Prediction of Climate Change in Western of Iran using Downscaling of HadCM3 Model under Different Scenarios

Abstract    Considering that water resources are at risk from climate change, the study of temperature and precipitation changes in the coming years can lead to droughts such as droughts, sudden floods, high evaporation and environmental degradation. To this end, global climate models (GCMs) are designed to assess climate change. The outputs of these models have low spatial accuracy. In order to increase the spatial accuracy of this data, downscaling methods are used which are divided into statistical and dynamic methods. One of the reasons for using these models is their quick and easy operation compared to other methods. Our study area consists of Kurdistan, Kermanshah and Hamedan provinces in the west of the country. In this study, observational data of minimum temperature, maximum temperature, precipitation and radiation of 6 synoptic stations in the studied area in the statistical period of 1961 until 2005. In this study, the LARS-WG model was used for downscaling of HadCM3 global model data. The LARS-WG model is one of the most popular weather generator models that which to generation for maximum and minimum temperature, rainfall and radiation are used daily under current and future climate conditions. This model as a downscaled version of the same process less complex and simulated data input and output, high ability to predict climate change. The HadCM3 model is also a type of atmospheric- oceanic circulation model developed at the Hadley Center for Climate Prediction and Research, which has a 2.5 degree latitude network at 3.75 degrees longitude. Also, three climate change scenarios A1B, A2 and B1 have been used, each of which reflects the characteristics of the world's economic growth, the world's population and social awareness. The methodology is that the model receives the monitored data of the basic course; by examining them the statistical characteristics of the data are extracted. Then, in order to validate and ensure the model's capability for the basic statistical period, the model is implemented to re-establish a series of artificial data in the base period. Then the outputs to evaluate the performance of the model in the reconstruction of the data, the statistical characteristics of observations to test and compare various criteria. MAE, MSE, RMSE and R2 criteria were used to evaluate and analyze the performance of the downscaling model.