Reference Evapotranspiraion (ET0) is a basic conception in hydrology. It is especially important for crop water requirement and agricultural irrigation. The Penman-Monteith Equation proposed by FAO supplies a standard method to compute ET0. But this method requires numerous meteorologic data, so it may get into difficulty in some area where meteorologic data is scare. Therefore, it is of great significance to find a new method which is simple, feasible and precison promising.Using ten-day meteorologic data of Xiaohe irrigation station, Shanxi province during 1978 and 2003, this paper gets ET0 for different time steps of ten days, one month and one year with three empirical equations, namely Hargreves Equation, Priestley-Taylor Equation and Markkink Equation. Using FAO Peman-Monteith ET0 (ET0, pm) as the basis, their performance is evaluated and their error fluctuation trend is anylyzed. Only main factors influencing ET0 are considered in the three equations, so their results are inaccurate compared with Penman-Monteith Equation. Gernerally, Hargreaves Equation overestimates ET0, which is especially notable in summer and autumn of rainy seasons with high temperature. The positive correlation between Hargreaves Equation's error and precipitation is notable. Priestley-Taylor ET0 is statistically consistent with ET0, pm to some degree. Its relative error varies as an inverse"U"shape in a year influenced by the ratio of the aerodynamic component to the radiation component in Peman-Monteith Equation. And the negative correlation between the relative error and the ratio can be expressed by an expotential polynomial. As to the Markkink Equation, it underestimates ET0 and the reasons require further study.Three different types of artificial neural network models are established for simulating and forcasting ET0, including back propagation network, radial basis function network based on project pursuit and dynamic radial basis function network. The three models have their own characters, but their main conclusions are similar as well as their computation precision. The mathematic experiments of them... |