COMPARISION OF LARS-WG AND SDSM DOWNSCALING MODELS IN KERMANSHAH (IRAN)Ahmad Rajabi, Hosein Sedghi, Saeid Eslamian and Habib Musavi
The main limitation of using the results of global circulation models (GCM) is the low spatial accuracy of these models. To use the results of these models one should downscale their results in different areas. Two downscaling models, Lars-WG and SDSM, were used to downscale the data of daily precipitation and minimum and maximum daily temperature in Kermanshah; and their downscaled results were compared. Due to the fact that the daily precipitation data are not normal and that the data of daily minimum and maximum temperature have autocorrelation, nonparametric tests were used. The investigation of errors in estimating the means was carried out by Wilcoxon rank test and investigation of errors in estimating the variances was carried out by Levenes test. The investigation of uncertainty in estimating the means and variances of observed and downscaled data is performed by using bootstrap nonparametric test. The results of the used tests indicate the superiority of Lars-WG model than SDSM in downscaling the data of daily precipitation and maximum and minimum daily temperature in Kermanshah (Iran).
Enter your contact information below to receive full paper.