Based on pan-evaporation data of past years and other meteorological element data of conventional surface meteorological measurement from 112 weather stations, orthogonal expansion method and canonical correlation are used to investigate the correlativity between monthly evaporation and conventional meteorological elements and to analyze and compare the capacity for each conventional meteorological element to explain potential evaporation, while regression analysis is applied to validate the foregoing analysis results and to seek the optimalizing element combination for conventional meteorological elements to simulate evaporation. As it turned out, there were 6 conventional meteorological elements which had notable effects on evaporation. The degrees of correlation between conventional meteorological elements and evaporation were as follow: average relative humidity(H)＞average temperature(T)＞average land surface temperature(Dt)＞sunshine duration(S)＞average speed of air(W)＞average vapor pressure(Vp)＞average atmospheric pressure(P)＞amount of precipitation(R). Among these essentials, essential H could explain 63.5% population variance of field variables for evaporation. T and Dt also had greater explanatory ability (more than 31% population variance interpreted). The explanatory ability of S, W, Vp and P to population variance of evaporation was relatively approximate about 20%. And precipitation had little effects on evaporation. All these were in accordance with the explanation of thermodynamics and dynamics in water evaporation.
　　According to correlation analysis and regressive simulation, average relative humidity took precedence over other conventional meteorological elements in simulating evaporation. The combination of relative humidity attached speed of air was probably the best two-essential combinatory simulating evaporation. Based on average relative humidity, average temperature, average speed of air, sunshine duration and average vapor pressure, the first three-essential combination and five-essential combination were respectively the optimum combination simulating evaporation by conventional meteorological elements under simply-universal requirement and high-precision requirement. The three-essential combination generated an average relative error of fitting, 2.77%, in this simulation while the five-essential combination made the identical error equal to 1.96%.
　　In the paper, mutuality of spatial distribution for essentials is analyzed by orthogonal expansion method and canonical correlation at the same time as correlativity of their temporal changes is investigated by regressive simulation (stepwise regression included). Both methods have validated and replenished opposite party reciprocally. So demonstration by reasoning is all-around and systematical here. The research has deepened the understanding related to potential evaporation, and is meaningful for reckoning evaporation and quantizing its spatial distribution.