Résumé
In arid regions where precipitation is weak and intermittent, the aquifer recharge is an important issueto evaluate groundwater resource and its future evolution in a context of excessive water requestincrease and climate change. A Multivariate statistical analysis of rainfall variables and atmosphericconditions factors was applied in an attempt to define the climate variability in Tataouine basin (107mm of mean annual rainfall from 1987 to 2014), and assess its impact on rainfall regimes and therecharge of the Continental Intercalaire aquifer (CI). The present work is based on a dataset of 11variables over the period from 1984 to 2003 (i.e. rainy days frequencies, P1 (1-10 mm), P2 (10 to 30mm), P3 (30 to 50 mm), P4 (> 50 mm) and P0 (number of no rainy day between 2 events), meanmonthly rainfall (Pm), mean monthly temperature (T), mean maximum temperature (T1), meanminimum temperature (T2), thermal amplitude (TA= T1-T2) and the mean monthly relative humidity(H)). The linear regression predictive model applied to assess the long-term trends of rainfall variationshowed that the annual variation of rainfall is related to the changes of Pm, P3, P4, T1, AT, and Hrather than P0, P1, P2 and T2. It appeared that annual rainfall variation is related to rainfall fr equencyand depends on the amount of rainfall events. Also the air temperature and the relative humidity of theair are factors of the temporal variability of the rainfall regimes. The analysis of rainfall variability basedon ‘Nicholson’ over the time series extending from 1987-2014 has allowed to highlight the presence ofan excess period from 1987 to 1999 and a deficit period from 1999 to 2014. It was shown that therecharge process is not correlated with soil humidity state but with higher rainfalls involving surfaceflow concentration through wadis, the main recharge mode, spatially limited in this context.