Statistical analysis on the performance of global and regional climate models for the Paute river basin in the south-ecuadorian andes

Abstract

Climate change impact in the Andes regions is expected to have a large influence on water resources as in many other regions of the world. However, a major problem for climate change impact studies on the region is the high spatial variability of rainfall and temperature, which is not well represented in climate models. Different Global Climate Model (GCM) simulations available from the IPCC AR4 as well as simulations from the Regional Climate Model (RCM) PRECIS have been analyzed for the Paute river basin in the South-Ecuadorian Andes. This river basin has a wide range of topographical elevations and covers different hydrological regimes, represented in this study by six weather stations. The control simulation results from the climate models were statistically checked for inconsistencies by quantifying the differences between control period and historical time series data of rainfall and temperature. This was done for annual, monthly and daily values regarding the return period. In addition statistics criteria of relative mean squared error, bias and correlation were also computed. Results show that no model performed well for all the criteria, but some models were generally better than others. The set of best performing models, however, differed from station to station. Surprisingly, several GCMs showed better results than the RCMs for rainfall. The results reveal that a strong increase in the climate model spatial resolution does not necessarily result in more accurate description of local climate properties. Therefore, statistical downscaling techniques are crucial in climate change impact studies for these regions.