中文摘要：

以柳圆口灌区2002～2005年的气象数据为基本资料，采用参考作物腾发量预测的前馈网络模型（BP—ET0），研究了参考作物腾发量对气象因子的敏感性。研究结果表明：在参考作物腾发量预测的6个气象因子中，相对湿度对BP—ET0预测效果影响最大，缺少它时预测合格率只有64．93％，其次是实际风速和日照时数，缺少它们时预测合格率分别为73．15％和76．98％，最高气温、最低气温和平均气温对预测效果影响较小，缺少它们时预测合格率均在80％以上。

英文摘要：

With the meteorological data from 2002 to 2005 of Liu Yuan Kou Irrigation District, using the reference crop evapotranspiration forecasting model for feed-forward neural networks （BP-ET0）, the sensi- tivity of meteorological factors on reference crop evapotranspiration is studied. The research results are as follow. In six meteorological factors on reference crop evapotranspiration, the relative humidity affects the forecasting result most, the forecasting passing rate is only 64. 93 percent without it. The actual wind speed and sunshine hours affect the forecasting result only less than the relative humidity, without them the forecasting passing rates are 73. 15 percent and 76. 98 percent respectively. The maximum temperature, minimum temperature and average temperature affect the forecasting result least, without any of them the forecasting passing rates are all more than 80 percent.