A hybrid downscaling approach for the estimation of climate change effects on droughts using a geo-information tool. Case study: Thessaly, Central Greece

Abstract

AbstractMultiple linear regression is used to downscale large-scale outputs from CGCM2 (second generation CGCM of Canadian centre for climate monitoring and analysis) and ECHAM5 (developed at the Max Planck Institute for Meteorology), statistically to regional precipitation over the Thessaly region, Greece. Mean monthly precipitation data for the historical period Oct.1960-Sep.2002 derived from 79 rain gauges were spatially interpolated using a geostatistical approach over the region of Thessaly, which was divided into 128 grid cells of 10 km × 10 km. The methodology is based on multiple regression of large scale GCM predictant variables with observed precipitation and the application of a stochastic time series model for precipitation residuals simulation (white noise). The methodology was developed for historical period (Oct.1960–Sep.1990) and validated against observed monthly precipitation for period (Oct.1990–Sep.2002). The downscaled proposed methodology was used to calculate the standardized precipitation index (SPI) at various timescales (3-month, 6-month, 9-month, 12-month, 24-month) in order to estimate climate change effects on droughts. Various evaluation statistics were calculated in order to validate the process and the results showed that the method is efficient in SPI reproduction but the level of uncertainty is quite high due to its stochastic component.