Identification of adequate Sahel climate prediction models based on CMIP5: Mali case

Abstract

In this work, a multi-model of evaluation of future normals of temperature and rainfall for Mali was built. For this, the normals of temperature and precipitations of twenty (20) climate models out of the forty-two (42) in the CMIP 5 were compared individually with those of seventeen (17) stations for the temperature and thirty (30) stations for precipitations. The objective function used is the standard deviation, which is even smaller than the model is more efficient. The standard deviation of normals for all models with observations exceeds 10% of the average of precipitations over the territory, however, for the temperature 85.0% of the models (17 out of 20) give standard deviations representing less than 10% of the territorial average of the latter. This which allowed classifying the models, their intrinsic performance for each element of the considered climate. The analysis has shown that most models occupy very different range from one classification to another. Thus, a general classification was made which has shown that the most efficient multi-model for temperature and rainfall are different but the one giving the best approximation for the whole of the two elements (with a minimum of standard deviation) is composed of the models of the Institute Max-Planck of the German meteorology, this is MPI-ESM-LR and MPI-ESM-MR. The performance of this multimodel, compared to that of each of the two models composing it, second rank for temperature and first for precipitations..