Estimation of the Monthly Mean Temperature Values of the Eastern Black Sea Basin with Statistical Downscaling Method Using EraInterim Re-analysis Data

Abstract

Statistical downscaling methods are based on determination of statistical relationships between low resolution atmospheric variables and measured climate parameters from meteorological stations. In this study, it was aimed to estimate the monthly mean temperature measured from 12 meteorological stations in and around the Eastern Black Sea Basin using atmospheric variables in the EraInterim re-analysis data set with grid resolution (0.75° x 0.75°). For this purpose, the variables of precipitation, temperature, sea surface pressure, surface air pressure and air temperature, geopotential height and relative humidity at 850, 500 and 200 hPa pressure levels in the EraInterim re-analysis data set were used as independent variables. Monthly mean temperature values measured from meteorological stations (1981-2010) were used as dependent variables. Multivariate adaptive regression splines (MARS) method selected as downscaling method. The root mean square error, scattering index, mean absolute error and Nash Sutcliffe (NS) efficiency coefficient statistics were used to evaluate the performance of the MARS model based on the station. The NS value calculated for all stations was in the range of 0.9-1.0. In addition, the global scale variables selected from the EraInterim data set were found to be quite successful in estimating local temperature values. The results obtained from the study showed that MARS statistical downscaling method can be used to downscale the coarse scale atmospheric variables to the regional scale.

Keywords

• erainterim
• mars
• eastern black sea basin
• climate change
• statistical downscaling
• temperature