 398 435 483 492 448 431 431 429 456 487 454
996087 1242836 454 450 495 550 557 503 482 481 475 509 550 513
1003087 1242836 399 398 442 504 522 462 433 431 437 484 517 467
1010087 1242836 263 267 300 367 408 350 311 311 337 396 406 338
1017087 1242836 162 171 196 266 323 268 225 228 265 330 321 241
1024087 1242836 172 183 208 279 321 272 232 235 267 330 322 252
1031087 1242836 186 198 223 294 326 281 243 245 274 334 328 266
1038087 1242836 194 206 230 300 328 284 248 250 277 335 329 272
1045087 1242836 195 207 230 297 322 281 246 249 274 329 324 269
1052087 1242836 196 207 229 293 317 277 244 246 270 323 318 264
1059087 1242836 196 207 227 288 311 272 241 243 266 317 312 259
1066087 1242836 195 205 224 281 305 267 237 238 261 310 306 253
1073087 1242836 193 202 220 273 298 260 231 232 254 303 299 245
954087 1249836 181 175 182 207 228 217 209 212 228 242 238 210
961087 1249836 206 200 211 239 260 246 235 238 253 270 269 239
968087 1249836 235 229 244 276 298 278 266 267 281 301 305 273
975087 1249836 269 262 282 318 339 314 300 300 312 335 344 311
982087 1249836 305 298 323 365 384 353 336 335 345 372 387 353
989087 1249836 341 334 365 412 430 392 372 369 378 409 431 395
996087 1249836 369 361 397 450 468 424 399 394 402 439 467 429
1003087 1249836 376 368 406 467 487 436 406 399 410 454 484 443
1010087 1249836 307 302 337 403 437 384 347 339 363 416 436 381
1017087 1249836 213 213 240 310 356 305 263 258 293 354 357 292
1024087 1249836 202 208 234 304 340 294 255 254 284 343 342 283
1031087 1249836 195 207 231 302 326 287 252 254 279 334 329 277
1038087 1249836 193 206 229 297 317 281 248 252 274 326 320 272
1045087 1249836 189 201 223 287 307 272 241 244 266 315 310 262
1052087 1249836 183 194 214 273 294 260 230 233 254 301 296 249
1059087 1249836 175 185 203 258 278 245 218 220 240 284 280 233
1066087 1249836 167 176 192 241 261 230 205 207 226 266 263 218
1073087 1249836 159 167 181 224 243 215 192 193 211 248 245 202
954087 1256836 195 189 197 224 246 234 224 226 243 259 257 229
961087 1256836 218 211 222 253 275 259 247 249 265 284 285 255
968087 1256836 242 235 249 284 307 287 273 273 288 311 316 283
975087 1256836 268 260 279 318 340 316 299 298 313 339 348 314
982087 1256836 293 286 308 352 374 345 325 322 337 367 380 344
989087 1256836 318 309 336 384 407 372 350 345 359 393 411 374
996087 1256836 335 326 357 411 434 394 367 360 375 414 437 397
1003087 1256836 340 330 364 423 449 404 372 361 380 425 451 409
1010087 1256836 319 306 341 407 437 389 350 335 361 414 439 395
1017087 1256836 253 245 275 344 381 334 292 279 313 372 386 334
1024087 1256836 224 225 254 324 354 311 272 263 292 350 359 307
1031087 1256836 223 231 256 325 348 311 276 274 301 354 355 304
1038087 1256836 208 217 240 303 324 292 260 261 285 333 330 282
Continued on the next page
 127

X Y Jan Febr Mar Apr May Jun Jul Aug Sept Oct Nov Dec
1045087 1256836 191 200 221 279 299 269 240 242 263 307 304 258
1052087 1256836 176 185 203 257 276 248 221 223 243 284 281 236
1059087 1256836 159 167 184 232 250 224 199 201 219 257 254 212
1066087 1256836 141 148 162 204 221 197 176 178 193 226 223 185
1073087 1256836 125 131 142 174 189 169 152 154 167 193 190 158
1080087 1256836 114 120 129 154 168 152 136 138 150 172 171 139
947087 1263836 189 181 186 213 235 226 216 219 236 252 249 222
954087 1263836 207 199 207 237 260 247 235 238 255 274 273 243
961087 1263836 226 219 230 263 287 271 257 258 275 297 299 267
968087 1263836 247 239 253 291 315 295 279 278 295 321 326 292
975087 1263836 268 260 277 318 344 319 300 298 315 344 353 317
982087 1263836 288 279 300 346 371 343 321 317 334 367 379 342
989087 1263836 306 297 321 371 397 364 339 333 351 388 404 364
996087 1263836 319 310 337 391 418 381 353 344 363 405 424 383
1003087 1263836 324 314 344 403 432 391 358 346 368 415 437 394
1010087 1263836 311 298 331 395 427 382 344 328 355 409 432 387
1017087 1263836 261 249 283 352 387 340 297 277 311 371 394 343
1024087 1263836 236 235 266 334 364 323 284 270 301 358 373 318
1031087 1263836 259 261 288 352 377 344 309 304 334 384 388 335
1038087 1263836 239 243 268 329 353 322 290 288 314 361 362 310
1045087 1263836 211 217 239 296 318 289 260 260 283 326 325 276
1052087 1263836 184 189 209 261 281 255 229 229 249 288 287 241
1059087 1263836 157 163 180 225 244 221 197 199 216 250 249 207
1066087 1263836 130 135 148 186 203 183 163 165 179 207 206 170
1073087 1263836 100 104 113 139 151 138 124 126 136 154 153 127
1080087 1263836 82 87 93 111 122 113 101 104 112 125 125 101
947087 1270836 199 191 196 225 249 239 227 231 248 267 265 236
954087 1270836 216 208 216 248 272 259 246 248 266 287 288 256
961087 1270836 234 226 236 272 297 281 265 266 284 309 312 278
968087 1270836 252 244 257 297 323 303 285 284 302 330 336 300
975087 1270836 270 261 278 321 348 324 303 301 320 351 360 322
982087 1270836 287 278 297 344 372 345 321 316 336 371 383 343
989087 1270836 302 292 315 366 395 363 337 330 350 389 404 362
996087 1270836 313 303 329 383 414 379 349 340 361 405 422 378
1003087 1270836 319 309 336 395 427 389 355 343 367 415 435 389
1010087 1270836 314 303 333 396 430 389 351 336 364 417 438 390
1017087 1270836 292 281 313 380 415 373 331 311 345 403 425 373
1024087 1270836 274 270 301 365 396 360 322 306 341 394 410 352
1031087 1270836 299 298 325 387 414 384 350 343 375 422 428 371
1038087 1270836 271 272 299 360 387 357 324 320 348 394 397 340
1045087 1270836 239 242 267 326 352 322 291 289 314 358 360 304
1052087 1270836 208 211 234 289 314 286 257 256 277 319 320 266
1059087 1270836 176 179 199 249 273 247 221 220 239 277 279 228
1066087 1270836 144 147 165 209 231 208 185 184 201 235 237 189
1073087 1270836 114 117 131 168 189 169 149 149 163 191 193 150
1080087 1270836 111 113 133 190 225 192 166 166 186 228 231 163
Continued on the next page
128 A Annex: Initial Data

X Y Jan Febr Mar Apr May Jun Jul Aug Sept Oct Nov Dec
961087 1277836 242 232 243 281 308 291 274 274 293 320 324 288
968087 1277836 258 249 262 303 332 311 291 290 310 341 346 309
975087 1277836 274 264 280 325 355 331 308 305 326 360 368 328
982087 1277836 289 279 298 347 377 350 324 319 341 379 390 347
989087 1277836 303 293 314 366 398 367 339 332 354 396 409 365
996087 1277836 314 303 327 383 416 382 351 341 365 411 427 380
1003087 1277836 321 311 337 396 430 394 360 347 373 423 440 392
1010087 1277836 325 313 341 404 439 402 364 349 377 431 450 400
1017087 1277836 327 316 345 409 446 408 367 350 383 439 458 405
1024087 1277836 420 409 431 481 509 493 460 453 494 529 535 483
1031087 1277836 349 344 374 434 463 436 402 396 429 473 477 418
1038087 1277836 300 299 330 394 423 392 358 353 381 428 431 369
1045087 1277836 263 263 293 356 386 353 320 316 341 388 391 328
1052087 1277836 229 230 258 317 347 315 284 281 303 348 351 290
1059087 1277836 197 198 223 279 308 277 248 245 266 309 312 252
1066087 1277836 167 168 190 242 272 243 215 212 232 273 278 217
1073087 1277836 139 140 161 212 245 215 186 183 204 247 253 188
1080087 1277836 122 123 141 188 218 190 165 163 182 220 225 166
961087 1284836 250 239 249 289 318 301 282 283 303 332 336 299
968087 1284836 265 255 267 311 341 320 299 298 319 352 358 318
975087 1284836 280 269 285 331 364 339 315 312 334 370 379 336
982087 1284836 294 283 301 352 385 357 330 325 348 388 399 354
989087 1284836 307 296 317 370 405 374 344 337 361 405 418 371
996087 1284836 318 308 330 387 423 390 357 347 373 420 435 386
1003087 1284836 328 317 342 402 439 404 368 355 383 434 450 400
1010087 1284836 336 325 351 414 452 416 377 362 391 446 464 412
1017087 1284836 347 337 364 429 467 431 390 374 405 461 480 426
1024087 1284836 398 388 416 475 507 482 446 436 471 516 525 468
1031087 1284836 373 368 403 469 500 469 434 429 459 505 508 443
1038087 1284836 316 315 352 420 452 416 382 376 401 452 455 387
1045087 1284836 277 277 311 376 408 372 339 334 357 406 410 343
1052087 1284836 243 243 274 335 365 331 300 295 317 364 368 304
1059087 1284836 212 213 240 296 326 294 264 260 280 325 330 268
1066087 1284836 184 185 209 260 289 259 231 227 247 289 294 234
1073087 1284836 159 161 180 224 251 224 199 196 214 252 257 202
1080087 1284836 137 139 153 190 214 191 169 167 183 216 221 173
954087 1291836 243 232 239 278 307 292 275 278 298 326 329 291
961087 1291836 258 247 257 299 330 312 292 292 314 345 349 310
968087 1291836 273 262 274 319 352 330 308 306 329 364 370 328
975087 1291836 287 276 291 339 374 349 323 319 343 382 390 346
982087 1291836 300 290 307 359 395 367 338 332 357 400 410 363
989087 1291836 313 302 323 377 415 384 352 344 370 417 429 379
996087 1291836 325 314 337 395 434 400 366 355 382 433 447 395
1003087 1291836 337 326 350 411 451 415 378 365 394 447 463 410
1010087 1291836 348 337 363 427 467 430 391 375 405 462 479 424
1031087 1291836 341 340 388 468 504 458 421 413 434 493 497 421
Continued on the next page
 129

X Y Jan Febr Mar Apr May Jun Jul Aug Sept Oct Nov Dec
1038087 1291836 308 308 351 425 458 416 381 374 395 450 454 382
1045087 1291836 278 278 315 380 412 374 341 335 356 407 411 344
1052087 1291836 248 248 280 339 369 334 304 299 318 365 370 307
1059087 1291836 220 221 248 300 328 296 269 264 282 326 331 272
1066087 1291836 195 197 219 264 289 261 236 232 249 288 293 240
1073087 1291836 176 178 194 231 252 229 208 204 220 254 259 213
1080087 1291836 156 158 170 198 216 197 179 176 190 219 223 185
954087 1298836 253 241 248 288 320 304 286 289 310 340 343 304
961087 1298836 267 255 265 309 342 323 302 302 325 359 363 321
968087 1298836 282 270 282 329 364 342 318 316 340 377 384 339
975087 1298836 295 284 299 349 386 360 333 329 354 395 404 356
982087 1298836 309 298 315 368 407 378 348 341 368 413 423 373
989087 1298836 322 311 331 387 428 395 362 354 381 430 442 390
996087 1298836 335 324 346 405 448 412 376 365 394 447 461 406
1003087 1298836 347 336 360 423 467 429 390 377 406 463 478 422
1010087 1298836 361 350 376 441 485 446 405 389 419 479 496 439
1017087 1298836 376 366 393 461 504 466 423 406 435 496 515 458
961087 1305836 277 265 274 320 355 335 313 313 337 373 378 334
968087 1305836 291 279 291 340 378 354 328 327 352 392 398 351
975087 1305836 306 294 308 360 400 373 344 340 366 410 418 369
982087 1305836 319 308 325 380 422 391 359 353 380 428 438 385
989087 1305836 333 321 341 399 443 409 374 365 394 446 458 402
996087 1305836 346 335 357 418 464 427 389 377 407 463 477 419
1003087 1305836 360 349 373 437 485 445 404 390 421 481 496 436
1010087 1305836 374 364 390 457 506 464 420 404 435 498 515 453
1017087 1305836 389 379 407 476 523 482 438 419 448 513 534 472
968087 1312836 303 290 302 353 393 368 341 339 365 407 414 365
1010087 1312836 390 379 407 475 529 482 436 420 453 520 536 469
1017087 1312836 405 394 424 495 550 502 453 434 467 538 556 487
 130 A Annex: Initial Data

Hydraulic Data - Pumping Test

In this section, indicates the pumping test information from six wells. The specific yield and
transmissivity parameters were estimated using Theis and Neuman methods.

San Antonio Well

Las Mercedes Well

 131

Penjamo Well

La Ponderosa Well
 132 A Annex: Initial Data

Campo Seco Well

Socorro Well
B. Annex: Statistical Analysis of the
Hydrological Model

AMA and Sobol’ Indices Series for each Parameter

Parameters AM AE AM AV AM Aγ AM AK Sobol’
qso 0.00785132 0.02195257 0.01612656 0.0572445 0.00018611
LnT e 0.00711278 0.02082543 0.01589391 0.05800311 0.00017285
m 0.00879224 0.02077318 0.01395075 0.05217205 0.00023078
Sro 0.00731151 0.01876407 0.01389315 0.04903818 0.00018132
Srmax 0.00703534 0.02425542 0.01281027 0.05354968 0.00015073
td 0.00720454 0.02285963 0.01409881 0.05340454 0.0002027
vch 0.00898599 0.02904335 0.01415603 0.05518942 0.0002894
vr 0.00808449 0.01829092 0.01587576 0.04834258 0.00019543
ko 0.0079966 0.0222415 0.01330691 0.05682577 0.00017769
CD 0.00691153 0.02106544 0.0096962 0.03976828 0.00015061

Analysis of the Four Statistical Moments

Day Mean Variance Skewness Kurtosis

1 1.80E-03 1.09E-06 3.29E-03 1.80E+00
150 6.21E-03 4.35E-06 4.48E-01 4.48E+00
300 6.76E-03 1.41E-06 1.79E+00 5.84E+00
1000 3.77E-03 1.25E-06 2.16E+00 6.76E+00
1250 4.66E-03 1.78E-06 2.00E+00 6.02E+00
1500 2.28E-03 8.99E-08 1.37E+00 4.91E+00
2000 7.02E-03 9.61E-07 1.94E+00 5.93E+00
2600 2.03E-03 1.07E-07 1.24E+00 5.63E+00
3000 2.00E-03 9.92E-07 3.44E-01 3.44E+00
 134 B Annex: Statistical Analysis of the Hydrological Model

Comparison of Index Values by Decomposition and Integral

Descomposition Integral Descomposition Integral

Index day Index
qso m Srmax qso m Srmax qso m Srmax qso m Srmax
1 0.4993 0.0055 0.0053 0.4528 0.0053 0.0052 3.7846 3.6115 3.2578 3.7352 3.2703 3.1475
150 0.1297 0.0560 0.1859 0.1114 0.0507 0.1610 3.1867 0.1590 5.5022 3.0052 0.1488 5.1229
300 0.0063 0.0043 0.1240 0.0050 0.0038 0.1085 0.1192 0.0491 3.5936 0.1005 0.0445 3.4480
1000 0.0029 0.0167 0.2061 0.0027 0.0153 0.1781 0.0149 0.0130 0.9485 0.0139 0.0121 0.9016
AM AE 1250 0.0028 0.0368 0.2013 0.0026 0.0335 0.1749 AM Aγ 0.0156 0.0300 0.8890 0.0146 0.0294 0.8419
1500 0.0011 0.0565 0.0839 0.0011 0.0511 0.0735 0.0197 0.3240 1.0517 0.0190 0.3084 1.0003
2000 0.0013 0.0025 0.0971 0.0012 0.0023 0.0844 0.0149 0.0137 0.9693 0.0138 0.0127 0.9213
2600 0.0012 0.0675 0.0963 0.0012 0.0602 0.0836 0.0282 0.6934 2.0245 0.0267 0.6618 1.9243
3000 0.0037 0.3367 0.1868 0.0035 0.3108 0.1615 0.0467 4.5832 4.2107 0.0438 4.4152 4.0070
1 0.9975 0.0084 0.0083 0.9470 0.0081 0.0080 0.0066 0.0063 0.0058 0.0060 0.0059 0.0057
150 0.3539 0.0470 0.6126 0.3308 0.0422 0.5774 0.3058 0.0473 0.9649 0.2789 0.0427 0.7772
300 0.0655 0.0369 0.9382 0.0494 0.0328 0.8925 0.0254 0.0289 11.8009 0.0232 0.0278 11.3524
1000 0.0200 0.0745 0.9736 0.0187 0.0687 0.9277 0.0234 0.0226 0.6355 0.0218 0.0208 0.6034
AM AV 1250 0.0185 0.0342 0.9653 0.0174 0.0333 0.9183 AM AK 0.0243 0.0300 0.6599 0.0227 0.0297 0.6273
1500 0.0171 0.2464 0.7295 0.0164 0.2351 0.6998 0.0215 0.0930 0.6041 0.0204 0.0891 0.5748
2000 0.0182 0.0681 0.9456 0.0174 0.0602 0.9006 0.0212 0.0234 0.5944 0.0195 0.0214 0.5662
2600 0.0187 0.2505 0.6910 0.0180 0.2460 0.6641 0.0185 0.1779 0.3127 0.0175 0.1728 0.2968
3000 0.0177 0.6246 0.5384 0.0169 0.5864 0.4846 0.0098 0.9353 0.3044 0.0094 0.9009 0.2880

1 1.0448 0.0001 0.0001 1.0448 0.0001 0.0001

150 0.3100 0.0389 0.6450 0.3100 0.0389 0.6450
300 0.0051 0.0010 0.9885 0.0051 0.0010 0.9885
1000 0.0001 0.0043 1.0300 0.0001 0.0043 1.0300
Sobol 1250 0.0001 0.0228 1.0200 0.0001 0.0228 1.0200
1500 0.0001 0.2600 0.7686 0.0001 0.2600 0.7686
2000 0.0001 0.0005 0.9960 0.0001 0.0005 0.9960
2600 0.0001 0.2645 0.7279 0.0001 0.2645 0.7279
3000 0.0001 0.6579 0.2871 0.0001 0.6579 0.2871
C. Annex: Hydrogeologic Model

Results Regularization Methods

 Observation
Points X Y Measure CZ 40 -PP 100 -PP 200 -PP 136
SVD Tikh. SVD-a SVD Tikh. SVD-a SVD Tikh. SVD-a SVD Tikh. SVD-a
Coyote 1059674 1265494 405.174 406.89 406.89 406.89 406.89 406.89 406.89 406.89 406.89 406.89 406.89 406.89 406.89
12 Oct 1032200 1293914 90.9937 92.2 92.32 92.31 92.3 92.42 92.43 92.24 92.36 92.33 92.31 92.55 92.45
Pte. Sogamoso 1032212 1293128 69.3122 69.99 69.99 69.99 69.98 69.99 69.99 69.98 69.99 69.99 70.17 70.03 70.01
Bello Monte 2 1025355 1289443 68.5806 68.96 68.96 68.96 68.96 68.96 68.97 68.96 68.96 68.96 69.11 69 68.98
Bello Monte 1024625 1289121 125.35 127.01 127.01 127.01 127 127.01 127 127 127 127 127 127 127
Bocas de Sog. 1019662 1287460 148.262 148.85 148.85 148.85 148.85 148.85 148.85 148.85 148.85 148.85 148.85 148.85 148.85
Bufalera 1032060 1292207 171.955 172.27 172.27 172.27 172.27 172.27 172.27 172.27 172.27 172.27 172.27 172.27 172.27
Ponderosa 988823 1207855 127.134 125.62 125.62 125.62 125.62 125.62 125.62 125.62 125.62 125.62 125.62 125.62 125.62
Cantagallo 1040376 1295548 117.674 115.86 115.86 115.86 115.86 115.86 115.86 115.86 115.86 115.86 115.86 115.86 115.86
Campo Seco 975497 1184466 113.178 114.24 114.24 114.24 114.24 114.24 114.24 114.24 114.24 114.24 114.24 114.24 114.24
Coyote 1047880 1256558 113.599 113 113 113 113 113 113 113 113 113 113 113 113
Coyote 1042872 1288351 112.158 112.12 112.12 112.12 112.12 112.12 112.12 112.12 112.12 112.12 112.12 112.12 112.12
Coyote 1037608 1286282 114.336 114.8 114.8 114.8 114.8 114.8 114.8 114.8 114.8 114.8 114.8 114.8 114.8
Coyote 1026557 1259046 114.039 114.57 114.57 114.57 114.57 114.57 114.57 114.57 114.57 114.57 114.57 114.57 114.57
Coyote 1055225 1268549 97.6181 96.59 96.59 96.59 96.59 96.59 96.59 96.59 96.59 96.59 96.59 96.59 96.59
Coyote 1035630 1256907 101.218 101.23 101.23 101.23 101.23 101.23 101.23 101.23 101.23 101.23 101.23 101.23 101.23
Coyote 1052395 1265100 106.195 107.01 107.01 107.01 107.01 107.01 107.01 107.01 107.01 107.01 107.01 107.01 107.01
Coyote 1047276 1271021 93.9767 93.85 93.85 93.85 93.85 93.85 93.85 93.85 93.85 93.85 93.85 93.85 93.85
Coyote 1049411 1267467 131 130.48 130.48 130.48 130.48 130.48 130.48 130.48 130.48 130.48 130.48 130.48 130.48
Coyote 1053508 1284723 101.333 102.67 102.67 102.67 102.67 102.67 102.67 102.67 102.67 102.67 102.67 102.67 102.67
Coyote 1038895 1261988 93.604 93.66 93.66 93.66 93.66 93.66 93.66 93.66 93.66 93.66 93.66 93.66 93.66
Coyote 1032195 1249690 84.9609 85.01 85.01 85.01 85.01 85.01 85.01 85.01 85.01 85.01 85.01 85.01 85.01
Coyote 1046672 1250824 92.2162 92.94 92.94 92.94 92.94 92.94 92.94 92.94 92.94 92.94 92.94 92.94 92.94
Coyote 1045435 1271996 82.3175 82.52 82.52 82.52 82.52 82.52 82.52 82.52 82.52 82.52 82.52 82.52 82.52
Coyote 1039207 1289151 102.894 102.1 102.09 102.1 102.09 102.08 102.1 102.07 102.11 102.11 102.09 102.11 102.12
Coyote 1035433 1283961 97.7521 98.72 98.72 98.72 98.72 98.72 98.72 98.72 98.72 98.72 98.72 98.72 98.72
Coyote 1039398 1279164 100.019 101.51 101.51 101.51 101.51 101.51 101.51 101.51 101.51 101.51 101.51 101.51 101.51
Coyote 1041091 1275895 86.4685 87.17 87.17 87.17 87.17 87.17 87.17 87.17 87.17 87.17 87.17 87.17 87.17
Coyote 1033985 1250501 208.854 210.72 210.72 210.72 210.72 210.72 210.72 210.72 210.72 210.72 210.72 210.72 210.72
Coyote 1042059 1259841 87.531 87.5 87.5 87.5 87.5 87.5 87.5 87.5 87.5 87.5 87.5 87.5 87.5
Coyote 1048583 1270162 88.025 88.51 88.51 88.51 88.51 88.51 88.51 88.51 88.51 88.51 88.51 88.51 88.51
Continued on the next page
C Annex: Hydrogeologic Model
Observation
Points X Y Measure CZ 40 -PP 100 -PP 200 -PP
SVD Tikh. SVD-a SVD Tikh. SVD-a SVD Tikh. SVD-a SVD Tikh. SVD-a
Coyote 1048518 1274638 71.3773 73.04 73.04 73.04 73.04 73.04 73.04 73.04 73.04 73.04 73.04 73.04 73.04
Coyote 1045559 1245147 128.733 129.24 129.24 129.24 129.24 129.24 129.24 129.24 129.24 129.24 129.24 129.24 129.24
Coyote 1053609 1279361 208.72 209.81 209.81 209.81 209.81 209.81 209.81 209.81 209.81 209.81 209.81 209.81 209.81
Coyote 1043381 1265883 130.401 129.55 129.55 129.55 129.55 129.55 129.55 129.55 129.55 129.55 129.55 129.55 129.55
Coyote 1035018 1266105 138.213 138.82 138.82 138.82 138.82 138.82 138.82 138.82 138.82 138.82 138.82 138.82 138.82
Coyote 1026421 1283408 116.813 118.17 118.17 118.17 118.17 118.17 118.17 118.17 118.17 118.17 118.17 118.17 118.17
Coyote 1057996 1274691 113.216 113.6 113.6 113.6 113.6 113.6 113.6 113.6 113.6 113.6 113.6 113.6 113.6
Coyote 1047880 1256558 100.028 100.43 100.43 100.43 100.43 100.43 100.43 100.43 100.43 100.43 100.43 100.43 100.43
Coyote 1052346 1246725 103 103.87 103.87 103.87 103.87 103.87 103.87 103.87 103.87 103.87 103.87 103.87 103.87
Coyote 1042036 1255430 123.697 123.7 123.7 123.7 123.7 123.7 123.7 123.7 123.7 123.7 123.7 123.7 123.7
Coyote 1042944 1283661 76.9657 76.82 76.82 76.82 76.82 76.82 76.82 76.82 76.82 76.82 76.82 76.82 76.82
Coyote 1052631 1242242 86.1388 86 86.19 86.18 86.06 86.15 86.16 85.98 86.04 85.96 85.7 86.37 86.13
Coyote 1055345 1258333 161.685 162.12 162.12 162.12 162.11 162.12 162.11 162.09 162.09 162.09 162.12 162.11 162.11
Coyote 1039884 1235905 70.0088 68.27 68.22 68.22 68.58 69.82 69.86 68.59 69.83 69.7 68.28 69.78 70.05
Coyote 1037860 1249611 70.1585 71.29 71.29 71.29 71.29 71.29 71.29 71.29 71.29 71.29 71.29 71.29 71.29
Coyote 1055462 1271893 132.943 136.23 136.22 136.22 136.2 136.16 133.67 136.18 132.99 136.15 136.18 133.15 132.97
Coyote 1029556 1268887 150.126 146.61 147.28 147.28 146.95 150.19 150.31 146.74 150.11 150.27 146.75 150.13 150.16
Coyote 1030678 1286282 74.8362 76.58 76.58 76.58 76.58 76.58 76.58 76.58 76.58 76.58 76.58 76.58 76.58
Coyote 1031122 1242879 83.9126 84.08 84.08 84.08 84.08 84.08 84.08 84.08 84.08 84.08 84.08 84.08 84.08
Coyote 1052300 1266521 84.0521 85.17 85.16 85.16 84.71 84.18 84.11 84.69 84.24 84.25 84.41 84.08 84.14
Coyote 1025074 1250409 69.3534 69.8 69.8 69.8 69.8 69.8 69.8 69.8 69.8 69.8 69.8 69.8 69.8
Coyote 1030907 1265404 69.1798 70.53 70.53 70.53 70.53 70.53 70.53 70.53 70.53 70.53 70.53 70.53 70.53
Coyote 1031734 1286051 68.9996 70.5 70.5 70.5 70.5 70.5 70.5 70.5 70.5 70.5 70.5 70.5 70.5
Coyote 1033897 1242729 69.0886 70.28 70.28 70.28 70.28 70.28 70.28 70.28 70.28 70.28 70.28 70.28 70.28
Coyote 1048130 1278346 69.4922 69.62 69.62 69.62 69.62 69.62 69.62 69.62 69.62 69.62 69.62 69.62 69.62
Coyote 1049641 1255613 73.7402 72.5 72.51 72.51 72.52 72.53 72.52 72.51 72.55 72.55 72.56 72.56 72.56
Coyote 1039477 1269166 69.6045 70.54 70.54 70.54 70.53 70.53 70.53 70.53 70.53 70.53 70.53 70.53 70.53
Coyote 1030522 1288586 70.5502 71.06 71.08 71.08 71.01 70.98 70.98 71 70.95 70.95 71.11 71.04 70.99
Coyote 1035506 1288469 95.3403 94.21 94.35 94.34 94.5 95.25 95.24 94.42 95.14 95.09 94.59 94.99 95.01
Coyote 1025846 1264554 94.0048 93.3 93.45 93.45 93.47 93.67 93.68 93.41 93.57 93.54 93.53 93.84 93.75
Coyote 1053472 1281422 70.4663 70.55 70.55 70.55 70.55 70.55 70.55 70.55 70.55 70.55 70.55 70.55 70.55
137

Coyote 1060084 1280926 104.869 104.45 104.45 104.45 104.45 104.45 104.45 104.45 104.45 104.45 104.45 104.45 104.45
Continued on the next page
 Observation
Points X Y Measure CZ 40 -PP 100 -PP 200 -PP
138
SVD Tikh. SVD-a SVD Tikh. SVD-a SVD Tikh. SVD-a SVD Tikh. SVD-a
Coyote 1043211 1246611 96.0018 97.17 97.17 97.17 97.17 97.17 97.17 97.17 97.17 97.17 97.17 97.17 97.17
Coyote 1028977 1256175 154.834 155.97 155.97 155.97 155.97 155.97 155.97 155.97 155.97 155.97 155.97 155.97 155.97
El esfuerzo 1024923 1289333 81.0006 81.32 81.32 81.32 81.32 81.32 81.32 81.32 81.32 81.32 81.32 81.32 81.32
el progreso 2 1029521 1292138 110.517 110.43 110.43 110.43 110.43 110.43 110.43 110.43 110.43 110.43 110.43 110.43 110.43
el progreso 3 1029592 1292270 145.581 145.9 145.9 145.9 145.9 145.9 145.9 145.9 145.9 145.9 145.9 145.9 145.9
Hda Ponderosa 991395 1198951 126.355 127.48 127.48 127.48 127.48 127.48 127.48 127.48 127.48 127.48 127.48 127.48 127.48
Escuela la lucha 1022847 1288743 94.303 94.14 94.14 94.14 94.14 94.14 94.14 94.14 94.14 94.14 94.14 94.14 94.14
Guane 1018862 1282366 87.0677 87.58 87.58 87.58 87.58 87.58 87.58 87.58 87.58 87.58 87.58 87.58 87.58
Guane 1020430 1287299 155.274 156.91 156.91 156.91 156.91 156.91 156.91 156.91 156.91 156.91 156.91 156.91 156.91
Las Mercedes 989394 1209989 83.0301 83.4 83.4 83.4 83.41 83.44 83.42 83.4 83.44 83.43 83.4 83.39 83.44
Penjamo 978367 1196624 79.8569 79.8 79.8 79.8 79.8 79.8 79.8 79.8 79.8 79.8 79.8 79.8 79.8
San Antonio 978474 1208615 82.0645 80.99 81.18 81.17 81.36 81.07 81.06 81.28 81.1 81.14 81.34 81.22 81.2
Hortensia 1031873 1291928 89.9279 88.9 88.92 88.92 88.94 88.88 88.88 88.92 88.88 88.89 88.91 88.89 88.86
La cascadera 1032695 1293017 85.4097 83.64 83.69 83.69 83.74 83.62 83.62 83.69 83.61 83.63 83.68 83.63 83.57
la ceiba 1022068 1288246 81.6372 81.15 81.58 81.56 81.93 81.4 81.4 81.8 81.46 81.54 81.9 81.74 81.69
la esmeralda 1022071 1288244 83 84.58 84.75 84.75 84.93 84.4 84.41 84.73 84.37 84.43 84.66 84.46 84.2
la esperanza 2 1031457 1292857 83.2303 83.39 83.64 83.63 83.89 83.24 83.25 83.65 83.22 83.32 83.57 83.31 83.01
La felicidad 1018987 1288157 84.6765 83.89 84.3 84.27 84.69 84.55 84.55 84.68 84.57 84.9 84.63 84.48 84.56
La felicidad 1019017 1288214 80.3412 79.93 79.93 79.93 79.93 79.93 79.94 79.92 79.88 79.88 79.89 79.88 79.88
La felicidad 2 1019012 1288214 81.5892 81.9 82.09 82.07 82.25 82.2 82.2 82.25 82.14 82.29 82.2 82.12 82.15
La lucha 1022205 1288404 82.1163 82.1 82.09 82.09 82.09 82.09 82.09 82.1 82.24 82.2 82.2 82.24 82.28
La lucha 4 1024111 1288555 87.318 87.19 87.18 87.18 87.17 87.16 87.15 87.2 87.23 87.21 87.22 87.22 87.24
La lucha 2 1022464 1288569 82.2149 80.85 80.85 80.85 80.85 80.85 80.85 80.85 80.85 80.85 80.85 80.85 80.85
La lucha 3 1022673 1288694 91.9346 91.33 91.29 91.3 91.26 91.27 91.26 91.26 91.28 91.26 91.27 91.29 91.29
la lucha 5 1023170 1288895 90.483 90.16 90.16 90.16 90.15 90.15 90.15 90.15 90.14 90.14 90.14 90.14 90.14
La lucha 6 1023972 1289417 90.0462 90.31 90.31 90.31 90.31 90.31 90.31 90.31 90.31 90.31 90.31 90.31 90.31
Las delicias 1024076 1288522 618.984 619.08 619.08 619.08 619.08 619.08 619.08 619.08 619.08 619.08 619.08 619.08 619.08
los mogalos 1032272 1294384 638.293 638.96 638.96 638.96 638.96 638.96 638.96 638.96 638.96 638.96 638.96 638.96 638.96
los mogalos 2 1032230 1294105 607.704 608.31 608.31 608.31 608.31 608.31 608.31 608.31 608.31 608.31 608.31 608.31 608.31
lucha 1022539 1288505 109.741 111.68 111.68 111.68 111.68 111.68 111.68 111.68 111.68 111.68 111.68 111.68 111.68
M-0007-14-101 1052159 1266329 266.269 266.57 266.57 266.57 266.57 266.57 266.57 266.57 266.57 266.57 266.57 266.57 266.57
M-0007-14-109 1032536 1286180 227.828 228.44 228.44 228.44 228.44 228.44 228.44 228.44 228.44 228.44 228.44 228.44 228.44
C Annex: Hydrogeologic Model

Continued on the next page

 Observation
Points X Y Measure CZ 40 -PP 100 -PP 200 -PP
SVD Tikh. SVD-a SVD Tikh. SVD-a SVD Tikh. SVD-a SVD Tikh. SVD-a
M-0007-14-11 1053689 1281102 317.976 319.24 319.24 319.24 319.24 319.24 319.24 319.24 319.24 319.24 319.24 319.24 319.24
M-0007-14-119 1050456 1280052 307.893 309.42 309.42 309.42 309.42 309.42 309.42 309.42 309.42 309.42 309.42 309.42 309.42
M-0007-14-120 1050652 1280092 313.992 315.77 315.77 315.77 315.77 315.77 315.77 315.77 315.77 315.77 315.77 315.77 315.77
M-0007-14-125 1042828 1268734 307.298 307.85 307.85 307.85 307.85 307.85 307.85 307.85 307.85 307.85 307.85 307.85 307.85
M-0007-14-129 1049540 1255546 257.57 257.73 257.73 257.73 257.73 257.73 257.73 257.73 257.73 257.73 257.73 257.73 257.73
M-0007-14-137 1055198 1261560 305.329 307.19 307.19 307.19 307.19 307.19 307.19 307.19 307.19 307.19 307.19 307.19 307.19
M-0007-14-140 1056335 1261488 304.622 304.86 304.86 304.86 304.86 304.86 304.86 304.86 304.86 304.86 304.86 304.86 304.86
M-0007-14-141 1056903 1263197 243.602 243.71 243.71 243.71 243.71 243.71 243.71 243.71 243.71 243.71 243.71 243.71 243.71
M-0007-14-147 1055957 1281024 239.587 239.35 239.35 239.35 239.35 239.35 239.35 239.35 239.35 239.35 239.35 239.35 239.35
M-0007-14-148 1054881 1281862 243.01 243.47 243.47 243.47 243.47 243.47 243.47 243.47 243.47 243.47 243.47 243.47 243.47
M-0007-14-15 1030162 1285752 242.506 243.42 243.42 243.42 243.42 243.42 243.42 243.42 243.42 243.42 243.42 243.42 243.42
M-0007-14-19 1029372 1286160 245.35 244.78 244.78 244.78 244.78 244.78 244.78 244.78 244.78 244.78 244.78 244.78 244.78
M-0007-14-47 1027008 1282516 255.679 255.34 255.34 255.34 255.34 255.34 255.34 255.34 255.34 255.34 255.34 255.34 255.34
M-0007-14-73 1025299 1279402 236.991 236.26 236.26 236.26 236.26 236.26 236.26 236.26 236.26 236.26 236.26 236.26 236.26
M-0007-14-9 1048986 1281049 252.417 252.6 252.6 252.6 252.6 252.6 252.6 252.6 252.6 252.6 252.6 252.6 252.6
M-0033-14-192 971967 1176580 253.787 253.52 253.52 253.52 253.52 253.52 253.52 253.52 253.52 253.52 253.52 253.52 253.52
M-2249-314 1027563 1286294 265.172 263.75 263.75 263.75 263.75 263.75 263.75 263.75 263.75 263.75 263.75 263.75 263.75
M-2249-315 1027268 1286107 263.24 263.97 263.97 263.97 263.97 263.97 263.97 263.97 263.97 263.97 263.97 263.97 263.97
M-2249-316 1027920 1286287 267.988 268.85 268.85 268.85 268.85 268.85 268.85 268.85 268.85 268.85 268.85 268.85 268.85
M-2249-317 1027618 1285767 236.52 237.09 237.09 237.09 237.09 237.09 237.09 237.09 237.09 237.09 237.09 237.09 237.09
M-2249-318 1027584 1285560 211.143 211.44 211.44 211.44 211.44 211.44 211.44 211.44 211.44 211.44 211.44 211.44 211.44
M-2249-319 1027939 1286126 264.601 265.29 265.29 265.29 265.29 265.29 265.29 265.29 265.29 265.29 265.29 265.29 265.29
M-2249-320 1028131 1285705 241.519 242.33 242.33 242.33 242.33 242.33 242.33 242.33 242.33 242.33 242.33 242.33 242.33
M-2249-321 1028112 1285781 234.233 234.05 234.05 234.05 234.05 234.05 234.05 234.05 234.05 234.05 234.05 234.05 234.05
M-2249-322 1027789 1285421 237.009 237.44 237.44 237.44 237.44 237.44 237.44 237.44 237.44 237.44 237.44 237.44 237.44
M-2249-323 1027557 1284818 244.632 244.94 244.94 244.94 244.94 244.94 244.94 244.94 244.94 244.94 244.94 244.94 244.94
M-2249-324 1027775 1284202 193.058 193.39 193.39 193.39 193.39 193.39 193.39 193.39 193.39 193.39 193.39 193.39 193.39
M-2249-325 1027608 1284348 281.691 281.92 281.92 281.92 281.92 281.92 281.92 281.92 281.92 281.92 281.92 281.92 281.92
M-2249-326 1025426 1282796 271.56 271.58 271.58 271.58 271.58 271.58 271.58 271.58 271.58 271.58 271.58 271.58 271.58
M-2249-327 1025549 1282686 266.915 268.68 268.68 268.68 268.68 268.68 268.68 268.68 268.68 268.68 268.68 268.68 268.68
M-2249-328 1025664 1282748 259.921 259.95 259.95 259.95 259.95 259.95 259.95 259.95 259.95 259.95 259.95 259.95 259.95
139

M-2249-329 1026367 1283144 251.541 251.34 251.34 251.34 251.34 251.34 251.34 251.34 251.34 251.34 251.34 251.34 251.34
Continued on the next page
Observation
Points X Y Measure CZ 40 -PP 100 -PP 200 -PP
140
SVD Tikh. SVD-a SVD Tikh. SVD-a SVD Tikh. SVD-a SVD Tikh. SVD-a
M-2249-330 1026451 1282824 250.217 250.92 250.92 250.92 250.92 250.92 250.92 250.92 250.92 250.92 250.92 250.92 250.92
M-2249-332 1026638 1282968 240.032 240.48 240.48 240.48 240.48 240.48 240.48 240.48 240.48 240.48 240.48 240.48 240.48
M-2249-63 1027584 1285560 237.513 237.67 237.67 237.67 237.67 237.67 237.67 237.67 237.67 237.67 237.67 237.67 237.67
M-4236-2 1061332 1264005 232.682 232.12 232.12 232.12 232.12 232.12 232.12 232.12 232.12 232.12 232.12 232.12 232.12
M-4236-3 1061281 1264074 183.13 183.03 183.03 183.03 183.03 183.03 183.03 183.03 183.03 183.03 183.03 183.03 183.03
M-4236-4 1061120 1264241 258.561 258.47 258.47 258.47 258.47 258.47 258.47 258.47 258.47 258.47 258.47 258.47 258.47
M-4236-5 1061070 1264389 278.079 278.58 278.58 278.58 278.58 278.58 278.58 278.58 278.58 278.58 278.58 278.58 278.58
M-4236-6 1061551 1263940 315.63 315.79 315.79 315.79 315.79 315.79 315.79 315.79 315.79 315.79 315.79 315.79 315.79
M-4644-4 1038432 1295689 108.473 108.33 108.33 108.33 108.33 108.33 108.33 108.33 108.33 108.33 108.33 108.33 108.33
M-5237-1 1040640 1225176 81.8296 80.84 80.91 80.9 80.78 80.71 80.7 80.75 80.67 80.65 80.62 80.77 80.7
M-5237-10 1041936 1229461 74.8237 73.09 73.06 73.06 73.3 74.31 74.08 73.01 74.48 74.47 74.24 74.32 74.66
M-5237-11 1041325 1228063 134.654 134.76 134.76 134.76 134.76 134.76 134.69 134.76 134.67 134.76 134.76 134.68 134.67
M-5237-12 1041342 1228135 128.471 128.37 128.37 128.37 128.37 128.37 128.37 128.37 128.37 128.37 128.37 128.37 128.37
M-5237-13 1041193 1227867 156.796 155.48 155.48 155.48 155.48 155.48 155.48 155.48 155.48 155.48 155.48 155.48 155.48
M-5237-14 1041004 1227679 156.431 156.1 156.1 156.1 156.1 156.1 156.1 156.1 156.1 156.1 156.1 156.1 156.1
M-5237-15 1040950 1227427 140.898 141.42 141.42 141.42 141.42 141.42 141.42 141.42 141.42 141.42 141.42 141.42 141.42
M-5237-16 1041049 1227442 140.241 140.25 140.25 140.25 140.25 140.25 140.25 140.25 140.25 140.25 140.25 140.25 140.25
M-5237-17 1040940 1227195 143.264 143.71 143.71 143.71 143.71 143.71 143.71 143.71 143.71 143.71 143.71 143.71 143.71
M-5237-18 1041039 1227340 143.845 143.8 143.8 143.8 143.8 143.8 143.8 143.8 143.8 143.8 143.8 143.8 143.8
M-5237-19 1041023 1227224 126.714 125.52 125.52 125.52 125.52 125.52 125.52 125.52 125.52 125.52 125.52 125.52 125.52
M-5237-2 1040602 1225288 145.704 144.42 144.42 144.42 144.42 144.42 144.42 144.42 144.42 144.42 144.42 144.42 144.42
M-5237-20 1040955 1227137 143.87 144.79 144.79 144.79 144.79 144.79 144.79 144.79 144.79 144.79 144.79 144.79 144.79
M-5237-21 1040949 1226799 149.932 150.29 150.29 150.29 150.29 150.29 150.29 150.29 150.29 150.29 150.29 150.29 150.29
M-5237-22 1040972 1226800 178.059 178.75 178.75 178.75 178.75 178.75 178.75 178.75 178.75 178.75 178.75 178.75 178.75
M-5237-23 1041171 1223732 165.781 165.39 165.39 165.39 165.39 165.39 165.39 165.39 165.39 165.39 165.39 165.39 165.39
M-5237-24 1041162 1223612 168.806 168.81 168.81 168.81 168.81 168.81 168.81 168.81 168.81 168.81 168.81 168.81 168.81
M-5237-25 1041114 1223519 130.586 130.36 130.36 130.36 130.36 130.36 130.36 130.36 130.36 130.36 130.36 130.36 130.36
M-5237-26 1041128 1223509 128.427 128.8 128.8 128.8 128.8 128.8 128.78 128.8 128.78 128.8 128.8 128.78 128.78
M-5237-27 1041080 1223529 128.956 128.93 128.93 128.93 128.93 128.93 128.93 128.93 128.93 128.93 128.93 128.93 128.93
M-5237-28 1040947 1223228 128.928 128.79 128.79 128.79 128.79 128.79 128.7 128.79 128.67 128.79 128.79 128.68 128.67
M-5237-29 1040969 1223113 105.781 106.75 106.75 106.75 106.75 106.75 106.75 106.75 106.75 106.75 106.75 106.75 106.75
M-5237-3 1040890 1225474 126.19 126.56 126.56 126.56 126.56 126.56 126.56 126.56 126.56 126.56 126.56 126.56 126.56
C Annex: Hydrogeologic Model

Continued on the next page

Observation
Points X Y Measure CZ 40 -PP 100 -PP 200 -PP
SVD Tikh. SVD-a SVD Tikh. SVD-a SVD Tikh. SVD-a SVD Tikh. SVD-a
M-5237-30 1040861 1222823 140.597 140.67 140.67 140.67 140.67 140.67 140.67 140.67 140.67 140.67 140.67 140.67 140.67
M-5237-31 1040722 1222654 128.654 129.44 129.44 129.44 129.44 129.44 129.44 129.44 129.44 129.44 129.44 129.44 129.44
M-5237-32 1040684 1222484 124.859 126.53 126.53 126.53 126.53 126.53 126.53 126.53 126.53 126.53 126.53 126.53 126.53
M-5237-33 1040628 1222426 153.983 156.07 156.07 156.07 156.07 156.07 156.07 156.07 156.07 156.07 156.07 156.07 156.07
M-5237-34 1040585 1222277 71.3105 71.36 71.36 71.36 71.36 71.36 71.36 71.36 71.36 71.36 71.36 71.36 71.36
M-5237-35 1040828 1225478 72.2645 72.35 72.34 72.34 72.31 72.22 72.23 72.31 72.15 72.15 72.32 72.18 72.19
141
D. Annex: Optimization Model Data
Water Demand and Deficit

Demand Deficit
Month ESC1 ESC2 ESC3 ESC4 ESC1 ESC2 ESC3 ESC4
January 6172 9875 12344 18516 -1503 -3878 -5667 -9710
February 6658 10653 13317 19975 -1490 -4276 -6272 -11368
March 5302 8483 10603 15905 -334 -1341 -2261 -4888
April 4913 7860 9826 14738 -202 -822 -1399 -3317
May 5312 8500 10624 15937 -456 -1763 -2853 -6197
June 4179 6686 8357 12536 -474 -1444 -2242 -4510
July 5236 8377 10472 15708 -518 -1897 -3023 -6430
August 4945 7912 9890 14835 -352 -1321 -2207 -4810
September 4883 7813 9766 14649 -298 -1041 -1696 -3671
October 5965 9544 11930 17895 -362 -1327 -2249 -4971
November 5040 8064 10080 15120 -374 -1399 -2313 -4994
December 5636 9018 11272 16908 -589 -2157 -3427 -7267
Total year 64240 102784 128481 192721 -6951 -22665 -35610 -72132

Water Allocation

Surface water allocation Groundwater allocation

Month ESC1 ESC2 ESC3 ESC4 ESC1 ESC2 ESC3 ESC4
January 2156 3048 2924 4384 2514 2950 3753 4422
February 2369 3174 3129 3979 2799 3203 3916 4628
March 3043 4220 3457 4935 1925 2921 4885 6081
April 3205 4693 3279 4801 1506 2346 5148 6620
May 2438 3623 3370 4468 2418 3114 4401 5272
June 2373 3157 2425 3483 1332 2085 3691 4543
July 2376 3451 3280 4260 2342 3030 4169 5017
August 2791 3808 3200 4522 1802 2783 4482 5502
September 3176 4589 3588 4856 1409 2183 4482 6122
October 3599 5096 3866 5652 2004 3122 5814 7272
November 2854 3843 3244 4575 1812 2822 4522 5551
December 2310 3573 3488 4442 2738 3288 4357 5199
Total year 32690 46274 39250 54359 24600 33845 53621 66230
 143

Deficit by Economic Sectors

ESC1
Aquac. Agricult. Const. Domest. O&G Indust. Minner Livestock Serv.
January 196 867 18 73 28 21 15 276 8
February 217 839 10 75 13 19 4 310 4
March 68 156 0 24 0 0 0 87 0
April 43 90 0 16 0 0 0 52 0
May 81 235 0 30 0 1 0 110 0
June 65 302 1 25 0 0 0 81 0
July 89 277 0 32 0 1 0 118 0
August 64 183 0 23 0 0 0 82 0
September 50 164 0 20 0 0 0 64 0
October 65 192 0 24 0 0 0 80 0
November 67 198 0 24 0 0 0 86 0
December 101 316 0 36 0 1 1 134 0

ESC2
Aquac. Agricult. Const. Domest. O&G Indust. Minner Livestock Serv.
January 464 1434 112 153 212 498 163 714 129
February 542 1404 135 175 249 588 195 843 146
March 202 305 43 53 79 197 61 355 48
April 125 182 28 34 53 126 37 205 31
May 268 421 56 66 106 267 79 440 59
June 196 451 43 50 81 203 62 314 44
July 285 481 58 70 109 283 90 454 66
August 193 326 40 49 75 187 59 344 48
September 151 278 32 39 59 146 46 256 34
October 193 335 42 51 77 190 61 330 48
November 207 347 42 52 80 199 62 360 51
December 320 546 68 81 125 328 102 510 76
144 D Annex: Optimization Model Data

ESC3
Aquac. Agricult. Const. Domest. O&G Indust. Minner Livestock Serv.
January 696 3031 74 302 150 243 127 940 104
February 774 3415 71 345 143 223 117 1080 102
March 370 1131 8 160 15 46 12 509 11
April 231 726 3 106 5 17 4 305 1
May 437 1427 17 190 34 82 34 600 32
June 309 1229 11 127 32 30 32 446 27
July 440 1561 16 190 35 79 41 629 32
August 356 1114 7 154 13 43 14 487 19
September 265 941 0 122 1 6 1 353 7
October 371 1153 5 162 9 34 8 495 14
November 367 1181 7 158 13 46 18 503 21
December 495 1759 20 216 47 91 47 710 43

ESC4
Aquac. Agricult. Const. Domest. O&G Indust. Minner Livestock Serv.
January 1086 5415 122 510 265 390 212 1527 184
February 1241 6536 130 593 270 408 233 1746 211
March 693 2562 29 312 58 128 59 994 53
April 540 1651 12 231 21 80 26 732 24
May 925 3129 41 414 101 189 94 1220 84
June 593 2411 47 247 99 140 81 830 62
July 948 3303 50 415 110 193 103 1223 86
August 696 2470 28 313 71 125 67 983 59
September 522 2026 6 231 21 36 32 762 35
October 736 2549 26 316 61 129 64 1032 59
November 716 2580 29 325 76 129 72 1004 62
December 1051 3721 61 456 135 229 119 1398 97
 145

Groundwater Allocation by Economic Sectors

ESC1
Aquac. Agricult. Const. Domest. O&G Indust. Minner Livestock Serv.
January 140 1122 94 119 179 420 135 196 109
February 142 1256 114 129 216 469 161 186 127
March 131 952 65 46 121 259 89 192 70
April 100 771 46 46 87 188 63 153 52
May 167 1146 81 101 152 322 113 244 92
June 80 628 50 26 97 211 66 127 48
July 151 1109 81 96 153 327 114 221 91
August 125 866 62 41 115 246 85 194 68
September 85 732 46 23 87 184 64 135 53
October 135 996 67 45 125 270 91 202 73
November 122 867 63 42 118 252 86 192 70
December 178 1301 94 115 177 381 130 256 105

ESC2
Aquac. Agricult. Const. Domest. O&G Indust. Minner Livestock Serv.
January 187 1093 82 749 155 278 100 223 82
February 160 1384 77 827 151 251 97 168 88
March 339 709 108 497 193 431 123 433 90
April 316 447 76 380 138 329 96 490 74
May 288 856 109 638 205 392 146 363 118
June 213 549 69 415 125 261 83 298 71
July 262 870 105 641 198 367 134 337 116
August 316 664 105 495 186 408 122 396 91
September 261 482 77 370 141 313 93 372 74
October 401 675 107 525 190 457 122 548 96
November 312 689 107 505 189 408 125 394 93
December 271 987 109 728 209 375 143 342 123
146 D Annex: Optimization Model Data

ESC3
Aquac. Agricult. Const. Domest. O&G Indust. Minner Livestock Serv.
January 20 1843 172 232 319 678 218 92 179
February 10 1851 195 221 362 750 255 69 203
March 175 2335 201 184 382 753 281 346 228
April 271 2573 188 149 356 725 258 420 208
May 78 2220 194 188 368 718 262 163 210
June 156 1724 155 165 281 604 197 248 161
July 76 2081 192 189 360 706 250 110 206
August 145 2151 189 180 359 697 259 297 205
September 229 2086 176 179 334 664 245 376 194
October 277 2816 229 200 436 855 319 431 252
November 143 2168 192 184 366 708 261 293 207
December 65 2184 205 201 380 757 267 83 215

ESC4
Aquac. Agricult. Const. Domest. O&G Indust. Minner Livestock Serv.
January 4 1993 236 316 415 938 289 4 226
February 0 1925 270 329 489 1044 326 0 247
March 105 2981 287 280 543 1061 383 132 309
April 184 3235 280 272 534 1025 383 399 310
May 16 2545 277 262 504 1009 352 23 282
June 82 2186 203 240 376 824 268 140 224
July 20 2356 264 262 486 991 336 28 274
August 67 2720 268 261 491 993 346 76 280
September 231 2846 273 256 508 1020 356 351 283
October 138 3623 329 318 613 1212 433 258 348
November 68 2728 273 264 496 1013 349 78 283
December 14 2413 277 280 506 1045 354 18 290
 147

Surface water Allocation by Economic Sectors

ESC1
Aquac. Agricult. Const. Domest. O&G Indust. Minner Livestock Serv.
January 70 1402 12 437 28 46 22 114 24
February 80 1563 10 475 24 38 22 136 22
March 151 1806 41 470 81 159 60 224 51
April 180 1838 52 439 100 200 74 261 61
May 102 1539 26 410 49 96 36 150 30
June 130 1366 32 375 62 119 50 189 48
July 106 1490 24 406 46 85 33 158 29
August 137 1668 37 440 73 144 54 193 45
September 187 1788 51 455 98 201 72 264 59
October 193 2090 52 539 101 200 76 284 64
November 143 1704 38 447 73 146 55 201 46
December 93 1479 19 423 37 63 27 145 25

ESC2
Aquac. Agricult. Const. Domest. O&G Indust. Minner Livestock Serv.
January 0 2899 2 105 8 3 14 0 16
February 0 3066 1 83 5 2 6 0 11
March 19 3647 18 315 51 42 54 18 57
April 77 3690 53 387 108 165 87 51 76
May 5 3393 5 162 12 12 13 4 18
June 32 2673 21 216 48 64 42 22 38
July 5 3251 4 143 11 11 11 5 11
August 13 3357 12 262 39 29 40 11 43
September 103 3533 46 387 96 158 80 114 72
October 36 4234 42 396 95 106 84 28 75
November 13 3394 12 265 38 30 39 12 42
December 3 3422 3 110 8 8 7 4 9
148 D Annex: Optimization Model Data

ESC3
Aquac. Agricult. Const. Domest. O&G Indust. Minner Livestock Serv.
January 0 2899 2 105 8 3 14 0 16
February 0 3066 1 83 5 2 6 0 11
March 19 3647 18 315 51 42 54 18 57
April 77 3690 53 387 108 165 87 51 76
May 5 3393 5 162 12 12 13 4 18
June 32 2673 21 216 48 64 42 22 38
July 5 3251 4 143 11 11 11 5 11
August 13 3357 12 262 39 29 40 11 43
September 103 3533 46 387 96 158 80 114 72
October 36 4234 42 396 95 106 84 28 75
November 13 3394 12 265 38 30 39 12 42
December 3 3422 3 110 8 8 7 4 9

ESC4
Aquac. Agricult. Const. Domest. O&G Indust. Minner Livestock Serv.
January 98 1909 0 723 0 54 0 139 1
February 94 2051 0 791 0 77 0 115 1
March 154 2360 2 736 5 38 4 152 5
April 146 2100 6 747 12 33 12 207 16
May 186 2190 0 704 1 39 1 245 2
June 86 1638 2 560 5 26 4 99 5
July 175 2112 1 689 2 41 2 255 2
August 151 2169 2 673 3 40 3 155 3
September 150 2339 19 694 36 101 27 199 24
October 139 2587 4 854 9 52 7 206 8
November 154 2189 2 685 4 42 3 161 4
December 183 2250 1 731 1 41 1 277 2
 149

Demands by Economic Sectors

ESC1
Aquac. Agricult. Const. Domest. O&G Indust. Minner Livestock Serv.
January 407 3391 123 629 234 487 173 586 142
February 439 3658 133 678 253 525 186 632 153
March 350 2913 106 540 201 418 148 503 122
April 324 2699 98 501 187 388 137 466 113
May 350 2919 106 541 202 419 149 504 122
June 276 2296 83 426 159 330 117 397 96
July 345 2877 105 534 199 413 146 497 120
August 326 2717 99 504 188 390 138 469 114
September 322 2683 98 498 185 385 137 463 112
October 393 3278 119 608 226 471 167 566 137
November 332 2769 101 514 191 398 141 478 116
December 372 3097 113 574 214 445 158 535 129

ESC2
Aquac. Agricult. Const. Domest. O&G Indust. Minner Livestock Serv.
January 651 5426 197 1006 375 779 276 937 227
February 702 5853 213 1086 404 841 298 1011 245
March 559 4661 169 864 322 669 237 805 195
April 518 4319 157 801 298 620 220 746 181
May 560 4670 170 866 323 671 238 807 195
June 441 3674 134 681 254 528 187 635 154
July 552 4603 167 854 318 661 234 795 192
August 522 4347 158 806 300 624 221 751 182
September 515 4293 156 796 297 617 219 741 180
October 629 5244 191 973 362 753 267 906 219
November 532 4431 161 822 306 636 226 765 185
December 595 4955 180 919 342 712 252 856 207
150 D Annex: Optimization Model Data

ESC3
Aquac. Agricult. Const. Domest. O&G Indust. Minner Livestock Serv.
January 814 6782 247 1258 469 974 345 1172 284
February 878 7317 266 1357 506 1051 372 1264 306
March 699 5826 212 1080 403 837 297 1006 244
April 648 5399 196 1001 373 775 275 933 226
May 701 5838 212 1083 403 838 297 1008 244
June 551 4592 167 852 317 660 234 793 192
July 690 5754 209 1067 398 826 293 994 241
August 652 5434 198 1008 375 781 277 939 227
September 644 5366 195 995 371 771 273 927 224
October 787 6555 238 1216 453 942 334 1132 274
November 665 5538 201 1027 383 796 282 957 232
December 743 6193 225 1149 428 890 315 1070 259

ESC4
Aquac. Agricult. Const. Domest. O&G Indust. Minner Livestock Serv.
January 98 1909 0 723 0 54 0 139 1
February 94 2051 0 791 0 77 0 115 1
March 154 2360 2 736 5 38 4 152 5
April 146 2100 6 747 12 33 12 207 16
May 186 2190 0 704 1 39 1 245 2
June 86 1638 2 560 5 26 4 99 5
July 175 2112 1 689 2 41 2 255 2
August 151 2169 2 673 3 40 3 155 3
September 150 2339 19 694 36 101 27 199 24
October 139 2587 4 854 9 52 7 206 8
November 154 2189 2 685 4 42 3 161 4
December 183 2250 1 731 1 41 1 277 2
References

Agencia Nacional de Hidrocarburos - ANH (2012). Escenarios de Oferta y Demanda de

Hidrocarburos en Colombia. Technical report, Agencia Nacional de Hidrocarburos.

Ahmed Suliman, A., Gumindoga, W., Katimon, A., & Darus, I. (2014). Semi-distributed
rainfall-runoff modeling utilizing ASTER DEM in Pinang catchment of Malaysia. Sains
Malaysiana, 43(9).

Al-Jawad, J. Y., Alsaffar, H. M., Bertram, D., & Kalin, R. M. (2019). A comprehensive
optimum integrated water resources management approach for multidisciplinary water
resources management problems. Journal of Environmental Management, 239:211–224.

Alameddine, I., Fayyad, A., Abou Najm, M., & El-Fadel, M. (2018). Sustainability of
basin level development under a changing climate. International Journal of Sustainable
Development and Planning, 13(3):394–405.

Alberti, L., Colombo, L., & Formentin, G. (2018). Null-space Monte Carlo particle tracking
to assess groundwater PCE (Tetrachloroethene) diffuse pollution in north-eastern Milan
functional urban area. Science of The Total Environment, 621:326–339.

Alcolea, A., Carrera, J., & Medina, A. (2006a). Inversion of heterogeneous parabolic-type
equations using the pilot points method. International Journal for Numerical Methods in
Fluids, 51(9-10):963–980.

Alcolea, A., Carrera, J., & Medina, A. (2006b). Pilot points method incorporating
prior information for solving the groundwater flow inverse problem. Advances in Water
Resources, 29(11):1678–1689.

Alcolea, A., Carrera, J., & Medina, A. (2008). Regularized pilot points method for
reproducing the effect of small scale variability: Application to simulations of contaminant
transport. Journal of Hydrology, 355(1-4):76–90.

Allen, R. G., Pereira, L. S., Raes, D., & Smith, M. (1998). Crop evapotranspiration-
guidelines fo computing crop water requirements. FAO Irrigation and Drainage Paper,
56:1–15.

Álvarez Mendiola, E. (2010). Diseño de una política eficiente de precios del agua integrando
 152 References

costes de oportunidad del recurso a escala de cuenca. Aplicación a la Directiva Marco

europea del Agua. PhD thesis, Universidad Politecnica de Valencia.

Amini, M., Johnson, A., Abbaspour, K. C., & Mueller, K. (2009). Modeling large
scale geogenic contamination of groundwater, combination of geochemical expertise
and statistical techniques. In International Congress on Modelling and Simulation:
Interfacing Modelling and Simulation with Mathematical and Computational Sciences,
pages 4100–4106. Geochmica et Cosmochimica Acta.

Andersen, J., Refsgaard, J. C., & Jensen, K. H. (2001). Distributed hydrological modelling
of the Senegal River Basin — model construction and validation. Journal of Hydrology,
247(3-4):200–214.

Anderton, S., Latron, J., & Gallart, F. (2002). Sensitivity analysis and multi-response,
multi-criteria evaluation of a physically based distributed model. Hydrological Processes,
16(2):333–353.

Arboleda-Obando, P. F. (2018). Determinando los efectos del cambio climático y del

cambio en usos del suelo en la Macro Cuenca Magdalena Cauca utilizando el modelo de
suelo-superficie e hidrológico MESH. PhD thesis, Universidad Nacional de Colombia ?
Sede Bogotá.

Arenas-Bautista, M., Pescador-Arévalo, J., Donado, L., Saavedra-Cifuentes, E., &

Arboleda-Obando, P. (2020). Hydrogeological Modeling in Tropical Regions via FeFlow.
Earth Sciences Research Journal, 24(1):42–52.

Arenas-Bautista, M. C., Arboleda Obando, P. F., Duque-Gardeazábal, N., Guadagnini, A.,

Riva, M., & Donado, L. D. (2017). Hydrological Modelling the Middle Magdalena Valley
(Colombia). In American Geophysical Union, New Orleans.

Arenas-Bautista, M. C., Arboleda-Obando, P. F., Duque-Gardeazábal, N., Guadagnini,

A., Riva, M., & Donado, L. D. (2018a). Sensitivity Analysis to uncertain Parameters
of TopModel in Tropical Regions with application to the Middle Magdalena Valley
(Colombia). In European Geosciences Union, Vienna.

Arenas-Bautista, M. C., Arboleda Obando, P. F., Duque-Gardeazábal, N., Saavedra, E., &
Donado, L. D. (2018b). Hydrological Modeling in Tropical Regions via TopModel. Study
Case: Central Sector of the Middle Magdalena Valley - Colombia. In EGU, editor, EGU,
Vienna. Preprints.

Arenas-Bautista, M. C., Pescador-Arévalo, J. P., Donado, L. D., Guadagnini, A., & Riva,
M. (2018c). Three-dimensional geological model applied for groundwater flow simulations
in the Middle Magdalena Valley, Colombia. In AGU Fall Meeting.
References 153

Asociacion Colombiana del Petroleo (2008). Historia del Petróleo En Colombia. Technical
report, ACIPET.

Assaf, H. & Saadeh, M. (2008). Assessing water quality management options in the
Upper Litani Basin, Lebanon, using an integrated GIS-based decision support system.
Environmental Modelling and Software, 23(10-11):1327–1337.

Ayantobo, O. O., Li, Y., Song, S., & Yao, N. (2017). Spatial comparability of drought
characteristics and related return periods in mainland China over 1961–2013. Journal of
Hydrology, 550:549–567.

Badjana, H., Fink, M., Helmschrot, J., Diekkrüger, B., Kralisch, S., Afouda, A., & Wala, K.
(2017). Hydrological system analysis and modelling of the Kara River basin (West Africa)
using a lumped metric conceptual model. Hydrological Sciences Journal, 62(7).

Ballinas-González, H., Alcocer-Yamanaka, V., & Pedrozo-Acuña, A. (2016). Uncertainty

analysis in data-scarce urban catchments. Water (Switzerland), 8(11).

Bao, J., Hou, Z., Ray, J., Huang, M., Swiler, L., & Ren, H. (2018). Soil moisture estimation
using tomographic ground penetrating radar in a MCMC–Bayesian framework. Stochastic
Environmental Research and Risk Assessment, 32(8):2213–2231.

Barthel, R. & Banzhaf, S. (2016). Groundwater and Surface Water Interaction at the
Regional-scale – A Review with Focus on Regional Integrated Models. Water Resources
Management, 30(1):1–32.

Beltrán, A., Lancheros, A., Lopez, C., Chaque, C., Patino, A., Guerra, A., Cabreras, J.,
Quintero, C., & Molano, S. (2014). Plancha Geologica 149- Puerto Serviez.

Beltrán, A. & Quintero, C. (2008). Plancha Geologica 119- Barrancabermeja.

Benoit, N., Marcotte, D., Boucher, A., D’Or, D., Bajc, A., & Rezaee, H. (2018). Directional
hydrostratigraphic units simulation using MCP algorithm. Stochastic Environmental
Research and Risk Assessment, 32(5):1435–1455.

Betancur, T., Mejia, O., & Palacio, C. (2009). Conceptual hydrogeology model to Bajo
Cauca antioqueño: a tropical aquifer system. Revista Facultad de Ingeniería, 48:107–118.

Beven, K. & Freer, J. (2001a). A dynamic TOPMODEL. Hydrological Processes,

15(10):1993–2011.

Beven, K. & Freer, J. (2001b). Equifinality, data assimilation, and uncertainty estimation in
mechanistic modelling of complex environmental systems using the GLUE methodology.
Journal of Hydrology, 249(1-4):11–29.

Blanco-Gutiérrez, I., Varela-Ortega, C., & Purkey, D. R. (2013). Integrated assessment

 154 References

of policy interventions for promoting sustainable irrigation in semi-arid environments: a

hydro-economic modeling approach. Journal of environmental management, 128:144–60.

Booker, J. F., Howitt, R. E., Michelsen, A. M., & Young, R. A. (2012). Economics And The
Modeling Of Water Resources And Policies. Natural Resource Modeling, 25(1):168–218.

Borgonovo, E., Lu, X., Plischke, E., Rakovec, O., & Hill, M. C. (2017). Making the most
out of a hydrological model data set: Sensitivity analyses to open the model black-box.
Water Resources Research, 53(9):7933–7950.

Boskidis, I., Gikas, G. D., Sylaios, G. K., & Tsihrintzis, V. A. (2012). Hydrologic and
Water Quality Modeling of Lower Nestos River Basin. Water Resources Management,
26(10):3023–3051.

Bossa, A., Diekkrüger, B., Giertz, S., Steup, G., Sintondji, L., Agbossou, E., & Hiepe, C.
(2012). Modeling the effects of crop patterns and management scenarios on N and P loads
to surface water and groundwater in a semi-humid catchment (West Africa). Agricultural
Water Management, 115:20–37.

Brouwer, R. & Hofkes, M. (2008). Integrated hydro-economic modelling: Approaches, key

issues and future research directions. Ecological Economics, 66(1):16–22.

Bulot, D., Gumiere, S., Périard, Y., Lafond, J. A., Gallichand, J., ARMALY-ST-GELAIS,
M.-H., & Caron, J. (2016). Relationships between soil hydraulic properties, drainage
efficiency and cranberry yields. Canadian Journal of Soil Science, pages 2016–0020.

Buytaert, W. (2011). topmodel: Implementation of the hydrological model TOPMODEL in

R.

Buytaert, W. & Beven, K. (2011). Models as multiple working hypotheses: hydrological

simulation of tropical alpine wetlands. Hydrological Processes, 25(11):1784–1799.

Cai, X. (2008). Implementation of holistic water resources-economic optimization models for

river basin management - Reflective experiences. Environmental Modelling and Software,
23(1):2–18.

Cánovas, C. R., Macías, F., Olías, M., López, R. P., & Nieto, J. M. (2017). Metal-fluxes
characterization at a catchment scale: Study of mixing processes and end-member analysis
in the Meca River watershed (SW Spain). Journal of Hydrology, 550:590–602.

Carniato, L., Schoups, G., Van De Giesen, N., Seuntjens, P., Bastiaens, L., & Sapion, H.
(2015). Highly parameterized inversion of groundwater reactive transport for a complex
field site. Journal of Contaminant Hydrology, 173:38–58.

Carrera, J., Alcolea, A., Medina, A., Hidalgo, J., & Slooten, L. (2005). Inverse problem in
hydrogeology. Hydrogeology Journal, 13(1):206–222.
References 155

Carrera, J., Hidalgo, J., Slooten, L., & Vázquez-Suñé, E. (2010). Computational and
conceptual issues in the calibration of seawater intrusion models | Problèmes conceptuels
et de calibration des modèles d’intrusion marines. Hydrogeology Journal, 18(1):131–145.

Carrera, J., Mousavi, S., Usunoff, E., Sánchez-Vila, X., & Galarza, G. (1993). A discussion
on validation of hydrogeological models. Reliability Engineering and System Safety,
42(2-3):201–216.

Carrera, J. & Neuman, S. (1986). Estimation of Aquifer Parameters Under Transient

and Steady State Conditions: 2. Uniqueness, Stability, and Solution Algorithms. Water
Resources Research, 22(2):211–227.

Casadevall, S. R. (2016). Improving the management of water multi-functionality through

stakeholder involvement in decision-making processes. Utilities Policy, 43:71–81.

Cediel, F., Barrero, D., & Caceres, C. (1998). Seismic Atlas of Colombia.

Chen, M., Izady, A., Abdalla, O. A., & Amerjeed, M. (2018). A surrogate-based sensitivity
quantification and Bayesian inversion of a regional groundwater flow model. Journal of
Hydrology, 557:826 – 837.

Cherry, J. A., Parker, B. L., Bradbury, K. R., Eaton, T. T., Gotkowitz, M. G., Hart,
D. J., & Borchardt, M. A. (2004). Role of Aquitards in the Protection of Aquifers from
Contamination: A “State of the Science” Report. AWWA Research Foundation Report,
pages 1–144.

Choi, J. M. (2018). Factors influencing public officials’ responses to requests for information
disclosure. Government Information Quarterly, 35(1):30–42.

Christensen, S. & Doherty, J. (2008). Predictive error dependencies when using pilot points
and singular value decomposition in groundwater model calibration. Advances in Water
Resources, 31(4).

Cobourn, K. M., Elbakidze, L., & Ghosh, S. (2017). Chapter 3.1.2 - Conjunctive Water
Management in Hydraulically Connected Regions in the Western United States. In
Ziolkowska, J. R. & Peterson, J. M., editors, Competition for Water Resources, pages
278–297. Elsevier.

Cooper, M. A., Addison, F. T., Alvarez, R., Coral, M., Graham, R., Hayward, A., Howe,
S., Martinez, J., Naar, J., Peñas, R., Pulham, A., & Taborda, A. (1995). Basin
development and tectonic history of the Llanos basin, and Middle Magdalena Valley,
Colombia. Petroleum basins of South America. AAPG. Memoir no. 62, 10(10):659–666.

Crespo, P., Feyen, J., Buytaert, W., Célleri, R., Frede, H.-G., Ramírez, M., & Breuer, L.
(2012). Development of a conceptual model of the hydrologic response of tropical Andean
 156 References

micro-catchments in Southern Ecuador. Hydrology and Earth System Sciences Discussions,

9(2):2475–2510.

Custodio, E., Andreu-Rodes, J. M., Aragón, R., Estrela, T., Ferrer, J., García-Aróstegui,
J. L., Manzano, M., Rodríguez-Hernández, L., Sahuquillo, A., & Del Villar, A. (2016).
Groundwater intensive use and mining in south-eastern peninsular Spain: Hydrogeological,
economic and social aspects. The Science of the total environment, 559:302–316.

da Silva, M. G., de Aguiar Netto, A. d. O., de Jesus Neves, R. J., do Vasco, A. N., Almeida,
C., & Faccioli, G. G. (2015). Sensitivity Analysis and Calibration of Hydrological Modeling
of the Watershed Northeast Brazil. Journal of Environmental Protection, 06(08):837–850.

Dai, H., Ye, M., Walker, A., & Chen, X. (2017). A new process sensitivity index to identify
important system processes under process model and parametric uncertainty. Water
Resources Research, 53(4).

Dakhlaoui, H., Ruelland, D., Tramblay, Y., & Bargaoui, Z. (2017). Evaluating the robustness
of conceptual rainfall-runoff models under climate variability in northern Tunisia. Journal
of Hydrology, 550:201–217.

DANE (2016). Metodología General Tercer Censo Nacional Agropecuario. Technical report,
Departamento Adminsitrativo Nacional de Estadistica, Bogotá D.C.

DANE (2017). Hoja metodológica de indicadores Productividad hídrica. Technical report,

Departamento Nacional de Estadísticas, Bogotá D.C.

DANE (2019). Boletín Técnico de Índice de Precios. Technical report, Departamento

Administrativo Nacional de Estadistica, Bogotá D.C.

Davidsen, C., Liu, S., Mo, X., Holm, P. E., Trapp, S., Rosbjerg, D., & Bauer-Gottwein, P.
(2015). Hydroeconomic optimization of reservoir management under downstream water
quality constraints. Journal of Hydrology, 529:1679–1689.

Davis, M. D. (2007). Integrated water resource management and water sharing. Journal of
Water Resources Planning And Management-Asce, 133(5):427–445.

De O. Torres, M., Howitt, R., & Rodrigues, L. N. (2016). Modeling the economic benefits
and distributional impacts of supplemental irrigation. Water Resources and Economics,
14:1–12.

de Sousa Fragoso, R. M. & de Almeida Noéme, C. J. (2018). Economic effects of

climate change on the Mediterranean’s irrigated agriculture. Sustainability Accounting,
Management and Policy Journal, 9(2):118–138.

Dell’Oca, A., Riva, M., & Guadagnini, A. (2017). Moment-based Metrics for Global
References 157

Sensitivity Analysis of Hydrological Systems. Hydrology and Earth System Sciences

Discussions, pages 1–41.

Dessu, S. B., Melesse, A. M., Bhat, M. G., Price, R. M., Seid, A. H., Debebe, S. A., &
McClain, M. E. (2019). Development and application of a priority rated optimization
model (PROM) for multi-sector water resource management systems. Environmental
Modelling & Software, 113:84–97.

Dewandel, B., Jeanpert, J., Ladouche, B., Join, J. L., & Maréchal, J. C. (2017). Inferring
the heterogeneity, transmissivity and hydraulic conductivity of crystalline aquifers from a
detailed water-table map. Journal of Hydrology, 550:118–129.

Dewandel, B., Maréchal, J., Bour, O., Ladouche, B., Ahmed, S., Chandra, S., & Pauwels,
H. (2012). Upscaling and regionalizing hydraulic conductivity and effective porosity
at watershed scale in deeply weathered crystalline aquifers. Journal of Hydrology,
416-417:83–97.

Doherty, J. (2003). Ground water model calibration using pilot points and regularization.
Ground Water, 41(2):170–177.

Doherty, J. (2004). PEST Model-Independent Parameter Estimation. Watermark Numerical

Computing, Idaho, 5th edition.

Donado, L. D., Arenas-Bautista, M. C., Pescador-Arévalo, J. P., Guadagnini, A., & Riva,
M. (2018). Hydrogeological Characterization in Tropical Regions with lack of information
subject to competing uses of groundwater. In AGU Fall Meeting, Washington.

Du, M., Zavattero, E., Ma, Q., Gourbesville, P., & Delestre, O. (2018). Groundwater
Modeling for a Decision Support System: The Lower Var Valley, Southeastern France.
In Advances in Hydroinformatics, pages 273–283. Springer Nature.

Dunne, T. (1983). Relation of field studies and modeling in the prediction of storm runoff.
Journal of Hydrology, 65(1):25–48.

Ehtiat, M., Mousavi, S., & Ghaheri, A. (2015). Ranking of conceptualized groundwater
models based on model information criteria. Journal of Water Supply: Research and
Technology - AQUA, 64(6):670–687.

Escriva-Bou, A., Pulido-Velazquez, M., & Pulido-Velazquez, D. (2017). Economic value

of climate change adaptation strategies for water management in Spain’s Jucar Basin.
Journal of Water Resources Planning and Management, 143(5).

Fabre, J., Ruelland, D., Dezetter, A., & Grouillet, B. (2016). Sustainability of water uses in
managed hydrosystems: Human- and climate-induced changes for the mid-21st century.
Hydrology and Earth System Sciences, 20(8):3129–3147.
 158 References

Fajraoui, N., Ramasomanana, F., Younes, A., Mara, T. A., Ackerer, P., & Guadagnini, A.
(2011). Use of global sensitivity analysis and polynomial chaos expansion for interpretation
of nonreactive transport experiments in laboratory-scale porous media. Water Resources
Research, 47(2):1–14.

Ferreira, D. M., Fernandes, C. V. S., Kaviski, E., & Fontane, D. (2019). Water quality
modelling under unsteady state analysis: Strategies for planning and management. Journal
of Environmental Management, 239:150–158.

Finsterle, S. & Zhang, Y. (2011). Error handling strategies in multiphase inverse modeling.
Computers & Geosciences, 37(6):724 – 730.

Freeze, R. & Cherry, J. (1979). Groundwater. Prentice-Hall.

Friedel, M. & Iwashita, F. (2013). Hybrid modeling of spatial continuity for application to
numerical inverse problems. Environmental Modelling and Software, 43:60–79.

Fu, Z. H., Zhao, H. J., Wang, H., Lu, W. T., Wang, J., & Guo, H. C. (2017). Integrated
planning for regional development planning and water resources management under
uncertainty: A case study of Xining, China. Journal of Hydrology, 554:623–634.

Gaganis, P. & Smith, L. (2006). Evaluation of the uncertainty of groundwater model

predictions associated with conceptual errors: A per-datum approach to model calibration.
Advances in Water Resources, 29(4):503–514.

Gallego, J., Jaramillo, H., & Patiño, A. (2015). Servicios Intensivos en Conocimiento en la
Industria del Petróleo en Colombia. Technical report, Banco Interamericano de Desarrollo.

García-González, M. L., Carvajal-Escobar, Y., & Jiménez-Escobar, H. (2007). Integrated

water resource management as a strategy for adaptation to climate change. Ingeniería y
Competitividad, 29(1):19–29.

Garrido-Rodríguez, J. C., López-Hernández, A. M., & Zafra-Gómez, J. L. (2019). The

impact of explanatory factors on a bidimensional model of transparency in Spanish local
government. Government Information Quarterly, 36(1):154–165.

Garrote, L. (2017). Managing Water Resources to Adapt to Climate Change: Facing

Uncertainty and Scarcity in a Changing Context. Water Resources Management,
31(10):2951–2963.

Gavidia, A., Porras, J., Perez, O., Pacheco, S., Mesa, L., Talero, C., Fuquen, J., Farfan, E.,
& Fonseca, H. (2008). Plancha Geologica 133- Puerto Berrio.

George, B., Malano, H., Davidson, B., Hellegers, P., Bharati, L., & Massuel, S. (2011). An
integrated hydro-economic modelling framework to evaluate water allocation strategies I:
Model development. Agricultural Water Management, 98(5):733–746.
References 159

Gil Morales, E. G. & Tobón Marín, C. (2016). Hydrological modelling with TOPMODEL
of Chingaza páramo, Colombia. Revista Facultad Nacional de Agronomía, 69(2).

Gleick, P. H. & Palaniappan, M. (2010). Peak water limits to freshwater withdrawal and
use. Proceedings of the National Academy of Sciences, 107(25):11155–11162.

Gogu, R., Carabin, G., Hallet, V., Peters, V., & Dassargues, A. (2001). GIS-based
hydrogeological databases and groundwater modelling. Hydrogeology Journal,
9(6):555–569.

Golmohammadi, A., Khaninezhad, M.-R., & Jafarpour, B. (2015). Group-sparsity

regularization for ill-posed subsurface flow inverse problems. Water Resources Research,
51(10):8607–8626.

Gómez, E., Jordan, T. E., Allmendinger, R. W., Hegarty, K., & Kelley, S. (2005). Syntectonic
Cenozoic sedimentation in the northern middle Magdalena Valley Basin of Colombia and
implications for exhumation of the Northern Andes. Bulletin of the Geological Society of
America, 117(5-6):547–569.

Gómez, E., Jordan, T. E., Allmendinger, R. W., Hegarty, K., Kelley, S., & Heizler, M. (2003).
Controls on architecture of the Late Cretaceous to Cenozoic southern Middle Magdalena
Valley Basin, Colombia. Bulletin of the Geological Society of America, 115(2):131–147.

Gonzalez, M., Sladarriaga, G., & Jaramillo, O. (2010). Estimación de la demanda del agua:
Conceptaulización y dimensionamiento de la demanda hídrica sectorial. Estudio Nacional
del Agua, pages 169–228.

González-Zeas, D., Erazo, B., Lloret, P., De Bièvre, B., Steinschneider, S., & Dangles, O.
(2019). Linking global climate change to local water availability: Limitations and prospects
for a tropical mountain watershed. Science of the Total Environment, 650:2577–2586.

Groves, D. G., Yates, D., & Tebaldi, C. (2008). Developing and applying uncertain global
climate change projections for regional water management planning. Water Resources
Research, 44(12):n/a–n/a.

Gunawardena, A., White, B., Hailu, A., Wijeratne, E., & Pandit, R. (2018). Policy choice and
riverine water quality in developing countries: An integrated hydro-economic modelling
approach. Journal of Environmental Management, 227:44–54.

Gupta, H. V. & Razavi, S. (2018). Revisiting the Basis of Sensitivity Analysis for Dynamical
Earth System Models. Water Resources Research, 54(11):8692–8717.

H. Hargreaves, G. & A. Samani, Z. (1985). Reference Crop Evapotranspiration from

Temperature. Applied Engineering in Agriculture, 1(2):96–99.

Han, D. & Cao, G. (2018). Phase difference between groundwater storage changes and
 160 References

groundwater level fluctuations due to compaction of an aquifer-aquitard system. Journal

of Hydrology, 566:89–98.

Hanemann, W. M. (2006). The economic conception of water. In Water Crisis: Myth or

Reality, number 1, chapter 4, pages 60–91. University of California, Berkeley.

Hargreaves, G. & Samani, Z. (1982). Estimating potential evapotranspiration. Journal of

the Irrigation & Drainage Division, 108(IR3):225–230.

Hargreaves, G. H. & Allen, R. G. (2003). History and Evaluation of Hargreaves

Evapotranspiration Equation. Journal of Irrigation and Drainage Engineering, 129(1).

Harmanny, K. S. & Malek, Z. (2019). Adaptations in irrigated agriculture in the

Mediterranean region: an overview and spatial analysis of implemented strategies.
Regional Environmental Change, 19(5):1401–1416.

Harou, J. J., Pulido-Velazquez, M., Rosenberg, D. E., Medellín-Azuara, J., Lund, J. R., &
Howitt, R. E. (2009). Hydro-economic models: Concepts, design, applications, and future
prospects. Journal of Hydrology, 375(3-4):627–643.

Hashimoto, T., Loucks, D. P., & Stedinger, J. R. (1982). Reliability , Resiliency , Robustness
, and Vulnerability Criteria for Water Resource Systems. Water Resources Research,
18(1):14–20.

Hassane, M. & Ackerer, P. (2017). Groundwater flow parameter estimation using refinement
and coarsening indicators for adaptive downscaling parameterization. Advances in Water
Resources, 100:139–152.

He, L., Huang, G., Lu, H., Wang, S., & Xu, Y. (2012). Quasi-Monte Carlo based global
uncertainty and sensitivity analysis in modeling free product migration and recovery from
petroleum-contaminated aquifers. Journal of hazardous materials, 219-220:133–40.

Hernandez, A., Neuman, S., Guadagnini, A., & Carrera, J. (2003). Conditioning mean study
state flow on hydraulic head and conductivity through geostatistical inversion. Stochastic
Environmental Research and Risk Assessment, 17(5):329–338.

Hernández-Bedolla, J., Solera, A., Paredes-Arquiola, J., Pedro-Monzonís, M., Andreu, J.,
& Sánchez-Quispe, S. T. (2017). The Assessment of Sustainability Indexes and Climate
Change Impacts on Integrated Water Resource Management. Water (Switzerland), 9(3).

Hollanda, M., Cecílio, R., Campanharo, W., Zanetti, S., de Andrade, L., & Garcia, G.
(2015). Evaluation of TOPMODEL for prediction of the runoff of a watershed under
different land uses | Avaliação do TOPMODEL na estimativa do escoamento superficial
em microbacia hidrográfica em diferentes usos. Revista Brasileira de Engenharia Agricola
e Ambiental, 19(5).
References 161

Hou, T., Zhu, Y., Lü, H., Sudicky, E., Yu, Z., & Ouyang, F. (2015). Parameter sensitivity
analysis and optimization of Noah land surface model with field measurements from Huaihe
River Basin, China. Stochastic Environmental Research and Risk Assessment, 29(5).

Hu, Z. & Chan, C. W. (2015). In-situ bioremediation for petroleum contamination: A

fuzzy rule-based model predictive control system. Engineering Applications of Artificial
Intelligence, 38:70–78.

Hughes, D. (2016). Hydrological modelling, process understanding and uncertainty in a

southern African context: lessons from the northern hemisphere. Hydrological Processes,
30(14).

Hutcheson, R. S. & McAdams, D. A. (2010). A Hybrid Sensitivity Analysis for Use in Early
Design. Journal of Mechanical Design, 132(11):111007.

Ibarra-Zavaleta, S., Landgrave, R., Romero-López, R., Poulin, A., & Arango-Miranda,
R. (2017). Distributed hydrological modeling: Determination of theoretical hydraulic
potential &amp; streamflow simulation of extreme hydrometeorological events. Water
(Switzerland), 9(8).

Ideam (2014). Estudio Nacional del Agua 2014. Technical report, Instituto de Hidrología,
Meteorología y Estudios Ambientales.

IDEAM (2019). Estudio Nacional del Agua 2018. Technical report, Instituto de Hidrología,
Meteorología y Estudios Ambientales – IDEAM, Bogotá D.C.

Iglesias, A. & Garrote, L. (2015). Adaptation strategies for agricultural water management
under climate change in Europe. Agricultural Water Management, 155:113–124.

Iglesias, A., Garrote, L., Diz, A., Schlickenrieder, J., & Martin-Carrasco, F. (2011).
Re-thinking water policy priorities in the Mediterranean region in view of climate change.
Environmental Science & Policy, 14(7):744–757.

Iglesias, A., Santillán, D., & Garrote, L. (2018). On the Barriers to Adaption to Less Water
under Climate Change: Policy Choices in Mediterranean Countries. Water Resources
Management, 32(15):4819–4832.

INGENIERIA GEOTEC, S. (2016). ESTUDIO DE IMPACTO AMBIENTAL PARA LAS

ACTIVIDADES EXPLORATORIAS A DESARROLLARSE EN EL BLOQUE VMM-9.
Technical report, PAREX, Bogota.

Ingrain (2012). Cuenca del Valle Medio del Magdalena - Integración Geológica de la
Digitalización y Análisis de Núcleos. Technical report, Agencia Nacional de Hidrocarburos.

Instituto Nacional de Salud (2018). Boletin de vigilancia de calidad del agua en colombia.
Technical report, Instituto Nacional de Salud, Bogotá D.C.
 162 References

Irsa, J. & Zhang, Y. (2012). A direct method of parameter estimation for steady state flow
in heterogeneous aquifers with unknown boundary conditions. Water Resources Research,
48(9).

Islam, M., Firoz, A., Foglia, L., Marandi, A., Khan, A., Schüth, C., & Ribbe, L. (2017).
A regional groundwater-flow model for sustainable groundwater-resource management in
the south Asian megacity of Dhaka, Bangladesh | Modèle régional d’écoulement des eaux
souterraines pour une gestion durable des ressources en eaux souterraines dans la mé.
Hydrogeology Journal, 25(3):617–637.

Jafarpour, B. (2013). Sparsity-promoting solution of subsurface flow model calibration

inverse problems. Advances in Hydrogeology, 9781461464:73–94.

Janetti, E., Riva, M., Straface, S., & Guadagnini, A. (2010). Stochastic characterization of
the Montalto Uffugo research site (Italy) by geostatistical inversion of moment equations
of groundwater flow. Journal of Hydrology, 381(1-2):42–51.

Jardani, A., Dupont, J., Revil, A., Massei, N., Fournier, M., & Laignel, B. (2012).
Geostatistical inverse modeling of the transmissivity field of a heterogeneous alluvial
aquifer under tidal influence. Journal of Hydrology, 472-473:287–300.

Jenson, S. K. (1991). Applications of hydrologic information automatically extracted from

digital elevation models. Hydrological Processes, 5(1):31–44.

Jeong, H. & Adamowski, J. (2016). A system dynamics based socio-hydrological model for
agricultural wastewater reuse at the watershed scale. Agricultural Water Management,
171:89–107.

Jeziorska, J. & Niedzielski, T. (2015). Applicability of TOPMODEL in the mountainous

catchments in the upper Nysa Kłodzka River basin (SW Poland). Acta Geophysica,
66(2):203–222.

Jia, Z., Cai, Y., Chen, Y., & Zeng, W. (2018). Regionalization of water environmental
carrying capacity for supporting the sustainable water resources management and
development in China. Resources, Conservation and Recycling, 134:282–293.

Jiao, J. & Zhang, Y. (2016). Direct Method of Hydraulic Conductivity Structure

Identification for Subsurface Transport Modeling. Journal of Hydrologic Engineering,
21(10).

Jiménez, S., Mariethoz, G., Brauchler, R., & Bayer, P. (2016). Smart pilot points using
reversible-jump Markov-chain Monte Carlo. Water Resources Research, 52(5):3966–3983.

Jung, Y., Ranjithan, R., & Mahinthakumar, G. (2011). Subsurface characterization using a
D-optimality based pilot point method. Journal of Hydroinformatics, 13(4):775–793.
References 163

Karay, G. & Hajnal, G. (2015). Modelling of Groundwater Flow in Fractured Rocks. Procedia
Environmental Sciences, 25:142 – 149.

Karlsson, I. B., Sonnenborg, T. O., Refsgaard, J. C., Trolle, D., Børgesen, C. D., Olesen,
J. E., Jeppesen, E., & Jensen, K. H. (2016). Combined effects of climate models,
hydrological model structures and land use scenarios on hydrological impacts of climate
change. Journal of Hydrology, 535:301–317.

Kashyap, D. & Vakkalagadda, R. (2009). New model of variogram of groundwater hydraulic

heads. Journal of Hydrologic Engineering, 14(8):872–875.

Kauffeldt, A., Wetterhall, F., Pappenberger, F., Salamon, P., & Thielen, J. (2016).
Technical review of large-scale hydrological models for implementation in operational flood
forecasting schemes on continental level. Environmental Modelling & Software, 75:68–76.

Kellner, J., Multsch, S., Kraft, P., Houska, T., Mueller, C., & Breuer, L. (2015). Uncertainty
Analysis of a Coupled Hydrological-plant Growth Model for Grassland under Elevated
CO2. Procedia Environmental Sciences, 29:79–80.

Khaninezhad, M. & Jafarpour, B. (2018). Field-scale history matching with sparse geologic
dictionaries. Journal of Petroleum Science and Engineering, 170:967–991.

Khaninezhad, M., Jafarpour, B., & Li, L. (2012). Sparse geologic dictionaries for subsurface
flow model calibration: Part II. Robustness to uncertainty. Advances in Water Resources,
39:122–136.

Khaninezhad, M.-R., Golmohammadi, A., & Jafarpour, B. (2018). Discrete Regularization

for Calibration of Geologic Facies Against Dynamic Flow Data. Water Resources Research,
54(4):2523–2543.

Khorashadi Zadeh, F., Nossent, J., Sarrazin, F., Pianosi, F., van Griensven, A., Wagener,
T., & Bauwens, W. (2017). Comparison of variance-based and moment-independent
global sensitivity analysis approaches by application to the SWAT model. Environmental
Modelling and Software, 91.

Ki-moon, B., Bokova, I., Jarraud, M., & Miletto, M. (2014). Water and Energy. Technical
report, UNESCO.

Kirshen, P., Aytur, S., Hecht, J., Walker, A., Burdick, D., Jones, S., Fennessey, N., Bourdeau,
R., & Mather, L. (2018). Integrated urban water management applied to adaptation to
climate change. Urban Climate, 24:247–263.

Klaas, D. K. S. Y., Imteaz, M. A., Sudiayem, I., Klaas, E. M. E., & Klaas, E. C. M. (2017).
Novel approaches in sub-surface parameterisation to calibrate groundwater models. IOP
Conference Series: Earth and Environmental Science, 82(1):012014.
 164 References

Kpegli, K. A. R., Alassane, A., van der Zee, S. E., Boukari, M., & Mama, D. (2018).
Development of a conceptual groundwater flow model using a combined hydrogeological,
hydrochemical and isotopic approach: A case study from southern Benin. Journal of
Hydrology: Regional Studies, 18:50 – 67.

Labadie, J. (1975). A Surrogate-Parameter Approach to Modeling Groundwater Basins.

JAWRA Journal of the American Water Resources Association, 11(1):97–114.

Lamb, R., Beven, K., & Myrabo, S. (1998). Use of spatially distributed water table
observations to constrain uncertainty in a rainfall–runoff model. Advances in Water
Resources, 22(4):305–317.

Le Ravalec-Dupin, M. (2010). Pilot block method methodology to calibrate stochastic

permeability fields to dynamic data. Mathematical Geosciences, 42(2):165–185.

Le Ravalec-Dupin, M. & Roggero, F. (2012). Parameterization in history matching: State

of the art and perspectives. In Integrated Reservoir Modelling: Are We Doing it Right?

Li, H. & Zhang, Y. (2017). Regionalising rainfall-runoff modelling for predicting daily runoff:
Comparing gridded spatial proximity and gridded integrated similarity approaches against
their lumped counterparts. Journal of Hydrology, 550:279–293.

Li, T., Yang, S., & Tan, M. (2019). Simulation and optimization of water supply and demand
balance in Shenzhen: A system dynamics approach. Journal of Cleaner Production,
207:882–893.

Linde, N., Ginsbourger, D., Irving, J., Nobile, F., & Doucet, A. (2017). On uncertainty
quantification in hydrogeology and hydrogeophysics. Advances in Water Resources,
110:166 – 181.

Linde, N., Lochbühler, T., Dogan, M., Van Dam, R. L., & Dam, R. L. V. (2015).
Tomogram-based comparison of geostatistical models: Application to the Macrodispersion
Experiment (MADE) site. Journal of Hydrology, 531:543–556.

Lionboui, H., Benabdelouahab, T., Elame, F., Hasib, A., & Boulli, A. (2018). Estimating
the economic impact of climate change on agricultural water management indicators.
Pertanika Journal of Science and Technology, 26(2):749–762.

Liu, J., Zhu, A.-X., Qin, C.-Z., Wu, H., & Jiang, J. (2016). A two-level parallelization method
for distributed hydrological models. Environmental Modelling & Software, 80:175–184.

Llopis-Albert, C., Merigó, J., & Xu, Y. (2016). A coupled stochastic inverse/sharp interface
seawater intrusion approach for coastal aquifers under groundwater parameter uncertainty.
Journal of Hydrology, 540:774–783.

Loosvelt, L., De Baets, B., Pauwels, V. R., & Verhoest, N. E. (2014). Assessing hydrologic
References 165

prediction uncertainty resulting from soft land cover classification. Journal of Hydrology,
517:411–424.
López, S. T., de los Angeles Barrionuevo, M., & Rodríguez-Labajos, B. (2019). Water
accounts in decision-making processes of urban water management: Benefits, limitations
and implications in a real implementation. Sustainable Cities and Society, 50:101676.
Lopez-Nicolas, A., Pulido-Velazquez, M., Rougé, C., Harou, J. J., & Escriva-Bou, A. (2018).
Design and assessment of an efficient and equitable dynamic urban water tariff. Application
to the city of Valencia, Spain. Environmental Modelling & Software, 101:137–145.
Loucks, D. P., Beek, E. V., Stedinger, J. R., Dijkman, J. P., & Monique T. Villars (2005).
Water Resources Systems Planning and Management: An Introduction to Methods, Models
and Applications. UNESCO, Paris.
Lund, J. R. & Israel, M. . (1995). Water Transfers in Water Resource Systems. Journal of
Water Resources Planning and Management, 121(2).
Ma, W. & Jafarpour, B. (2017). Conditioning multiple-point geostatistical facies simulation
on nonlinear flow data using pilot points method. In SPE Western Regional Meeting
Proceedings, volume 2017-April, pages 288–305.
Ma, W. & Jafarpour, B. (2018). Pilot points method for conditioning multiple-point
statistical facies simulation on flow data. Advances in Water Resources, 115:219–233.
Mahmoud, F. & Saleem, M. (1993). Regularization method for solving radon integral
equations. Computers and Mathematics with Applications, 25(4):3–13.
Malagón-Navarro, J. P. (2017). Análisis hidrogeoquímico-multivariado del agua subterránea
del sistema acuífero del Valle Medio del Magdalena Colombia. PhD thesis, Universidad
Nacional de Colombia - Sede Bogotá.
Maliva, R. G. (2014). Economics of managed aquifer recharge. Water (Switzerland),
6(5):1257–1279.
Marchant, B. P. & Bloomfield, J. P. (2018). Spatio-temporal modelling of the status of
groundwater droughts. Journal of Hydrology, 564:397 – 413.
Martin-Carrasco, F. & Garrote, L. (2006). Drought-Induced Water Scarcity in Water
Resources Systems. In NATO Science Series, editor, Extreme Hydrological Events: New
Concepts for Security, pages 301–311. Springer Netherlands, Dordrecht, 78 edition.
Martinsen, G., Liu, S., Mo, X., & Bauer-Gottwein, P. (2019). Joint optimization of water
allocation and water quality management in Haihe River basin. Science of The Total
Environment, 654:72–84.
MAVDT (2013). Metodología para la estimación y evaluación del caudal ambiental en
 166 References

proyectos que requieren Licencia Ambiental. Technical report, Ministerio de Medio

Ambiente Vivienda y Desarrollo TerritoriL, Bogotá D.C.

Mazzoleni, M., Alfonso, L., Chacon-Hurtado, J., & Solomatine, D. (2015). Assimilating
uncertain, dynamic and intermittent streamflow observations in hydrological models.
Advances in Water Resources, 83:323–339.

Medina, A., Galarza, G., Carrera, J., Jódar, J., & Alcolea, A. (2001). The inverse problem in
hydrogeology: Applications | El problema inverso en hidrología subterránea. Aplicaciones.
Boletin Geologico y Minero, 112(SPECIAL ED):93–106.

Meeks, J., Moeck, C., Brunner, P., & Hunkeler, D. (2017). Infiltration under snow cover:
Modeling approaches and predictive uncertainty. Journal of Hydrology, 546:16–27.

Meerveld, I. T. v. & Weiler, M. (2008). Hillslope dynamics modeled with increasing

complexity. Journal of Hydrology, 361(1-2):24–40.

Merritt, W., Croke, B., & Jakeman, A. (2005). Sensitivity testing of a model for exploring
water resources utilisation and management options. Environmental Modelling & Software,
20(8):1013–1030.

Metcalfe, P., Beven, K., & Freer, J. (2015). Dynamic TOPMODEL: A new implementation
in R and its sensitivity to time and space steps. Environmental Modelling & Software,
72:155–172.

Metcalfe, P., Beven, K., Hankin, B., & Lamb, R. (2017). A modelling framework
for evaluation of the hydrological impacts of nature-based approaches to flood risk
management, with application to in-channel interventions across a 29 km2 scale catchment
in the United Kingdom. Hydrological Processes, 31(9):1734–1748.

Meyer, R., Engesgaard, P., Høyer, A.-S., Jørgensen, F., Vignoli, G., & Sonnenborg, T. O.
(2018). Regional flow in a complex coastal aquifer system: Combining voxel geological
modelling with regularized calibration. Journal of Hydrology, 562:544 – 563.

Ministerio de Ambiente, V. y. D. T. (2007). Resolucion N. 2115 del 2007. Technical report,

MAVDT, Bogotá D.C.

Ministerio de Ambiente y Desarrollo Sostenible (2017). Decreto 1155 de 2017.

Ministerio de Vivienda Ciudad y Territorio (2018). Resolución CRA 825 de 2017. Technical
report, Ministerio de Vivienda, Ciudad y Territorio, Bogotá D.C.

Mishra, V. & Lilhare, R. (2016). Hydrologic sensitivity of Indian sub-continental river basins
to climate change. Global and Planetary Change, 139:78–96.

Mockler, E. M., O’Loughlin, F. E., & Bruen, M. (2016). Understanding hydrological

References 167

flow paths in conceptual catchment models using uncertainty and sensitivity analysis.
Computers & Geosciences, 90:66–77.
Mohtar, W. H. M. W., Maulud, K. N. A., Muhammad, N. S., Sharil, S., & Yaseen, Z. M.
(2019). Spatial and temporal risk quotient based river assessment for water resources
management. Environmental Pollution, 248:133–144.
Morales, L. (1958). General Geology and Oil Occurrences of Middle Magdalena Valley,
Colombia: South America. Asociación Colombiana de Geólogos y Geofisicos del Petróleo
(ACGGP), page 55.
Morales-Marín, L. A., Wheater, H. S., & Lindenschmidt, K. E. (2017). Assessment of nutrient
loadings of a large multipurpose prairie reservoir. Journal of Hydrology, 550:166–185.
Moreno, N., Silva, A., Mora, A., Tesón, E., Quintero, I., Rojas, L. E., Lopez, C., Blanco,
V., Castellanos, J., Sanchez, J., Osorio, L., Namson, J., Stockli, D., & Casallas, W.
(2013). Interaction between thin- and thick-skinned tectonics in the foothill areas of an
inverted graben. The Middle Magdalena Foothill belt. Geological Society, London, Special
Publications, 377(1):221–255.
Mukherjee, S., Mukherjee, S., Garg, R., Bhardwaj, A., & Raju, P. (2013). Evaluation of
topographic index in relation to terrain roughness and DEM grid spacing. Journal of
Earth System Science, 122(3).
Mukundan, R., Acharya, N., Gelda, R. K., Frei, A., & Owens, E. M. (2019). Modeling
streamflow sensitivity to climate change in New York City water supply streams using a
stochastic weather generator. Journal of Hydrology: Regional Studies, 21:147–158.
Navarro-Chaparro, K., Rivera, P., & Sánchez, R. (2015). Water management analysis of
the city of Tijuana , Baja California : Critical factors and challenges Análisis del manejo
de agua en la ciudad de Tijuana , Baja California : Factores críticos y retos. Estudios
Fronterizos Nueva Época, 17(33):1–20.
Neverre, N. & Dumas, P. (2015). Projecting and valuing domestic water use at regional scale:
A generic method applied to the Mediterranean at the 2060 horizon. Water Resources and
Economics, 11:33–46.
Ortiz, R., Luis, J., Ruiz, S., & Ricardo, J. (2009). Modelo Geológico De Los Campos
Maduros De Lisama, Tesoro, Nutria Y Peroles, Valle Medio Del Magdalena. Colombia.
. . . Exploracion Petrolera en . . . .
Padrón, R. S., Wilcox, B. P., Crespo, P., & Célleri, R. (2015). Rainfall in the Andean
Páramo: New Insights from High-Resolution Monitoring in Southern Ecuador. Journal of
Hydrometeorology, 16(3):985–996.
Panzeri, M., Guadagnini, A., & Riva, M. (2012). Optimization of pilot points location for
 168 References

geostatistical inversion of groundwater flow. In IAHS-AISH Publication, volume 355, pages

34–40.
Parra, E. A. & Carvajal, I. A. (2012). Modelamiento y manejo de las interacciones entre
la hidrología, la ecología y la economía en una cuenca hidrográfica para la estimación de
caudales ambientales. PhD thesis, Universidad Nacional de Colombia.
Payne, J. T., Wood, A. W., Hamlet, A. F., Palmer, R. N., & Lettenmaier, D. P. (2004).
Mitigating the Effects of Climate Change on the Water Resources of the Columbia River
Basin. Climatic Change, 62(1-3):233–256.
Pedro-Monzonís, M., Ferrer, J., Solera, A., Estrela, T., & Paredes-Arquiola, J. (2015). Key
issues for determining the exploitable water resources in a Mediterranean river basin.
Science of The Total Environment, 503-504:319–328.
Pedro-Monzonís, M., Solera, A., Ferrer, J., Andreu, J., & Estrela, T. (2016). Water
accounting for stressed river basins based on water resources management models. Science
of The Total Environment, 565:181–190.
Peña-Haro, S., Pulido-Velazquez, M., & Llopis-Albert, C. (2011). Stochastic hydro-economic
modeling for optimal management of agricultural groundwater nitrate pollution
under hydraulic conductivity uncertainty. Environmental Modelling and Software,
26(8):999–1008.
Peña-Haro, S., Pulido-Velazquez, M., & Sahuquillo, A. (2009). A hydro-economic modeling
framework for optimal management of groundwater nitrate pollution from agriculture.
Journal of Hydrology, 373(1-2):193–203.
Pereira, H., Figueira, J. R., & Marques, R. C. (2019). Multiobjective Irrigation Model:
Alqueva River Basin Application. Journal of Irrigation and Drainage Engineering, 145(7).
Pérez-Blanco, C. D. & Gutiérrez-Martín, C. (2017). Buy me a river: Use of multi-attribute
non-linear utility functions to address overcompensation in agricultural water buyback.
Agricultural Water Management, 190:6–20.
Pérez-Sánchez, J. & Senent-Aparicio, J. (2015). Integrated water resources management
on a local scale: a challenge for the user community—a case study in Southern Spain.
Environmental Earth Sciences, 74(7):6097–6109.
Pérez-Uresti, S. I., Ponce-Ortega, J. M., & Jiménez-Gutiérrez, A. (2019). A multi-objective
optimization approach for sustainable water management for places with over-exploited
water resources. Computers & Chemical Engineering, 121:158–173.
Pianosi, F., Beven, K., Freer, J., Hall, J. W., Rougier, J., Stephenson, D. B., & Wagener, T.
(2016). Sensitivity analysis of environmental models: A systematic review with practical
workflow. Environmental Modelling & Software, 79:214–232.
References 169

Pianosi, F., Sarrazin, F., & Wagener, T. (2015). A Matlab toolbox for Global Sensitivity
Analysis. Environmental Modelling and Software, 70.

Pool, M., Carrera, J., Alcolea, A., & Bocanegra, E. (2015). A comparison of deterministic
and stochastic approaches for regional scale inverse modeling on the Mar del Plata aquifer.
Journal of Hydrology, 531:214–229.

Pulido-Velazquez, D., Collados-Lara, A.-J., & Alcalá, F. J. (2018). Assessing impacts of

future potential climate change scenarios on aquifer recharge in continental Spain. Journal
of Hydrology, 567:803–819.

Pulido-Velazquez, D., Garrote, L., Andreu, J., Martin-Carrasco, F.-J. J., & Iglesias, A.
(2011). A methodology to diagnose the effect of climate change and to identify adaptive
strategies to reduce its impacts in conjunctive-use systems at basin scale. Journal of
Hydrology, 405(1-2):110–122.

Pulido-Velazquez, M., Andreu, J., Sahuquillo, A., & Pulido-Velazquez, D. (2008).

Hydro-economic river basin modelling: The application of a holistic surface–groundwater
model to assess opportunity costs of water use in Spain. Ecological Economics, 66(1):51–65.

Räsänen, A., Juhola, S., Monge Monge, A., Käkönen, M., Kanninen, M., & Nygren, A.
(2017). Identifying mismatches between institutional perceptions of water-related risk
drivers and water management strategies in three river basin areas. Journal of Hydrology,
550:704–715.

Refsgaard, J. C. (1997). Parameterisation, calibration and validation of distributed

hydrological models. Journal of Hydrology, 198(1-4):69–97.

Relly, J. E. & Sabharwal, M. (2009). Perceptions of transparency of government

policymaking: A cross-national study. Government Information Quarterly, 26(1):148–157.

Reynaud, A. & Leenhardt, D. (2008). MoGIRE : A Model for Integrated Water Management.
American Geophysical Union, 2:576–583.

Riva, M., Guadagnini, A., De Gaspari, F., & Alcolea, A. (2010). Exact sensitivity matrix and
influence of the number of pilot points in the geostatistical inversion of moment equations
of groundwater flow. Water Resources Research, 46(11).

Riva, M., Guadagnini, A., Neuman, S., Janetti, E., & Malama, B. (2009). Inverse analysis
of stochastic moment equations for transient flow in randomly heterogeneous media.
Advances in Water Resources, 32(10):1495–1507.

Riva, M., Panzeri, M., Guadagnini, A., & Neuman, S. P. (2011). Role of model selection
criteria in geostatistical inverse estimation of statistical data- and model-parameters.
Water Resources Research, 47(7).
 170 References

Rogers, B. P., Hall, A. W., & Global Water Partnership (2003). Effective Water Governance.
Technical Report 7, Global Water Partnership.
Romero, J., Jerez, S., Muñoz, O., Castro, O., Bermudez, C., & Pardo, Y. (2015). Mapa
Sismotectónico en el Sector Norte del Valle Medio del Magdalena. Technical report,
Servicio Geológico Colombiano -SGC, Bogotá D.C.
Rosenberg, D. E., Howitt, R. E., & Lund, J. R. (2008). Water management with water
conservation, infrastructure expansions, and source variability in Jordan. Water Resources
Research, 44(11):1–11.
Ruelland, D., Ardoin-Bardin, S., Billen, G., & Servat, E. (2008). Sensitivity of a lumped
and semi-distributed hydrological model to several methods of rainfall interpolation on a
large basin in West Africa. Journal of Hydrology, 361(1-2):96–117.
Sahoo, S. & Jha, M. (2017). Numerical groundwater-flow modeling to evaluate potential
effects of pumping and recharge: implications for sustainable groundwater management
in the Mahanadi delta region, India | Modélisation numérique des écoulements d’eau
souterraine pour évaluer les . Hydrogeology Journal, 25(8):2489–2511.
Salazar, J. (2016). Una metodología para la estimación de curvas de duración de
caudales (cdc) en cuencas no instrumentadas. Caso de aplicación para Colombia en los
departamentos de Santander y norte de Santander. PhD thesis, Universidad Nacional de
Colombia.
Sanchez-León, E., Leven, C., Haslauer, C., & Cirpka, O. (2016). Combining 3D Hydraulic
Tomography with Tracer Tests for Improved Transport Characterization. Groundwater,
54(4):498–507.
Sarmiento-Rojas, L. F. (2011). Middle Magdalena Basin-Vol. 11 Petroleum Geology of
Colombia. Technical report, Agencia Nacional de Hidrocarburos, Bogotá D.C.
Schwabe, K., Knapp, K., & Luviano, I. (2017). Chapter 2.1.2 - The Water–Energy Nexus
and Irrigated Agriculture in the United States: Trends and Analyses. In Ziolkowska, J. R.
& Peterson, J. M., editors, Competition for Water Resources, pages 80–104. Elsevier.
Şen, Z. (2015). Basic Porous Medium Concepts. In Practical and Applied Hydrogeology,
chapter Water Scie, pages 43–97. Elsevier.
Servicio Geológico Colombiano (2014). Plan Estratégico del Conocimiento Geológico del
territorio Colombiano. Technical report, Servicio Geológico Colombiano, Bogotá.
Servicios Integrales Hidrogeológicos, S. (2015). Elaborar la Caracterización Hidrológica e
Hidrogeológica de las Áreas de Interés de las Zonas Hidrográficas del Meta y Magdalena
Medio en el Marco del Estudio Nacional del Agua en el Marco del Estudio Nacional del
Agua. Technical report, IDEAM, Bogota.
References 171

SGC & MINMINAS (2016). Historia geológica de los Andes colombianos en los alrededores
de Ibagué. Technical Report July, Servicio Geologico Colombiano, Bogota.
Sheikholeslami, R., Yassin, F., Lindenschmidt, K.-E., & Razavi, S. (2017). Improved
understanding of river ice processes using global sensitivity analysis approaches. Journal
of Hydrologic Engineering, 22(11).
Sieber, J., Yates, D., Huber-Lee, a., & Purkey, D. (2005). WEAP a demand, priority,
and preference driven water planning model: Part 1, model characteristics. Water
International, 30(4):487–500.
Sigdel, A., Jha, R., Bhatta, D., Abou-Shanab, R. A. I. Sapireddy, V. R., & Jeon, B. H.
(2011). Applicability of TOPMODEL in the catchments of nepal: Bagmati river basin.
Geosystem Engineering, 14(4):181–190.
Simmons, J., Harley, M., Marshall, L., Turner, I., Splinter, K., & Cox, R. (2017). Calibrating
and assessing uncertainty in coastal numerical models. Coastal Engineering, 125.
Singh, A., Walker, D., Minsker, B., & Valocchi, A. (2008). Interactive multi-objective inverse
groundwater modelling - Incorporating subjective knowledge and conceptual uncertainty.
In World Environmental and Water Resources Congress 2008: Ahupua’a - Proceedings of
the World Environmental and Water Resources Congress 2008, volume 316.
Sobol, I. (2001). Global sensitivity indices for nonlinear mathematical models and their
Monte Carlo estimates. Mathematics and Computers in Simulation, 55(1-3):271–280.
Sobol, I. M. (1993). Sensitivity analysis for nonlinear mathematical models. Math. Model.
[Link], 1(4):407–414.
Sochala, P. & Le Maître, O. (2013). Polynomial Chaos expansion for subsurface flows with
uncertain soil parameters. Advances in Water Resources, 62:139–154.
Song, X., Zhang, J., Zhan, C., Xuan, Y., Ye, M., & Xu, C. (2015). Global sensitivity
analysis in hydrological modeling: Review of concepts, methods, theoretical framework,
and applications. Journal of Hydrology, 523.
Sordo-Ward, A., Granados, A., Iglesias, A., Garrote, L., & Bejarano, M. D. (2019).
Adaptation effort and performance of water management strategies to face climate change
impacts in six representative basins of Southern Europe. Water (Switzerland), 11(5).
Spulber, D. F. (1988). Optimal environmental regulation under asymmetric information.
Journal of Public Economics, 35(2):163–181.
Sudret, B. (2008). Global sensitivity analysis using polynomial chaos expansion. Reliability
Engineering & System Safety, 93:964–979.
Suliman, A., Katimon, A., Darus, I., & Shahid, S. (2016). TOPMODEL for Streamflow
 172 References

Simulation of a Tropical Catchment Using Different Resolutions of ASTER DEM:

Optimization Through Response Surface Methodology. Water Resources Management,
30(9).

Sun, D., Zhao, C., Wei, H., & Peng, D. (2011). Simulation of the relationship between
land use and groundwater level in Tailan River basin, Xinjiang, China. Quaternary
International, 244(2):254–263.

Sun, L., Nistor, I., & Seidou, O. (2015). Streamflow data assimilation in SWAT model using
Extended Kalman Filter. Journal of Hydrology, 531:671–684.

Switzman, H., Coulibaly, P., & Adeel, Z. (2015). Modeling the impacts of dryland agricultural
reclamation on groundwater resources in Northern Egypt using sparse data. Journal of
Hydrology, 520:420–438.

Tavakoli, R. & Reynolds, A. (2009). History matching with parametrization based on the
SVD of a dimensionless sensitivity matrix. In SPE Reservoir Simulation Symposium
Proceedings, volume 1, pages 455–476.

Teng, J., Jakeman, A., Vaze, J., Croke, B., Dutta, D., & Kim, S. (2017). Flood inundation
modelling: A review of methods, recent advances and uncertainty analysis. Environmental
Modelling and Software, 90.

Tian, Y., Zheng, Y., & Zheng, C. (2016). Development of a visualization tool for integrated
surface water and groundwater modeling. Computers & Geosciences, 86:1–14.

Tiedeman, C. & Green, C. (2013). Effect of correlated observation error on parameters,

predictions, and uncertainty. Water Resources Research, 49(10):6339–6355.

Tonkin, M., Doherty, J., & Moore, C. (2007). Efficient nonlinear predictive error variance
for highly parameterized models. Water Resources Research, 43(7).

Tortajada, C., González-Gómez, F., Biswas, A. K., & Buurman, J. (2019). Water demand
management strategies for water-scarce cities: The case of Spain. Sustainable Cities and
Society, 45:649–656.

Tóth, A., Havril, T., Simon, S., Galsa, A., Monteiro-Santos, F. A., Muller, I., &
Mádl-Szonyi, J. (2016). Groundwater flow pattern and related environmental phenomena
in complex geologic setting based on integrated model construction. Journal of Hydrology,
539:330–344.

Tsai, F.-C. (2006). Enhancing random heterogeneity representation by mixing the kriging
method with the zonation structure. Water Resources Research, 42(8).

Tsai, F.-C. & Yeh, W.-G. (2011). Model calibration and parameter structure identification in
References 173

characterization of groundwater systems. Groundwater Quantity and Quality Management,

pages 159–202.

Tzabiras, J., Vasiliades, L., Sidiropoulos, P., Loukas, A., & Mylopoulos, N. (2016).
Evaluation of Water Resources Management Strategies to Overturn Climate Change
Impacts on Lake Karla Watershed. Water Resources Management, 30(15):5819–5844.

Usman, M., Reimann, T., Liedl, R., Abbas, A., Conrad, C., & Saleem, S. (2018). Inverse
parametrization of a regional groundwater flow model with the aid of modelling and
GIS: Test and application of different approaches. ISPRS International Journal of
Geo-Information, 7(1).

Valdés-Pineda, R., Pizarro, R., García-Chevesich, P., Valdés, J. B., Olivares, C., Vera, M.,
Balocchi, F., Pérez, F., Vallejos, C., Fuentes, R., Abarza, A., & Helwig, B. (2014). Water
governance in Chile: Availability, management and climate change. Journal of Hydrology,
519(APRIL):2538–2567.

Valencia, C. (2014). Más petróleo, menos agua.

Van Der Heijden, S. & Haberlandt, U. (2015). A fuzzy rule based metamodel for monthly
catchment nitrate fate simulations. Journal of Hydrology, 531:863–876.

Vargas, N. (2006). Zonas Hidrogeológicas Homogéneas de Colombia. Boletín Geológico y

Minero, 117(1):47–61.

Vargas Martínez, N. O., Campillo Pérez, A. K., García Herrán, M., & Jaramillo Rodríguez,
O. (2013). Aguas Subterráneas en Colombia: una Visión General. IDEAM, Bogota.

Velázquez, M. A. P., Álvarez, J. A., & Herraiz, A. S. (2003). Optimización económica de la

gestión del uso conjunto de aguas superficiales y subterráneas en un sistema de recursos
hídricos: contribución al análisis económico propuesto en la Directiva Marco Europea del
Agua. PhD thesis, Universidad Politecnica de Valencia.

Vieira, S. M. (2009). World Water Assesment Programme. The United Nations World Water
Development Report, 5(3):153.

Wagner, P. D., Fiener, P., Wilken, F., Kumar, S., & Schneider, K. (2012). Comparison and
evaluation of spatial interpolation schemes for daily rainfall in data scarce regions. Journal
of Hydrology, 464-465:388–400.

Walker, D., Jakovljević, D., Savić, D., & Radovanović, M. (2015). Multi-criterion water
quality analysis of the Danube River in Serbia: A visualisation approach. Water research,
79:158–72.

Wang, S., Huang, G., Baetz, B., & Huang, W. (2015). A polynomial chaos ensemble
 174 References

hydrologic prediction system for efficient parameter inference and robust uncertainty
assessment. Journal of Hydrology, 530:716–733.

Wang, X., Jardani, A., & Jourde, H. (2017a). A hybrid inverse method for hydraulic
tomography in fractured and karstic media. Journal of Hydrology, 551:29 – 46.

Wang, X.-S., Wan, L., Jiang, X.-W., Li, H., Zhou, Y., Wang, J., & Ji, X. (2017b). Identifying
three-dimensional nested groundwater flow systems in a Tóthian basin. Advances in Water
Resources, 108:139 – 156.

Ward, D., Goldsmith, R., Jimeno, A., Cruz, J., Restrepo, H., & Gomez, E. (1977). Plancha
Geologica 109- Rionegro.

Ward, F. A. (2009). Economics in integrated water management. Environmental Modelling

& Software, 24(8):948–958.

White, J. T. (2018). A model-independent iterative ensemble smoother for efficient

history-matching and uncertainty quantification in very high dimensions. Environmental
Modelling & Software, 109:191–201.

White, J. T., Fienen, M. N., & Doherty, J. E. (2016). A python framework for environmental
model uncertainty analysis. Environmental Modelling & Software, 85:217 – 228.

Wiant, H. V. J. & Harner, E. J. (1979). Percent Bias and Standard Error in Logarithmic
Regression. Forest Science, 25(1):167–168.

Wongsa, S. (2015). Impact of Climate Change on Water Resources Management in the

Lower Chao Phraya Basin, Thailand. Journal of Geoscience and Environment Protection,
03(10):53–58.

Woodrow, K., Lindsay, J. B., & Berg, A. A. (2016). Evaluating DEM conditioning techniques,
elevation source data, and grid resolution for field-scale hydrological parameter extraction.
Journal of Hydrology, 540:1022–1029.

Wu, Q., Liu, S., Cai, Y., Li, X., & Jiang, Y. (2017). Improvement of hydrological model
calibration by selecting multiple parameter ranges. Hydrology and Earth System Sciences,
21(1).

Wu, X., Jeuland, M., & Whittington, D. (2016). Does political uncertainty affect water
resources development? The case of the Eastern Nile. Policy and Society, 35(2):151–163.

Xie, S., Hu, Y., Jiang, M., & Liu, Q. (2006). Study on the inverse method to permeability
coefficient of groundwater system: A case study of uranium gangue site in Southern China.
In 5th International Conference on Environmental Informatics, ISEIS 2006.

Xie, Y., Cook, P. G., Simmons, C. T., Partington, D., Crosbie, R., & Batelaan, O. (2018).
References 175

Uncertainty of groundwater recharge estimated from a water and energy balance model.
Journal of Hydrology, 561:1081–1093.
Xue, L., Yang, F., Yang, C., Wei, G., Li, W., & He, X. (2018). Hydrological simulation
and uncertainty analysis using the improved TOPMODEL in the arid Manas River basin,
China. Scientific Reports, 8(1):1–12.
Yao, L. & Guo, Y. (2014). Hybrid algorithm for parameter estimation of the groundwater
flow model with an improved genetic algorithm and gauss-newton method. Journal of
Hydrologic Engineering, 19(3):482–494.
Yasari, E., Pishvaie, M. R., Khorasheh, F., Salahshoor, K., & Kharrat, R. (2013). Application
of multi-criterion robust optimization in water-flooding of oil reservoir. Journal of
Petroleum Science and Engineering, 109:1–11.
Yeh, W.-G. (2015). Review: Optimization methods for groundwater modeling and
management | Revue: Méthodes d’optimisation pour la modélisation et la gestion des
eaux souterraines | Revisão: Métodos de otimização para modelagem e gerenciamento de
águas subterrâneas | Revisión. Hydrogeology Journal, 23(6):1051–1065.
Yi, L., Zhang, W.-C., & Yan, C.-A. (2017). A modified topographic index that incorporates
the hydraulic and physical properties of soil. Hydrology Research, 48(2):370–383.
Yoon, H. & McKenna, S. (2012). Highly parameterized inverse estimation of hydraulic
conductivity and porosity in a three-dimensional, heterogeneous transport experiment.
Water Resources Research, 48(10).
Young, R. (1996). Measuring economic benefits for water investments and policies. World
Bank technical paper, 1(338):140.
Zhang, J., Fan, S.-k., Yang, J.-c., Du, X.-m., Li, F.-s., & Hou, H. (2014). Petroleum
contamination of soil and water, and their effects on vegetables by statistically analyzing
entire data set. The Science of the total environment, 476-477:258–65.
Zhang, M. & Burbey, T. (2016). A comparison of three hydraulic parameter optimization
schemes for Las Vegas valley. Environmental and Engineering Geoscience, 22(3):173–194.
Zhang, X., Sun, A. Y., & Duncan, I. J. (2016). Shale gas wastewater management under
uncertainty. Journal of Environmental Management, 165:188–198.
Zhou, H., Gómez-Hernández, J. J., & Li, L. (2014). Inverse methods in hydrogeology:
Evolution and recent trends. Advances in Water Resources, 63.
Zimmerman, J. (2016). GIS Topographic Wetness Index (TWI) Exercise Steps.