Abstract/Summary

Pan evaporation plays a critical role in estimating water budget and modeling crop water requirements. However, it has been measured at a very limited number of meteorological stations. Estimation of pan evaporation from measured meteorological variables offers an important alternative and drawn increasing attention in the recent years. This paper investigated the performance of support vector machine (SVM) in the estimation of monthly pan evaporation using commonly measured meteorological variables in Three Gorges Reservoir Area in China. Evaluation suggested that SVM models showed remarkable performances and significantly outperformed the empirical model. The SVM model with polynomial as kernel function outperformed that with radial basis function. In the case of unavailable measurements of pan evaporation and meteorological variables to construct the SVM model, pan evaporation can be well-estimated by SVM model developed using data at other sites. The results indicated that the SVM method would be a promising alternative over the traditional approaches for estimating pan evaporation from measured meteorological variables.