Assessment of GCM performances for the Arabian Peninsula, Brazil, and Ukraine and indications of regional climate change

Abstract

This study presents an assessment of global climate model (GCM) performances for the regions Arabian Peninsula, Brazil and Ukraine, all facing problems related to water availability as well as quality and so are the focus regions for investigations within the Integrated Water Resources Management Project ‘International Water Research Alliance Saxony’. Results from the model performance assessment have been applied to derive estimates of the future climate in the regions by means of ensemble averaging. The study uses GCM data from the third Climate Model Intercomparison Project. Model performances were assessed by the Model Performance Index (MPI) as well as Taylor diagrams applying different reference data sets like reanalysis data as well as observational data sets. The model performances were analysed for different domain sizes. Climate variables of different tropospheric levels were classified by applying projections of the overall model ensemble as well as projections of a subset of models (selected according to the MPI) in variables with different levels of certainty of change. The MPI as well as the reliability ensemble averaging method were employed to calculate ensemble averages for variables with high certainties of change. Results are presented for mean values within the time periods 2040–2060 an 2080–2100. Applying different data sets as reference data shows that model performance crucially depends on the reference data set. The size of the model domain is of minor importance. Model rankings vary between regions. Certain changes are found for temperatures and geopotential heights in all levels but, e.g., not for precipitation.