An Assessment of Uncertainty of Regional Climate Change Models, Error Correction Methods and Forecasting Climate Change in Birjand Township

One of the most important consequences increased radiative forcing due to anthropogenic activities is global warming, that created varied and challenging changes such as climate change on ecosystems. Predictions show that the increased radiative forcing will continue. Regarding to the existence of several climate models and error correction methods, selection of the right model is one of the key challenges. In this study, the accuracy and efficiency of the fifth report on regional climate models including CanESM2, CSIRO Mk3, EC- EARTH, IPSL, MIROC51, HadGEM2, MPI, NorESM1 and GFDL and statistical downscaling error correction methods including Linear Scaling (LS), Change Factor (CF), Distribution Mapping (DM),  Local Intensity Scaling(LI), Power Transformation (PT) and Variance Scaling (VS) Using T and F tests, the Taylor diagram and 10 statistical indices during two control (1956-2005) and validation (2006-2015) periods were assessed in Birjand station. The results show that; comparing the average of monthly period increases the efficiency of models and methods exaggeratedly. The best model is different based on statistical indicators and time series period. CF downscaling method is not accurate in validation period. LS and VS downscaling methods are appropriate selection for precipitation and temperature parameters respectively. Also, average accuracy of all models for both precipitation and temperature parameters is better than a single model. MPI and Earth climate models have good performance in simulating precipitation and temperature data.