Multiple Linear Regression Based Statistical Downscaling of Daily Precipitation in a Canal Command

Abstract

The climate impact studies, particularly in hydrology, often require climate information at fine scale for present as well as future scenario. Global Climate Model (GCM) estimates climate change scenarios on coarse spatial resolution. Therefore, different techniques have been evolved to downscale the coarsegrid scale GCM data to finer scale surface variables of interest. In the present study, the Statistical Downscaling Model (SDSM) has been applied to downscale daily precipitation from simulated GCM data. SDSM utilizes Multiple Linear Regression (MLR) technique. The daily precipitation data (1961–2001) representing Tawa region has been considered as input (predictand) to the model. The model has been calibrated (1961–1991) and validated (1992–2001) with screened large-scale predictors of (National Centre for Environmental Prediction (NCEP) reanalysis data. The prediction of future daily rainfall for the study area has been carried out for the period 2020s, 2050s and 2080s corresponding to HadCM3 A2 variables. The calibration and validation results confirm the SDSM model acceptability slightly at a lower degree. The results of the downscaled daily precipitation for the future period indicate an increasing trend in the mean daily precipitation.