Iran Meteorology Society (IMS) Journal of Meteorology and Atmospheric Science 2645-7261 1 1 2018 05 22 Performance comparison of GP, ANN, BCSD and SVM models for temperature simulation Comparison performance of GP, ANN, BCSD and SVM models in temperature simulation Performance comparison of GP, ANN, BCSD and SVM models for temperature simulation Comparison performance of GP, ANN, BCSD and SVM models in temperature simulation 53 64 83636 FA S.M. Taabtabaei Department of Water Science and Engineering, Faculty of Agriculture, University of Birjand M. Nazeri Tahroudi Department of Water Science and Engineering, Faculty of Agriculture, University of Birjand M. Dastourani Department of Water Science and Engineering, Faculty of Agriculture, University of Birjand Journal Article 2018 03 09 In this study, GP, ANN, BCSD and SVM were used to evaluate the performance of simulation models. The modeling based on the large scale data of atmospheric general circulation and the mean daily temperature of the Ahwaz synoptic station was conducted during the period of 1971-2004, and the evaluation of each model was based on the correlation coefficient and modeling error between simulated and observational data. The results of modeling indicated that the correlation coefficient between observation and simulation data in the SVM model is more than in other models, and its value is 0.9960. The correlation coefficient for the GP, ANN and BCSD models is equal to 0.9393, 0.9384, and 0.4936 respectively. Moreover, the results of the evaluation of the simulation error were calculated using the RMSE and NSE criteria for SVM to be 0.677 and 0.955 degrees Celsius respectively. Similarly, these values are 1.644 and 0.969 for GP, 1.657 and 0.968 for ANN, and 6.174 and 0.661 for BCSD. Therefore, SVM has better performance in modeling the mean daily temperature than other methods, and modeling the mean daily temperature by the BCSD method is less accurate than other methods. The GP method has a weak advantage over ANN, and it is recommended to use the minimum and maximum temperatures for the more precise performance of these two models. In this study, GP, ANN, BCSD and SVM were used to evaluate the performance of simulation models. The modeling based on the large scale data of atmospheric general circulation and the mean daily temperature of the Ahwaz synoptic station was conducted during the period of 1971-2004, and the evaluation of each model was based on the correlation coefficient and modeling error between simulated and observational data. The results of modeling indicated that the correlation coefficient between observation and simulation data in the SVM model is more than in other models, and its value is 0.9960. The correlation coefficient for the GP, ANN and BCSD models is equal to 0.9393, 0.9384, and 0.4936 respectively. Moreover, the results of the evaluation of the simulation error were calculated using the RMSE and NSE criteria for SVM to be 0.677 and 0.955 degrees Celsius respectively. Similarly, these values are 1.644 and 0.969 for GP, 1.657 and 0.968 for ANN, and 6.174 and 0.661 for BCSD. Therefore, SVM has better performance in modeling the mean daily temperature than other methods, and modeling the mean daily temperature by the BCSD method is less accurate than other methods. The GP method has a weak advantage over ANN, and it is recommended to use the minimum and maximum temperatures for the more precise performance of these two models. https://www.ims-jmas.net/article_83636_0d88c6193f1a828ef949c4d7fd6f5431.pdf