Improvement of multiple linear regression method for statistical downscaling of monthly precipitation

Abstract

This article aims at proposing an improved statistical model for statistical downscaling of monthly precipitation using multiple linear regression (MLR). The proposed model, namely Monthly Statistical DownScaling Model (MSDSM), has been developed based on the general structure of Statistical DownScaling Model (SDSM). In order to improve the performance of the model, some statistical modifications have been incorporated including bias correction using variance correction factor (VCF) to improve the computed variance pattern. We illustrate the effectiveness of the proposed model through its application to 288 rain gauge stations scattered in different climatic zones of Iran. Comparison between the results of SDSM and the proposed MSDSM has indicated superiority of the proposed model in reproducing long-term mean and variance of monthly precipitation. We found that the weakness of MLR method in estimating variance has been considerably improved by applying VCF. We showed that the proposed model provides a promising alternative for statistical downscaling of precipitation at monthly time scale. An investigation of the effects of climate change in different climatic zones of Iran by the use of Representative Concentration Pathways (RCPs) has shown that the most significant change is an increase in precipitation in fall and that the largest share of this increase belongs to arid climate.