The use of a deterministic fractal-multifractal (FM) representation to model high-resolution rain... more The use of a deterministic fractal-multifractal (FM) representation to model high-resolution rainfall time series via projections of fractal interpolating functions weighed by multifractal measures is reported. It is shown that the intrinsic shape and variability of an 8-hour Boston storm recorded every 15 s on October 25, 1980, may be encoded wholistically, employing the fractal geometric methodology. It is illustrated that the FM methodology provides very faithful descriptions of both major trends and small (noisy) fluctuations for this storm, resulting in preservation of not only classical statistical characteristics of the records but also rnultifractal and chaotic properties present in them. These results, and those for other storms, suggest that a stochastic framework for rainfall may be bypassed in favor of a deterministic representation based on projections. 1. Introduction Modeling the structure of temporal rainfall has attracted much attention in the literature for the last 30 years. Although several sophisticated (deterministic and stochastic) rainfall models exist, they are built to preserve some characteristics of the records and consequently do not capture all the variability observed at a fixed location when a storm passes by. Typically, these models preserve some statistics of the available data set but do not keep the intermittent details present in the records. Since knowing details may be important in predicting the rainfall process, an alternative approach to temporal rainfall should be considered. Existing temporal rainfall representations may be classified dependcling on their basic building blocks, as follows: (1) physically based, (2) point process, (3) chaotic, and (4) fractal geometric. Physically based models approximate the physical processes giving rise to the observed rainfall [e.g., Georgakakos and Bras, 1984]. By employing basic thermodynamics, cloud microphysics principles, and integration of the cloud on top of a recording station, these representations produce intermittent outcomes which nonetheless do not capture the inherent details present in the records. This happens, in part, because of approximated parameterizations which reflect our lack of understanding of all the processes taking place within the atmosphere. Typical stochastic point process models approximate the irregular and complex rain patterns (of a fractal and/or multifractal nature) by superimposing randomly arriving Euclidean objects (e.g., rectangular pulses). Models with Poisson and cluster-based arrival processes have been defined such that some statistical features of the rainfall time series (e.g., mean, variance, autocorrelations, e.tc.) are preserved (see, for instance, Eagleson [1978]; Kavvas and Delleur [1981]; Smith and Karr [1983]; and Rodriguez-Iturbe [1986], among others). Although it has been found that cluster-based procedures are better than Poisson-based ones [e.g., •aldes et al., 1985; Rodriguez-Iturbe et al., 1987], inconsistencies have been ob-
The article published in Vol. 10, No. 1 (2002), pp. 91-102, has some figures misplaced relative t... more The article published in Vol. 10, No. 1 (2002), pp. 91-102, has some figures misplaced relative to their captions, as follows. The graphs on Figs. 7 and 8 should be on Figs. and, and vice versa.
Usage of a deterministic fractal-multifractal (FM) procedure to model high-resolution rainfall ti... more Usage of a deterministic fractal-multifractal (FM) procedure to model high-resolution rainfall time series, as derived distributions of multifractal measures via fractal interpolating functions, is reported. Four rainfall storm events having distinct geometries, one gathered in Boston and three others observed in Iowa City, are analyzed. Results show that the FM approach captures the main characteristics of these events, as the fitted storms preserve the records' general trends, their autocorrelations and spectra, and their multifractal character.
Natural Hazards and Earth System Sciences, Mar 23, 2016
In this paper a method is proposed to identify mountainous watersheds with the highest flood risk... more In this paper a method is proposed to identify mountainous watersheds with the highest flood risk at the regional level. Through this, the watersheds to be subjected to more detailed risk studies can be prioritised in order to establish appropriate flood risk management strategies. The prioritisation is carried out through an index composed of a qualitative indicator of vulnerability and a qualitative flash flood/debris flow susceptibility indicator. At the regional level, vulnerability was assessed on the basis of a principal component analysis carried out with variables recognised in literature to contribute to vulnerability, using watersheds as the unit of analysis. The area exposed was obtained from a simplified flood extent analysis at the regional level, which provided a mask where vulnerability variables were extracted. The vulnerability indicator obtained from the principal component analysis was combined with an existing susceptibility indicator, thus providing an index that allows the watersheds to be prioritised in support of flood risk management at regional level. Results show that the components of vulnerability can be expressed in terms of three constituent indicators: (i) socio-economic fragility, which is composed of demography and lack of well-being; (ii) lack of resilience and coping capacity, which is composed of lack of education, lack of preparedness and response capacity, lack of rescue capacity, cohesiveness of the community; and (iii) physical exposure, which is composed of exposed infrastructure and exposed population. A sensitivity analysis shows that the classification of vulnerability is robust for watersheds with low and high values of the vulnerability indicator, while some watersheds with intermediate values of the indicator are sensitive to shifting between medium and high vulnerability.
Hydrology and Earth System Sciences Discussions, 2016
A distributed model (TETIS), a semi-distributed model (TOPMODEL) and a lumped model (HEC HMS soil... more A distributed model (TETIS), a semi-distributed model (TOPMODEL) and a lumped model (HEC HMS soil moisture accounting) were used to simulate the discharge response of a tropical high mountain basin characterized by soils with high water storage capacity and high conductivity. The models were calibrated with the Shuffle Complex Evolution algorithm, using the Kling and Gupta efficiency as objective function. Performance analysis and diagnostics were carried out using the signatures of the flow duration curve and through analysis of the model fluxes in order to identify the most appropriate model for the study area for flood early warning. The impact of varying grid sizes was assessed in the TETIS model and the TOPMODEL in order to chose a model with balanced model performance and computational efficiency. The sensitivity of the models to variation in the precipitation input was analysed by forcing the models with a rainfall ensemble obtained from Gaussian simulation. The resulting discharge ensembles of each model were compared in order to identify differences among models structures. The results show that TOPMODEL is the most realistic model of the three tested, albeit showing the largest discharge ensemble spread. The main differences among models occur between HEC HMS soil moisture accounting and TETIS, and HEC HMS soil moisture accounting and TOPMODEL, with HEC HMS soil moisture accounting producing ensembles in a range lower than the other two models. The ensembles of TETIS and TOPMODEL are more similar.
This chapter presents an approach to forecast criminal patterns that combines the time series fro... more This chapter presents an approach to forecast criminal patterns that combines the time series from clustering method with a computational intelligence-based prediction. In this approach, clusters of criminal events are parametrized according to simple geometric prototypes. Cluster dynamics are captured as a set of time series. The size of this set corresponds to the number of clusters multiplied by the number of parameters per cluster. One of the main drawbacks of clustering is the difficulty of defining the optimal number of clusters. The paper also deals with this problem by introducing a validation index of dynamic partitions of crime events that relates the optimal number of clusters with the foreseeability of time series by means of non-linear analysis. The method as well as the validation index was tested over two cases of reported urban crime. Our results showed that crime clusters can be predicted by forecasting their representative time series using an evolutionary adaptive neural fuzzy inference system. Thus, we argue that the foreseeability of these series can be anticipated satisfactorily by means of the proposed index.
Contexto: Se ha observado que la modelación de sistemas socio-ecológicos complejos a partir de si... more Contexto: Se ha observado que la modelación de sistemas socio-ecológicos complejos a partir de simulaciones basadas en agentes, tienen la ventaja de integrar diferentes procesos, escalas, variables y la posibilidad de generar escenarios con actores en el marco de procesos de modelación participativa.Método: Se diseña una herramienta computacional para la planificación y apoyo a los procesos de toma de decisión relacionados con la gestión de los recursos hídricos, específicamente para el caso de las inundaciones en la ecorregión de la Mojana (Colombia). Son desarrollados talleres de participación social relacionados con creencias, valores, redes sociales y resiliencia.Resultados: El modelo tiene dos componentes: el primero representa la hidrodinámica de las inundaciones por medio de la plataforma numérica ISIS 2D; el segundo, en relación con los aspectos sociales de la región, se maneja a través de modelado de sistemas de agentes. Ambos sistemas están integrados en la plataforma Net...
Considering the evaluation of acquired knowledge as a fundamental element of Computerized Tutorin... more Considering the evaluation of acquired knowledge as a fundamental element of Computerized Tutoring Systems, the classical Bayesian methods for adaptive testing present several deficiencies, such as arbitrary assignments or prior probabilities, non representativity, the effect of the context and the need to include all the concepts under evaluation. An alternative model of the adaptive test is proposed here based on content-balanced and imprecise interpretations of the information level and prior distribution of each item using Gil's fuzzy relationships instead of Bayesian relationships.
Successful usage of a large family of deterministically generated measures to model complex nonli... more Successful usage of a large family of deterministically generated measures to model complex nonlinear phenomena (e.g. rainfall, turbulence and groundwater contaminant transport) has been reported recently.1–7 As these measures, generated as derived distributions of multifractal measures via fractal interpolating functions (FIF), i.e. the fractal-multifractal (FM) approach, have been found to share the inherent character of natural sets, the present study further investigates their dynamical (chaotic or stochastic) properties. It is shown, through a variety of examples and via the use of power spectrum, mass exponents and false nearest neighbors in the state-space, that the FM approach indeed generates deterministic measures whose behavior (depending on their parameters) may be classified as low-dimensional and chaotic or as high-dimensional and stochastic. These results suggest the general suitability of the FM approach for understanding and modeling nonlinear natural phenomena.
... Felipe Prada-Sarmiento, Nelson Obregón-Neira. ... compared to the observed rescaled discharge... more ... Felipe Prada-Sarmiento, Nelson Obregón-Neira. ... compared to the observed rescaled discharges and to the initially trained ANN model results, resorting to some statistical parameters such as , mean absolute error ; root mean square error ; scatter index ; and Nash index: , where ...
Resumen: Los Algoritmos Evolutivos forman parte de la Inteligencia Artificial la cual a su vez se... more Resumen: Los Algoritmos Evolutivos forman parte de la Inteligencia Artificial la cual a su vez se encuentra integrada a un tipo de Sistema Inteligente, la Inteligencia Computacional. En este contexto, este trabajo estudia modelos derivados de los Algoritmos Genéticos (AG) y de la Programación Genética (PG). En particular, se revisan sus potencialidades de aplicación en estudios hidroinformáticos. Específicamente en problemas de optimización matemática global no lineal altamente dimensional, los cuales caracterizan los problemas inversos generados en procesos de calibración de modelos hidrológicos e hidráulicos. Para ilustrar la aplicación hidroinformática de los AG y la PG, se proporcionan ejemplos ilustrativos para el caso de modelos de balance hídrico y de obtención de relaciones funcionales entre variables hídricas de entrada y salida, respectivamente. De esta forma, se discute cómo en los procesos de gestión y particularmente de modelación de sistemas hidroinformáticos, los mode...
RESUMEN: En el diseño y manejo de sistemas de drenaje urbano es necesario conocer distribución es... more RESUMEN: En el diseño y manejo de sistemas de drenaje urbano es necesario conocer distribución espacial de la precipitación. Para obtener esta información, en la ciudad de Bogotá se emplean con frecuencia métodos de interpolación cuya entrada son los registros puntuales de estaciones pluviométricas o pluviográfi cas. En el presente trabajo se presenta una revisión de los parámetros básicos que los modeladores deben asumir antes de realizar una interpolación espacial. Se utilizó información diaria y subhoraria, considerando los métodos de la distancia inversa cuadrática, Kriging y V4 de MATLAB®. Los resultados obtenidos permiten defi nir algunos criterios sencillos en cuanto a la orientación de la malla, el tamaño de la celda a emplear, el análisis estructural previo que debe realizársele a la información y el método de interpolación a utilizar.
Este Articulo muestra los resultados obtenidos en la implementación del modelo neurodifuso ANFIS ... more Este Articulo muestra los resultados obtenidos en la implementación del modelo neurodifuso ANFIS y de las redes neuronales para la predicción de caudales medios mensuales. Se desarrolla e Implementa el modelo ANFIS y se evalúa el comportamiento de 6 modelos al variar el número de entradas, el número y tipo de conjuntos difusos ( funciones de pertenencia), que son los parámetros fundamentales del modelo ANFIS y se compara los resultados con los obtenidos con las redes neuronales perceptron multicapa.
Carbofuran and oxadixyl pesticides are used in Colombia to control pests and fungi, but their mob... more Carbofuran and oxadixyl pesticides are used in Colombia to control pests and fungi, but their mobility through the soil profile is poorly understood. This study showed degradation and adsorption processes of these compounds in a Melanudands soil (0-100 cm) from Colombia using laboratory incubation and the batch equilibrium methods. First-order kinetic models indicated that the degradation rates of carbofuran (0.013-0.006 day-1) and oxadixyl (0.013-0.008 day-1) decreased at deeper soil layers, suggesting that the pesticides were more persistent in the sub-surface (60-100 cm) than in the surface layers (0-40 cm). The thermodynamic approach showed that the adsorption of both pesticides was similar, an exothermic and spontaneous process. The carbofuran and oxadixyl coefficient of distribution (5.8-0.3 L kg-1) and the percentage of adsorption (71.2-11.3%) were very similar in the surface layers (0-40 cm) and decreased with the soil depth. The organic carbon (OC) and clay content showed a...
This article presents the results of the research work of the GIGA research group of the Distrita... more This article presents the results of the research work of the GIGA research group of the Distrital University in Bogota Colombia in the framework of the advanced doctoral research in the engineering doctorate of the Javeriana University and shows the development followed and the corresponding analysis determine the effectiveness of the M5P algorithm within the decision tree approach for the management of BIG DATA and Machine Learning for the determination of massive property valuation. The analysis was carried out with data provided by the Special Administrative Unit of the Distrital Cadaster (UAECD) and the Technical Cadastral Observatory corresponding to three sectors of the city of Bogotá D.C. The results of the J48-based model were compared by means of a statistical analysis with the traditionally used linear regression method, obtaining satisfactory results with errors well below the linear regressions in the training, validation and forecast stages.
ASME-ARPEL 2021 International Pipeline Geotechnical Conference, 2021
Cenit Transporte y Logística de Hidrocarburos (CENIT), operator of about 7000 km of hydrocarbon t... more Cenit Transporte y Logística de Hidrocarburos (CENIT), operator of about 7000 km of hydrocarbon transport systems, which constitutes it the largest operator in Colombia, has developed a strategic alliance to structure an adaptive geotechnical susceptibility zoning using supervised learning algorithms. Through this exercise, has been implemented operational decision inferences with simple linguistic values. The difficulties proposed by the method considers the hydroclimatology of Colombia, which is conditioned by several phenomena of Climate Variability that affect the atmosphere at different scales such as the Oscillation of the Intertropical Convergence Zone - ITCZ (seasonal scale) and the occurrence of macroclimatic phenomena such as El Niño-La Niña Southern Oscillation (ENSO) (interannual scale). Likewise, it considers the geotechnical complexity derived from the different geological formation environments, the extension and geographical dispersion of the infrastructure, and its ...
The use of a deterministic fractal-multifractal (FM) representation to model high-resolution rain... more The use of a deterministic fractal-multifractal (FM) representation to model high-resolution rainfall time series via projections of fractal interpolating functions weighed by multifractal measures is reported. It is shown that the intrinsic shape and variability of an 8-hour Boston storm recorded every 15 s on October 25, 1980, may be encoded wholistically, employing the fractal geometric methodology. It is illustrated that the FM methodology provides very faithful descriptions of both major trends and small (noisy) fluctuations for this storm, resulting in preservation of not only classical statistical characteristics of the records but also rnultifractal and chaotic properties present in them. These results, and those for other storms, suggest that a stochastic framework for rainfall may be bypassed in favor of a deterministic representation based on projections. 1. Introduction Modeling the structure of temporal rainfall has attracted much attention in the literature for the last 30 years. Although several sophisticated (deterministic and stochastic) rainfall models exist, they are built to preserve some characteristics of the records and consequently do not capture all the variability observed at a fixed location when a storm passes by. Typically, these models preserve some statistics of the available data set but do not keep the intermittent details present in the records. Since knowing details may be important in predicting the rainfall process, an alternative approach to temporal rainfall should be considered. Existing temporal rainfall representations may be classified dependcling on their basic building blocks, as follows: (1) physically based, (2) point process, (3) chaotic, and (4) fractal geometric. Physically based models approximate the physical processes giving rise to the observed rainfall [e.g., Georgakakos and Bras, 1984]. By employing basic thermodynamics, cloud microphysics principles, and integration of the cloud on top of a recording station, these representations produce intermittent outcomes which nonetheless do not capture the inherent details present in the records. This happens, in part, because of approximated parameterizations which reflect our lack of understanding of all the processes taking place within the atmosphere. Typical stochastic point process models approximate the irregular and complex rain patterns (of a fractal and/or multifractal nature) by superimposing randomly arriving Euclidean objects (e.g., rectangular pulses). Models with Poisson and cluster-based arrival processes have been defined such that some statistical features of the rainfall time series (e.g., mean, variance, autocorrelations, e.tc.) are preserved (see, for instance, Eagleson [1978]; Kavvas and Delleur [1981]; Smith and Karr [1983]; and Rodriguez-Iturbe [1986], among others). Although it has been found that cluster-based procedures are better than Poisson-based ones [e.g., •aldes et al., 1985; Rodriguez-Iturbe et al., 1987], inconsistencies have been ob-
The article published in Vol. 10, No. 1 (2002), pp. 91-102, has some figures misplaced relative t... more The article published in Vol. 10, No. 1 (2002), pp. 91-102, has some figures misplaced relative to their captions, as follows. The graphs on Figs. 7 and 8 should be on Figs. and, and vice versa.
Usage of a deterministic fractal-multifractal (FM) procedure to model high-resolution rainfall ti... more Usage of a deterministic fractal-multifractal (FM) procedure to model high-resolution rainfall time series, as derived distributions of multifractal measures via fractal interpolating functions, is reported. Four rainfall storm events having distinct geometries, one gathered in Boston and three others observed in Iowa City, are analyzed. Results show that the FM approach captures the main characteristics of these events, as the fitted storms preserve the records' general trends, their autocorrelations and spectra, and their multifractal character.
Natural Hazards and Earth System Sciences, Mar 23, 2016
In this paper a method is proposed to identify mountainous watersheds with the highest flood risk... more In this paper a method is proposed to identify mountainous watersheds with the highest flood risk at the regional level. Through this, the watersheds to be subjected to more detailed risk studies can be prioritised in order to establish appropriate flood risk management strategies. The prioritisation is carried out through an index composed of a qualitative indicator of vulnerability and a qualitative flash flood/debris flow susceptibility indicator. At the regional level, vulnerability was assessed on the basis of a principal component analysis carried out with variables recognised in literature to contribute to vulnerability, using watersheds as the unit of analysis. The area exposed was obtained from a simplified flood extent analysis at the regional level, which provided a mask where vulnerability variables were extracted. The vulnerability indicator obtained from the principal component analysis was combined with an existing susceptibility indicator, thus providing an index that allows the watersheds to be prioritised in support of flood risk management at regional level. Results show that the components of vulnerability can be expressed in terms of three constituent indicators: (i) socio-economic fragility, which is composed of demography and lack of well-being; (ii) lack of resilience and coping capacity, which is composed of lack of education, lack of preparedness and response capacity, lack of rescue capacity, cohesiveness of the community; and (iii) physical exposure, which is composed of exposed infrastructure and exposed population. A sensitivity analysis shows that the classification of vulnerability is robust for watersheds with low and high values of the vulnerability indicator, while some watersheds with intermediate values of the indicator are sensitive to shifting between medium and high vulnerability.
Hydrology and Earth System Sciences Discussions, 2016
A distributed model (TETIS), a semi-distributed model (TOPMODEL) and a lumped model (HEC HMS soil... more A distributed model (TETIS), a semi-distributed model (TOPMODEL) and a lumped model (HEC HMS soil moisture accounting) were used to simulate the discharge response of a tropical high mountain basin characterized by soils with high water storage capacity and high conductivity. The models were calibrated with the Shuffle Complex Evolution algorithm, using the Kling and Gupta efficiency as objective function. Performance analysis and diagnostics were carried out using the signatures of the flow duration curve and through analysis of the model fluxes in order to identify the most appropriate model for the study area for flood early warning. The impact of varying grid sizes was assessed in the TETIS model and the TOPMODEL in order to chose a model with balanced model performance and computational efficiency. The sensitivity of the models to variation in the precipitation input was analysed by forcing the models with a rainfall ensemble obtained from Gaussian simulation. The resulting discharge ensembles of each model were compared in order to identify differences among models structures. The results show that TOPMODEL is the most realistic model of the three tested, albeit showing the largest discharge ensemble spread. The main differences among models occur between HEC HMS soil moisture accounting and TETIS, and HEC HMS soil moisture accounting and TOPMODEL, with HEC HMS soil moisture accounting producing ensembles in a range lower than the other two models. The ensembles of TETIS and TOPMODEL are more similar.
This chapter presents an approach to forecast criminal patterns that combines the time series fro... more This chapter presents an approach to forecast criminal patterns that combines the time series from clustering method with a computational intelligence-based prediction. In this approach, clusters of criminal events are parametrized according to simple geometric prototypes. Cluster dynamics are captured as a set of time series. The size of this set corresponds to the number of clusters multiplied by the number of parameters per cluster. One of the main drawbacks of clustering is the difficulty of defining the optimal number of clusters. The paper also deals with this problem by introducing a validation index of dynamic partitions of crime events that relates the optimal number of clusters with the foreseeability of time series by means of non-linear analysis. The method as well as the validation index was tested over two cases of reported urban crime. Our results showed that crime clusters can be predicted by forecasting their representative time series using an evolutionary adaptive neural fuzzy inference system. Thus, we argue that the foreseeability of these series can be anticipated satisfactorily by means of the proposed index.
Contexto: Se ha observado que la modelación de sistemas socio-ecológicos complejos a partir de si... more Contexto: Se ha observado que la modelación de sistemas socio-ecológicos complejos a partir de simulaciones basadas en agentes, tienen la ventaja de integrar diferentes procesos, escalas, variables y la posibilidad de generar escenarios con actores en el marco de procesos de modelación participativa.Método: Se diseña una herramienta computacional para la planificación y apoyo a los procesos de toma de decisión relacionados con la gestión de los recursos hídricos, específicamente para el caso de las inundaciones en la ecorregión de la Mojana (Colombia). Son desarrollados talleres de participación social relacionados con creencias, valores, redes sociales y resiliencia.Resultados: El modelo tiene dos componentes: el primero representa la hidrodinámica de las inundaciones por medio de la plataforma numérica ISIS 2D; el segundo, en relación con los aspectos sociales de la región, se maneja a través de modelado de sistemas de agentes. Ambos sistemas están integrados en la plataforma Net...
Considering the evaluation of acquired knowledge as a fundamental element of Computerized Tutorin... more Considering the evaluation of acquired knowledge as a fundamental element of Computerized Tutoring Systems, the classical Bayesian methods for adaptive testing present several deficiencies, such as arbitrary assignments or prior probabilities, non representativity, the effect of the context and the need to include all the concepts under evaluation. An alternative model of the adaptive test is proposed here based on content-balanced and imprecise interpretations of the information level and prior distribution of each item using Gil's fuzzy relationships instead of Bayesian relationships.
Successful usage of a large family of deterministically generated measures to model complex nonli... more Successful usage of a large family of deterministically generated measures to model complex nonlinear phenomena (e.g. rainfall, turbulence and groundwater contaminant transport) has been reported recently.1–7 As these measures, generated as derived distributions of multifractal measures via fractal interpolating functions (FIF), i.e. the fractal-multifractal (FM) approach, have been found to share the inherent character of natural sets, the present study further investigates their dynamical (chaotic or stochastic) properties. It is shown, through a variety of examples and via the use of power spectrum, mass exponents and false nearest neighbors in the state-space, that the FM approach indeed generates deterministic measures whose behavior (depending on their parameters) may be classified as low-dimensional and chaotic or as high-dimensional and stochastic. These results suggest the general suitability of the FM approach for understanding and modeling nonlinear natural phenomena.
... Felipe Prada-Sarmiento, Nelson Obregón-Neira. ... compared to the observed rescaled discharge... more ... Felipe Prada-Sarmiento, Nelson Obregón-Neira. ... compared to the observed rescaled discharges and to the initially trained ANN model results, resorting to some statistical parameters such as , mean absolute error ; root mean square error ; scatter index ; and Nash index: , where ...
Resumen: Los Algoritmos Evolutivos forman parte de la Inteligencia Artificial la cual a su vez se... more Resumen: Los Algoritmos Evolutivos forman parte de la Inteligencia Artificial la cual a su vez se encuentra integrada a un tipo de Sistema Inteligente, la Inteligencia Computacional. En este contexto, este trabajo estudia modelos derivados de los Algoritmos Genéticos (AG) y de la Programación Genética (PG). En particular, se revisan sus potencialidades de aplicación en estudios hidroinformáticos. Específicamente en problemas de optimización matemática global no lineal altamente dimensional, los cuales caracterizan los problemas inversos generados en procesos de calibración de modelos hidrológicos e hidráulicos. Para ilustrar la aplicación hidroinformática de los AG y la PG, se proporcionan ejemplos ilustrativos para el caso de modelos de balance hídrico y de obtención de relaciones funcionales entre variables hídricas de entrada y salida, respectivamente. De esta forma, se discute cómo en los procesos de gestión y particularmente de modelación de sistemas hidroinformáticos, los mode...
RESUMEN: En el diseño y manejo de sistemas de drenaje urbano es necesario conocer distribución es... more RESUMEN: En el diseño y manejo de sistemas de drenaje urbano es necesario conocer distribución espacial de la precipitación. Para obtener esta información, en la ciudad de Bogotá se emplean con frecuencia métodos de interpolación cuya entrada son los registros puntuales de estaciones pluviométricas o pluviográfi cas. En el presente trabajo se presenta una revisión de los parámetros básicos que los modeladores deben asumir antes de realizar una interpolación espacial. Se utilizó información diaria y subhoraria, considerando los métodos de la distancia inversa cuadrática, Kriging y V4 de MATLAB®. Los resultados obtenidos permiten defi nir algunos criterios sencillos en cuanto a la orientación de la malla, el tamaño de la celda a emplear, el análisis estructural previo que debe realizársele a la información y el método de interpolación a utilizar.
Este Articulo muestra los resultados obtenidos en la implementación del modelo neurodifuso ANFIS ... more Este Articulo muestra los resultados obtenidos en la implementación del modelo neurodifuso ANFIS y de las redes neuronales para la predicción de caudales medios mensuales. Se desarrolla e Implementa el modelo ANFIS y se evalúa el comportamiento de 6 modelos al variar el número de entradas, el número y tipo de conjuntos difusos ( funciones de pertenencia), que son los parámetros fundamentales del modelo ANFIS y se compara los resultados con los obtenidos con las redes neuronales perceptron multicapa.
Carbofuran and oxadixyl pesticides are used in Colombia to control pests and fungi, but their mob... more Carbofuran and oxadixyl pesticides are used in Colombia to control pests and fungi, but their mobility through the soil profile is poorly understood. This study showed degradation and adsorption processes of these compounds in a Melanudands soil (0-100 cm) from Colombia using laboratory incubation and the batch equilibrium methods. First-order kinetic models indicated that the degradation rates of carbofuran (0.013-0.006 day-1) and oxadixyl (0.013-0.008 day-1) decreased at deeper soil layers, suggesting that the pesticides were more persistent in the sub-surface (60-100 cm) than in the surface layers (0-40 cm). The thermodynamic approach showed that the adsorption of both pesticides was similar, an exothermic and spontaneous process. The carbofuran and oxadixyl coefficient of distribution (5.8-0.3 L kg-1) and the percentage of adsorption (71.2-11.3%) were very similar in the surface layers (0-40 cm) and decreased with the soil depth. The organic carbon (OC) and clay content showed a...
This article presents the results of the research work of the GIGA research group of the Distrita... more This article presents the results of the research work of the GIGA research group of the Distrital University in Bogota Colombia in the framework of the advanced doctoral research in the engineering doctorate of the Javeriana University and shows the development followed and the corresponding analysis determine the effectiveness of the M5P algorithm within the decision tree approach for the management of BIG DATA and Machine Learning for the determination of massive property valuation. The analysis was carried out with data provided by the Special Administrative Unit of the Distrital Cadaster (UAECD) and the Technical Cadastral Observatory corresponding to three sectors of the city of Bogotá D.C. The results of the J48-based model were compared by means of a statistical analysis with the traditionally used linear regression method, obtaining satisfactory results with errors well below the linear regressions in the training, validation and forecast stages.
ASME-ARPEL 2021 International Pipeline Geotechnical Conference, 2021
Cenit Transporte y Logística de Hidrocarburos (CENIT), operator of about 7000 km of hydrocarbon t... more Cenit Transporte y Logística de Hidrocarburos (CENIT), operator of about 7000 km of hydrocarbon transport systems, which constitutes it the largest operator in Colombia, has developed a strategic alliance to structure an adaptive geotechnical susceptibility zoning using supervised learning algorithms. Through this exercise, has been implemented operational decision inferences with simple linguistic values. The difficulties proposed by the method considers the hydroclimatology of Colombia, which is conditioned by several phenomena of Climate Variability that affect the atmosphere at different scales such as the Oscillation of the Intertropical Convergence Zone - ITCZ (seasonal scale) and the occurrence of macroclimatic phenomena such as El Niño-La Niña Southern Oscillation (ENSO) (interannual scale). Likewise, it considers the geotechnical complexity derived from the different geological formation environments, the extension and geographical dispersion of the infrastructure, and its ...
Uploads
Papers by Nelson Obregon