Future scenarios of air temperature maximums and minimums for Georgia based on statistical downscaling

Abstract

In this article monthly maximums and minimums of 2-meter air temperature from three GCMs of CMIP5 [1] database has been statistically downscaled using RCMES [2] package, with four different methods for 27 selected meteorological stations on the territory of Georgia. Stations have been selected all over the territory of Georgia, based on the completeness of their air temperature series throughout the entire period of 1961−2010, their credibility (measured by the number of non-missing data) and to cover as much complex climate features of the territory as possible. The downscaling methods have been trained for the period of 1961-1985 and validated for the period of 1986-2010. Some statistical parameters have been calculated by applying R statistics environment to compare observed and simulated time series and to evaluate temporal and spatial goodness of each method. Downscaling model, driven by the validation study was used for future Tmin and Tmax time series construction for the 2021-2070 period under RCP4.5 and RCP8.5 scenarios. Temperatures time series have been constructed from a multimodel ensemble, with mean and spread. Future change tendencies have been assessed in comparison of the period of 1986–2010 but was also compared with previous 25-years period (1961-1985) to compare future changes with the magnitudes of past tendencies.