Monday, 15 January 2007
Assessment of global climate models performance from the characterization of duration, intensity and frequency of daily rainfall. Application to African Sahel
Exhibit Hall C (Henry B. Gonzalez Convention Center)
The sahelian climate in western Africa is characterized by a strong variability and extremes occurrence in rainfall regime at inter-annual to decadal time scales. As illustrated in recent decades, this semi-arid region undergoes periodically severe droughts, with its fate of famine and of loss of life, and all related environmental calamities such as locust invasions that revealed its strong vulnerability to climate variability and its potential change. In this context, as suggested in all recent IPCC reports, accurate regional climate analysis and climate scenarios development are needed for the vulnerability, impact and adaptation studies related to rainfall variability and its potential change including extremes. Therefore, measures aimed at managing these risks are of the utmost importance. In this context, a collaborative and multidisciplinary project has been initiated to strengthen adaptive capacity to climate changes in sahelian countries that participate in the CILSS organization (Comité permanent Inter-États de Lutte contre la Sécheresse au Sahel), as well as to help to manage the risk faced to climate change and promote the concerted action at the community level by integrating local knowledge and practices with respect to current climate variability. It involves the AGRHYMET research centre (AGRiculture HYdrology METeorology located in Niamey, Niger), Environment Canada and the University of Quebec at Montreal (Canada), with financial support from the Canadian International Development Agency. The climate science work of this project consists first on the development of climate indices-based extremes and their using to characterize both the recent climate variability and changes, as well to evaluate the performance of global climate models (GCMs) to reproduce the current climate before to construct the climate scenarios information over key sahelian areas. The climate indices developed from daily values have been chosen for their relevance to link the vulnerability and risk assessment in agriculture and food security sectors, and to underpin the relationship between climate variability and socio-economic criteria of vulnerability. Results on the validation of three different GCMs (CGCM2, ECHAM4, and HadCM3) as well as two sets of reanalyses products (NCEP/NCAR and ECWMF/ERA40) with comparison between simulated and observed climatology (over the period 1961-1990) is discussed in this paper, in using duration, intensity and frequency of daily rainfall indices, i.e. characterizing wet days, consecutive dry days, daily intensity index, and three extreme indices of precipitation. Statistical criteria are considered as mean bias error (MBE), mean absolute error (MAE), root mean square error (RMSE) and normalized difference (ND) to measure the models/reanalyses performance related to observations. The observed trends of all climate indices over different regions of Sahel is also presented to better document the characteristics of monsoon regime in western Africa at intra- and interannual scales, in particular the recent variability in extremes of precipitation. In general, the results show that the simulated climate indices are quite different between the three GCMs and the two sets of reanalysis, with a systematic overestimation of wet days and intensity of precipitation during the dry period and in the first part of the wet season compared to observed values. At spatial scale, in spite of the fact that the mean feature of precipitation is relatively well represented, all GCMs tend to underestimate the precipitation occurrence and intensity over southern areas, and overestimate in western and in sub-saharian regions.