Design Flood Estimation Using RORB
7th-8th December 2016
Melbourne
Day 1 Session 1
Introduction to Runoff Routing
Overview
• Context
• Formation of actual floods
• Flood simulation - basic concepts
• Overview of modelling components
• Runoff routing models:
– background
– model structure, parameters, inputs
January 2011: Lochyer Valley, Qld
0
min
January 2011: Lochyer Valley, Qld
0 +11
min min
January 2011: Lochyer Valley, Qld
0 +11
min min
+18
min
January 2011: Lochyer Valley, Qld
0 +11
min min
+18 +22
min min
The context for flood estimation
• Flood management (eg zoning, mitigation structures) is
concerned with future events
• A risk analysis of some form is used to decide how severe an
event will be used for design
– for some structures (eg stormwater drainage systems) a peak
flow is sufficient.
– where flood volumes (and/or flow durations) are an issue,
design hydrographs are needed
• The standard method in Australia for design hydrograph
estimation is runoff routing
Flood Formation
What factors are important in the formation
of actual floods?
– Climate inputs (especially rainfall):
• sequence of storm events
• temporal variability of rainfall within storms
• high degree of spatial variations of rainfall
– Catchment characteristics:
• great variety of catchment features
• high degree of spatial variability
• land use changes
Flood Formation
– Hydrologic processes:
• continuum of many different processes
• physically complex, non-linear
• interactions (eg. surface/sub-surface hydrology)
• spatially variable within a catchment
• differences between catchments (depending on
catchment & climate characteristics)
=> Flood simulation represents a challenge
The Flood Event
12000 0
Rainfall
Discharge 2
10000
4
8000
Discharge (ML/d)
Rainfall (mm/h)
6
6000 8
10
4000
12
2000
14
0 16
108
120
132
144
156
168
180
192
204
12
24
36
48
60
72
84
96
0
Catchment response
Time (h)
to rainfall
Flood Simulation
Hydrologic Processes:
Dominant processes are represented by
3 separate conceptual models:
• Runoff production model (or loss model):
converts local rainfall to runoff (or rainfall excess)
• Hydrograph formation (or runoff routing) model:
transforms local runoff to h/g at catchment outlet
• Baseflow model:
reflects sub-surface contribution to streamflow
Flood Simulation - Components
Rainfall Input
(design storms)
Runoff Production
(loss model)
Baseflow
Production Rainfall Excess
Hydrograph Formation
Baseflow
(Routing Model)
Flood Output
Nature of Flood Event Modelling
• Configuration of a suitable model that simulates flood response of
a catchment in response to rainfall
• Can be used to estimate:
– Floods from historical rainfalls
– Floods of a specified annual exceedance probability (AEP)
– Impacts of urbanisation or land-use change (difficult)
• “Design Flood” is term given to a flood peak of specified AEP which
is derived from “Design Rainfall” (eg BoM IFD2013)
• Design floods traditionally estimated using “design event”
(deterministic) procedures
• Increasing focus given to using “joint probability” (stochastic)
approaches
The Rational Method Equation
Q = 1/3600 * C.I.A.
• Q (m3/s)
• C (dimensionless)
• I (mm/h) (assuming duration equals time of concentration)
• A (hectares, ha) 1ha = 104m2 = 10-2km2
• No flow data is required
• Useful (in theory) if only peak flow is required from a given design
rainfall IFD for small catchments e.g. Urban and rural drainage design
• Widely used in practice, but has low confidence when applied to
ungauged catchments
Unit Hydrograph Models
• Unit hydrograph is the hydrograph resulting from unit depth of
surface runoff produced by a storm of uniform intensity and
specified duration
• It assumes linearity in the transformation of rainfall excess to
flood response
• Discussed in just about all engineering hydrology texts
• Its use in Australia has been replaced by runoff-routing models,
but it remains popular overseas (in conjunction with network
routing models)
• It treats (sub-)catchment as a “lumped” system
• Not easily modified for catchments of different size to those used
in calibration, or for regional purposes
• Not easily adapted to urban systems
Network Model
Muskingum-Cunge method used to Unit hydrograph model used to convert
route hydrographs along stream channels Rainfall into sub-catchment hydrograph
Lag and route models
• Do not have strong rigorous physical basis but do provide
reasonable results
• Translation with (non-linear) concentrated storage at
outlet
– ILLUDAS
– ILLSAX
Runoff Routing Models
Concepts:
• The effect of a catchment on the input rainfall-
excess is a storage effect, ie:
– attenuation of the input; and
– translation in time
• Can simulate this storage effect by routing
rainfall-excess through a series of
concentrated storages
Simulation of catchment storage
Rainfall (mm/hr)
Streamflow (mm/hr)
Time (hr)
Simulation of catchment storage
Losses
Rainfall (mm/hr)
Net Rainfall (Re)
Re S1 S2 S3 Storage routing model
Q1
Attenuation
Q2
Streamflow (mm/hr)
Q3 Translation and attenuation of
hydrograph moving downstream
(shown for three locations Q1, Q2, Q3)
Translation
Time (hr)
Runoff Routing Models
Storage elements are
selected to represent
catchment (node-link
structure)
Representation of Reach Storage
(S = kQm)
• Between each node on the stream network is
a storage of form S = kQm
• The form of the storage equation S = kQm is
justifiable by both theoretical and empirical
means.
• For example, application of a uniform flow
equation in a prismatic channel will yield an
equation of this form
Value of exponent m
• Triangular cross-section m = 0.75
• Trapezoidal cross-section m = 0.74
• Parabolic cross-section m = 0.69
• Wide rectangular channel m = 0.60
The results are for uniform flow in open channels. However,
similar m values, ranging from 0.68 to 0.8, have been reported
in field studies using natural catchments
Normally, an m value of 0.8 is adopted in practice (unless there is
evidence to the contrary)
Value of k
• k ( Equation A-3) dependent on channel roughness, cross-
section shape, bed-slope, and length (prismatic channels)
• in natural river channels the effects of slope and roughness
often tend to be compensatory; in such cases, the k value can
be taken as proportional to L.
• in urban catchments with lined channels, this compensating
effect is missing so the effect of slope on reach storage must
be specifically included.
Splitting k into kc and kr
• there are advantages in specifying k as the
product of kc and kr (ie k = kc x kr)
• kc - a catchment parameter (rural) related to
catchment lag formulae (applicable to entire
catchment and stream network)
• kr - a reach parameter (usually reach length)
expressed in dimensionless form
– it can be adjusted for changes in reach condition – eg
forming, lining, drowning)
– applicable to individual reaches
Runoff Routing - Basic Steps
• Development/application of a runoff routing
model involves 3 basic steps:
1. Defining model structure
2. Deriving model parameters
3. Specifying model inputs
Model Structure
• Select storage elements to represent catchment (node-link
structure):
Guidelines for Catchment Subdivision
• At least 4 subcatchments upstream of any
streamflow gauge or point of interest
• Subcatchments all of similar size (generally
largest subcatchment no more than 3 times
the area of the smallest subcatchment)
• Sufficient number of subcatchments to
capture spatial and temporal variability in
observed rainfalls
Guidelines for Catchment Subdivision
• Divide catchment into several sub-catchments
(generally between 5 and 20)
– Sub-catchment boundaries along watershed divides
– Sub-catchments should be reasonably similar in size
– Indicative sub-area ranges (though this varies widely!):
• 4 sub-areas for 0.1 km2
• 7 sub-areas for 10 km2
• 15 sub-areas for 10 km2
• Place model nodes at:
– Centroid of each sub-catchment
– Confluence of streams
– Significant inflow/outflow or artificial storages
Model Structure
• Select storage elements to represent catchment (node-link
structure):
Model Structure - Issues
• Many possible model structures could
be used to represent a catchment but:
– Different types of elements imply
different S-Q relationships
=> different results when extrapolated
– Which representation performs best for:
• modelling different land use conditions?
• estimating larger floods?
• application to ungauged catchments?
parameters = f(catchment characteristics)
Summary – Session 1
• Flood simulation - basic concepts
• Modelling components
• Runoff routing models:
– background
– model structure, parameters, inputs
