Validation of the precipitation estimated by CMIP5 climate models in Colombia

Abstract

The prediction of possible changes in precipitation in the coming years and decades due to global climate change associated with the greenhouse effect, deforestation and other anthropic perturbations is a highly complex practical and scientific problem of huge consequences. The traditional strategy has focused on the use of coupled atmosphere-ocean general circulation models, and while such models have demonstrated important improvements for representing past and present climate, they are not without limitations. In particular, the estimation of precipitation in a tropical and mountainous country like Colombia has proved to be very difficult. In order to check the credibility of such predictions, we evaluated the historical precipitation in some of the models of the fifth phase of the Coupled Models Intercomparison Project (CMIP5) through a systematic comparison with observed precipitation as registered in the database of the national hydrologic model proposed by Hurtado (2009), mostly based on rain gauge observations. We evaluated different error measures: the mean absolute error, the mean squared error, the concordance index, the systematic error (bias), the correlation coefficient and the coefficient of determination. The results of the seasonal and regional assessments question the credibility of precipitation prediction through general circulation models (GCM), and give an insight on the use of derived data for relevant decision-making, which emphasizes the need for a national research program to answer to the important question of the impact of climate change on Colombian precipitation patterns.