Comparison of methods of statistical downscaling of climate change models to simulate climate elements in Northwest Iran

Linking resolution global climate models with local scale is a micro climatic process that itself is a significant issue. Recently, attempts have been made by the climatology community to develop dynamics and statistical downscaling methods for expressing climate change has taken place at a local and regional scale. Two general techniques are used for downscaling of the output of general circulation models (GCM). The prior is using of statistical methods in which the output of a statistical model (MOS) and a planned approach to weather short-term numerical prediction is presented. The second is regional climate model (RCM), that same is limited GCM model in a subnet of the network global model and by dynamic method uses climatic conditions temporal changes according to GCM model. Both methods Play an important role in Determine the potential effects of climate change caused by increased greenhouse gas emissions. Much work is done to use this method for downscaling of the global model output in different areas In which the performance of the model is assessed and uncertainty analysis has been done on these methods or were compared by other statistical methods.
In this study, three approaches to statistical downscaling methods are provided. The first approach uses random generation of climate models based on time series and Fourier series delivers. LARS-WG statistical model(Rskv et al., 1991, 27) is one of the ways is built on the basis of this approach,. In this model, the empirical distribution of daily series of dry and wet precipitation and solar radiation use is desirable. The minimum and maximum daily temperatures as the daily stochastic process with mean and standard deviations are taken daily. Seasonal cycles by means of finite Fourier series are of the order of 3 models and model residuals (model errors) is approximated by a normal distribution.
The second approach is regression model or transfer function that is more used, which uses the relationship between atmospheric parameters and synoptic (predictor variables) and climatology Parameter that it is necessary to have a vision of the future(Instant predictor variable) is a transfer function is provided. One of the applications that combines these two approaches based on statistical downscaling model (SDSM) is. The meteorological station data required as input and output in seven steps GCM model on the basis of daily data in the area are downscaled.
The third downscaling model is artificial neural network (ANN), developed by Coulibaly et al., 2005. This model is a non-linear regression type in which a relationship is developed between a few selected large-scale atmospheric predictors and basin scale meteorological predictands. In developing that relationship a time lagged recurrent network is used in which inputs are supplied through tap delay line and the network is trained using a variation of backpropagation algorithm (Principle et al., 2000). A slightly different approach is used in selecting predictors for the case of neural network downscaling.
To compare data generated models and observations can be compared to an average of two non-parametric test Mann-Whitney society that is using correlation analysis. For the observed data and the model can be generated from correlation Spearman used. The basic correlation analysis based on linear correlation coefficient of the two variables. One of the important indicators that can be used for performance evaluation model, index model mean square error (RMSE) is defined as follows:The area North West of Iran, which includes the provinces of East and West Azerbaijan, Ardebil, Zanjan and a part of Kurdistan is the geographical coordinates '30 ˚49 '07 ˚44 to the East and the '00 ˚36 to '50 ˚39 North, is located. To study the effects of climate change in the region, using statistical models mentioned the need for a minimum period of 1961-1990 is based. In addition to the complete statistical period synoptic meteorological stations of old climate data confirmed the country's Meteorological Agency has been helping though some regional stations are multi-year statistical vacuum.
The results show that according to the Mann-Whitney test the performance of three models for minimum temperature in the study area are close. Spearman correlation test results for minimum temperature show that the number of correlation, in all stations for LARS-WG model is less than the other two and demonstrate low performance LARS-WG model is in this respect. The average number of months with significant correlation for ANN model with seven months of the year, the best performance among the three models in this respect. SDSM model with a four-month correlation table in the middle. In terms of RMSE index for the minimum temperature, LARS-WG and ANN models have average values are close together and show the error of sum of squares closer together the two models. RMSE values are less than the SDSM model and this shows the SDSM model is less than the other two models.
According to our evaluation, according to Mann-Whitney test data generated in which the difference between the observed and tested model placed, Parameters for minimum and maximum temperatures, three models have not different performance. But the results were somewhat different in different stations. Correlation data for SDSM and ANN models for maximum high temperature and minimum temperature for solidarity in SDSM model is less than ANN model. However, because the same structure prediction methods and large-scale use of such an outcome was not unexpected.
Mann-Whitney test for precipitation results show that significant differences observed and modeled data for ANN model is much more than the other two, which reflects the low performance of this model. SDSM and LARS-WG model and have similar good performance in this regard. The Spearman correlation test, all three models have a low correlation was observed and the model and represents the three models in the study area in this respect is low. According to the RMSE, the SDSM model is better than the other two models have average performance.