Assessment of CMIP5 climate models with observed precipitation in Iran

The changes in precipitation that depend on future climatic changes highly affect environmental processes and the use of ecosystem services, especially water sources. Because providing necessary material for human beings is mostly dependent on water sources, the reliable prediction of precipitation and water sources, affected by climate change, is of considerable importance. Nowadays, there are centers and various models worldwide that simulate the state of future climate of the earth by different scenarios, e.g., the scenario of release physical and computational structure. Simulations of world climate models have been archived by CMIP project, which are regarded as one of the most important sources to study the climate condition of the 21st century. The simulations from models of general atmospheric circulation, which is a part of CMIP5, are as the basis for the conclusions of international committee related to future climatic changes. The data can be used to assess future climatic changes in local or regional scales, whether directly or after presenting downscale. Although the predictions of general circulation models are reliable enough, ignoring some important features of each region, especially developing countries and the ones with high environmental diversity like Iran, make the data of these models need accurate assessment in various spatial and temporal scales.
Therefore, the present study aims to assess the accuracy of precipitation data from eight models (BCC-CSM1.1, CCSM4, CESM1-BGC, CESM1-CAM5, CMCC-CM, EC-EARTH, MIROC5and MIR- CGCM3) of general atmosphere circulation according to high spatial accuracy for Iran applying statistical tests. Statistic indices like R, R^2, RMSE, BIAS, EF, NARMSE, SLOPE, and IA were applied to choose the most appropriate model, out of eight, up close to real data of the country.
The findings reveal that Although the models used to calculate rainfall has not high reliability, but also it is too weak to estimate the stations’ extreme events. Besides, in a similar study, Hidalgo and Alfaro (2014) believe that most of the CMIP5 models have low ability to estimate the precipitation of central regions of the USA. Regionally, output accuracy for north-eastern and western regions is more than other parts of the country. Besides, the accuracy for coastal regions of Iran (Oman and Caspian) is very low and practically useless, which is due to the special geographical condition and the contrast of land and water in these regions. In fact, the assessment of future precipitation output of these models under scenario 4.5 and 8.5 presents the same findings; the correctness of predictions in northern half (except for Caspian beaches) is more than the southern half. Scenario 4.5 shows better results in northeast and west while scenario 8.5 shows better results in southern beaches, especially southeast of the country. The findings from the process of precipitation from CCSM4 model, under scenarios 4.5 and 8.5, show that the process of future precipitation changes will not be significant for any region and the slope is from weak to average.