This work describes the results obtained by the statistical downscaling technique for the assessment of changes in precipitation (P), potential evaporation (PE). In turn P and PE are used for computing two indexes of water availability, namely the index of water deficit (WDI) and the aridity index (AI). The analysis is carried out for the Capitanata plain (South-East of Italy) and the A2 scenario of the IPCC Assessment Report 4 (AR4). The large-scale temperature at the 1000hPa level and sea level pressure fields are used as predictors. The local precipitation and potential evaporation time series are used as predictands. The statistical downscaling technique used is based on Canonical Correlation Analysis. A validation procedure of the model is performed and the same technique is used for climatic projections of P, PE and consequently WDI and AI. Climate analysis and projections at this localspace scale is an important issue not only for current water management and planning, but also for improving the irrigation efficiency considering future climate change scenarios.