Performance Evaluation of Statistical Downscaling Model (SDSM) in Forecasting Temperature Indexes in Two Arid and Hyper Arid Regions

Statistical downscaling methods are widely used for prediction of climatic variables e.g. temperature because of importance of these factors in environmental planning and management. In this study, the performance of Statistical Downscaling Model (SDSM) was investigated to predict temperature. The input data of the study include minimum, maximum and mean temperature of Kerman and Bam Synoptic stations, NCEP (National Centers for Environmental Prediction) data and the A2 and B2 emission scenarios HadCM3 for the reference period, 1971-2001. The first 15 years data (1971-1985) was applied for the calibration and the second 15 years data (1986-2001) for model validation. Temperature for three periods including 2010-2039, 2040-2069 and 2070-2099 was predicted and then compared with the temperature data of reference period i.e.1971-2001 using HadCM A, B2 data. Statistical measures of model performance such as Mean Absolute Error (MAE), Root Mean Squared Error (RMSE), Nash-Sutcliffe efficiency (NS) and the analysis of output results from SDSM model shown that this model is able to predict temperature indexes more accurately in arid climate than in hyper-arid climate. The results indicate temperatures rising in all months for both stations.