Hydrological modelling using artificial intelligence techniques for the characterisation and forecasting of maximum extreme events
2022
Hamitouche, Mohamed
This document addresses the high-flow hydrological extremes. The overarching objective is to develop a methodology within Artificial Intelligence (AI) for their prediction, to estimate their associated uncertainty and to quantify the impact of their causal factors. For that, this document firstly provides a comprehensive review of the state-of-the-art AI methods. Secondly, it shows the development of a method called “HydroPredicT_Extreme”. Finally, it presents a case study to demonstrate its forecasting capabilities. As a result, AI methods such as Bayesian methods show a high predictive capability. The “HydroPredicT_Extreme” method, based on Bayesian Causal Modelling, and applied to the semi-arid Upper Andarax watershed, has demonstrated a satisfactory predictive capacity for both low flows and extremal behaviour, and showed that the basin response is more dependent on the rainfall intensity than on antecedent moisture conditions. This methodology may be coupled to Early Warning and Forecasting Systems for the reduction of flooding risk.
Mostrar más [+] Menos [-]Palabras clave de AGROVOC
Información bibliográfica
Este registro bibliográfico ha sido proporcionado por Instituto Agronómico Mediterráneo de Zaragoza