Statistic was done for the precipitation and evaporation monitoring data of Yuncheng from 1971 to 2007. The first-order seasonal autoregressive models were set up, considering separately for the normal and skewed distribution of rainfall and evaporation. Long series of monthly precipitation and evaporation sequences were generated. Compared the average, standard deviation and other parameters of simulation sequences with measured sequence, the results of skewed simulation had better agreement with the measured one. Then it proved that skewness model maintains the main statistical characteristics of the measured sequence. 20 groups of equal length with the measured sequence of precipitation and evaporation sequence could be generated through the skewness model. The sequence of irrigation water of the winter wheat multiple corns planting type had been calculated, using the “α” value method. And comparative analysis with the measured result was performed. The simulation computed result was found in good coincidence with the measure computed result.