IWRA Proceedings

AbstractNowadays, climate changes are widely considered a reality by scientists (IPCC, 2001). Since the preindustrial period, observations have shown a significant evolution of mean annual pluviometry (higher or lower depending on the geographical area). Nevertheless, scientists cannot easily evaluate its impact on the extreme hydrometeorologic phenomena using classical statistical techniques. This problem comes from a lack of distancing in the observed rain event time series to evaluate the tendency of extreme phenomena. Besides, global climatic models show many difficulties in modelling extreme meteorological processes. Only studying the flow is not sufficient because of changes in basin, independently of climate change (urbanisation, fire, deforestation, hydraulic installations…). An original approach to study climate change is proposed by modelling flood generator processes using an hourly rainfall generation model adapted to different climates [1] coupled with rainfall-runoff modelisation [2]. The main objective of the research is to evaluate, by rainfall-runoff modelisations, which impacts climate changes might have on floods. For example, evolution of climatic boundaries, which are important in the translation in hydrologic risks, will be studied. To show the detection of climate changes, parameters of the rainfall stochastic generator are evaluated from many ranges of time and calculated by different statistical tests. In contrast to classical methods, the model parameters are calculated by mean climatic characteristics and not by extreme values. So the three parameters: occurrences of events, duration of events, and the intensity of rainfall, which are particularly well adapted to examine rain signals, are not sensitive to the sample. By the variability of the characteristics of rain detected by chronicles of observed rainfalls or anticipated by atmospheric models and by rainfall-flow modelisation, impacts on hydrological risk are studied. Moreover, tests on future scenarios are possible, to evaluate the impact of climate changes and also of landscape use. Our previous research has allowed us to product climatic typology of rainfall by using model parameter settings. The rainfall and rainfall-runoff models have already been regionalised, with actual climate hypothesis, on the part of French territory, but also on the Italian piemont , in a Spanish region, in the Caribbean and on Reunion Island. On these territories, the cartographic evolution of hydrologic risk can be evaluated with climatic change hypothesis. References: [1] Arnaud P., Fine J.A., Lavabre J., 2007. An hourly rainfall generation model adapted to all types of climate. Accepted in Atmospheric Research. [2] Arnaud P., Lavabre J., 2002. Coupled rainfall model and discharge model for flood frequency estimation. Water Resources Research, 38(6) 1075-1085.