基于2003-2007年千烟洲涡度相关通量塔观测的气象数据和蒸散发数据, 评价了常用的蒸散发模型模拟森林生态系统蒸散发的适用性, 包括Priestly-Taylor、Blaney-Criddle、Hargreaves-Samani、Jensen-Haise、Hamon、Turc、Makkink和Thornthwaite模型。 结果表明, 日尺度上Priestly-Taylor模型的模拟效果较好, 相关系数达0.953; 月尺度上Makkink模型的模拟效果较好, 相关系数达0.995; 而Thornthwaite模型在月尺度上模拟误差较大, 均方根误差与平均偏差分别为15.559和13.436; Jensen-Haise模型在日、月尺度上模拟效果均较差。采用偏相关法分析气象因子与蒸散发值的关系, 得出森林生态系统蒸散发驱动因子的贡献排序为: 辐射>温度>水气压>风速>土壤温度>相对湿度>白天时长, 即辐射对蒸散发的影响较为显著, 与基于辐射法的Priestly-Taylor和Makkink模型分别在日、月尺度上适用性较好的结论一致。

Using meteorological and evapotranspiration (ET) data acquired at the Eddy Covariance Flux tower in Qian- yanzhou, Jiangxi Province, for 2003 to 2007, we evaluated the applicability of 8 widely used evapotranspiration simulation models (Priestly-Taylor, Blaney-Criddle, Hargreaves-Samani, Jensen-Haise, Hamon, Turc, Makkink and Thornthwaite) for a forest ecosystem. Among these 8 models, the Priestly-Taylor model was the best (R=0.953) on a daily time scale, the Makkink model was the best (R=0.995) on a monthly scale, and the Thornthwaite model was the worst on a monthly scale (RMSE=15.559, MBE=13.436). The Jensen-Haise model failed in simulation of ET on both day and month scales. Partial correlation analysis of simulated ET against meteorological factors showed that the order of factors contributing to ET for the forest ecosystem was radiation>air temperature>surface pressure>wind speed>soil temperature>relative humidity>daytime length. Radiation was the most important driving factor for ET, which is consistent with the performance of radiation-based ET models (e.g., the Priestly-Taylor and Makkink models) being better than other models.