干旱区生态环境知识资源中心

The symbolic regression (SR) method has been applied to calculate reference crop evapotranspiration (ET0) in arid china, and was proved suitable to find site-specific equations. However, its application in humid regions has not been discussed and the difference between the equations in different regions is still unknown. The SR method was engaged to establish equations with the same inputs to HargreavesSamani (HS) equation for ET0 estimation in humid China. For each station, the most precise equation, the simplest equation and the trade-off equation (to balance the complexity and accuracy) were selected out and evaluated. The results showed that the SR equations could be used for ET0 calculation in humid China, and performed better than the local calibrated HS equation. The most precise equations largely varied among stations in both structures and terms in humid and arid China, and they included more terms with high complex index (CI) value in humid China. The simplest equations were uniform in structure either in arid or in humid stations, but the variables included were the daily minimum air temperature (T-mi) and temperature range (TR) in humid China rather than maximum air temperature (T-ma) and the extraterrestrial radiation (R-a) in arid China. For trade-off equations, three different structures were found for five stations in humid China. The variables of T-mi, TR and Ra was included in trade-off equations in humid China instead of T-ma. and R-a in arid China. It indicated that Ra and T-ma. were the important variables in determining reference evapotranspiration in arid region, while T-mi, TR and R-a played a vital role in estimating ET0 in humid region.