中文摘要：

在北京近郊区海淀公园架设涡度相关仪，连续观测城市公园绿地水热通量及其环境影响因素数值日变化。分别采用神经网络、相关分析和灰关联方法研究植被生长季节影响水热通量的环境因素权重，结果表明：公园绿地水热通量受多个环境因素的综合影响，净辐射影响最大，其次是下垫面温度，再次相对湿度和气温影响潜热，风速对潜热的影响较弱，而气温和风速对显热的影响也相对较弱。神经网络方法直接给出非线性关系权值，分析结果较准确，优于统计分析方法。

英文摘要：

Eddy covariance system was set up in Haidian Park of Beijing urban area. Water and heat fluxes between underlying surface and atmosphere and environmental variables were measured continuously. The influent weights of environmental variables for the fluxes were gotten with neural network analysis, correlation analysis, and gray relevancy analysis during growth season. The results indicated that the fluxes were affected by several environmental variables. The most important environmental variable affecting the fluxes was net radiation, followed was ground temperature, and third were relative humidity and atmospheric temperature for latent heat. Wind speed had less influence on latent heat. Also atmospheric temperature and wind speed had less influence on sensible heat. Neural network analysis showed nonlinear accurate weights directly, was more effective than statistic analysis.