Prediction of Wind Speed Meteorological Variable in Zayandehrud Basin Using Time Series Analysis

Considering the effect of wind speed on important climatic issues such as evaporation, wind speed was investigated in seven meteorological stations of Zayandehrud basin using time series method.  In this study the periods includes 7 days, 15 days, 30 days, seasonal and daily. Outlier data were determined by box diagram, normal and Grubs beck methods. The modeling was performed by examining the autocorrelation and partial autocorrelation diagrams using the criteria of Akaike, Schwartz and Hannan Quinn. Kolmogorov-Smirnov and Jark Bera tests that were used to check the normality of data and Durbin Watson and Pert Manto tests were used to check the accuracy of the model. Trend and homogeneity tests were performed using Matlab and modeling was performed using Minitab, Eviews. For validation at daily and 7-day intervals, 5% of data were considered, at 15 days and 30 days, 10% of data and at seasonal intervals, 20% of data. The trend analysis was performed by non-parametric Man Kendall method and the trend was observed in the majority of intervals. also, based on the results obtained (the coefficient of determination above 0.7 and the mean square error and error percentage below 20%) the model of SARIMA was introduced as the best model in the 7-day, 15-day and 30-day time intervals, and ARIMA model in daily and seasonal  intervals