Climate change has been recognized as having a profound impact on the hydrologic cycle at different temporal and spatial scales and Global Climate Models (GCMs) have been commonly used in various studies for assessing its potential impacts. However, outputs from these models are considered too coarse (generally greater than 200km) and hence are not suitable for climate change impact studies at a given site or over a catchment area. Consequently, several downscaling techniques have been proposed to downscale these GCM outputs to the precipitation series at a given location of interest. Nevertheless, there is still no general agreement about which downscaling method is the best approach for describing accurately the observed precipitation characteristics at a given site in the climate change context, depending mainly on the specific study objectives and the specific climatology of the particular study area. The present study proposes therefore a new statistical model, herein referred to as SDGAM, using the Generalized Additive Modeling methods in order to address the shortcomings of existing downscaling methods. The feasibility and accuracy of the proposed new approach were evaluated using the National Center for Environmental Prediction (NCEP) re-analysis data and the observed daily precipitation records available for the 1961–2000 period from a network of ten raingages located in Southern Quebec-Ontario region in Canada. Results of this numerical application have indicated that the proposed SDGAM model was able to describe well many features of the daily precipitation process, including its occurrence frequency, intensity, and extremes. In addition, it has been demonstrated that the suggested SDGAM model could provide more accurate results than those given by the popular Statistical Downscaling Model (SDSM) in the modeling of the daily precipitation process based on both numerical and graphical performance criteria.
