A stochastic weather generator model for hydroclimatic prevision in urban floods risk assessment in Abidjan District (Cote d’Ivoire).

Abstract

Flood risk occurrence is very often related to heavy precipitation; and available future weather data is a potential source for long term flood risk prediction. The aim of this paper was to determine and analyze trends in rainfall, temperature and PET under present and future climatic conditions using Long Ashton Research Science-Weather Generator (LARS-WG) software, in prediction of flood risk occurrence in Abidjan. This work was based on the integration of Hydro climatic daily data within LARS-WG software. The processing steps are: (1) calibrating and validating the model using 50 years measured data, (2) generating baseline data for 50 years, (3) processing future scenario data based on baseline already set using HADCM3 and (4) Comparing baseline and generated scenario data. The resulting statistics show that temperature will increase by 0.32, 1.36 and 2.54 °C for the periods 2011–2030, 2046–2065 and 2080–2099 respectively. Then rainfall in the same period will increase by 4 %, 6 % and 10 % respectively. The mean and high flooding risk will then increase in long term within this urban area. Thus this future large extension of flooding occurrence imposes to take future weather scenario into account in prediction and management of flooding risk in Abidjan District.