中文摘要：

利用AXWG03自动气象站于2007年3～6月对烟台市一果园内表层土壤温度及其环境气象因子进行了观测，分析了土壤温度与气象因子之间的相关性，并建立了基于气象因子的土壤温度预测模型。结果表明在相对湿度、大气温度、风速、太阳总辐射和大气压5个常规气象因子中，相对湿度、大气温度、风速、太阳总辐射与土壤温度存在极显著相关关系，大气压与土壤温度相关关系不显著；故以相对湿度、大气温度、风速、太阳总辐射等4个气象因子为输入变量，分别利用多元线性回归和BP人工神经网络方法建立土壤温度预测模型，其中BP人工神经网络模型预测值最大相对误差不足1．24％，多元线性回归模型预测值相对误差在1．93％以上。因此建立的神经网络模型具有很高的精度，能很好地满足土壤温度的预测要求。表4，参13。

英文摘要：

The surface soil temperature and the weather factor were measured with AXWG03 automatic weather station from May to June, 2007 in an orchard of Yantai, and the relativity between the soil temperature and the weather factor was analyzed. The model of soil temperature based on weather factor was established. The result showed that the correlation between 4 normal weather factors including relative humidity, air temperature, solar radiation, wind speed and soil temperature was very significant, and the correlation between atmospheric pressure and soil temperature was not significant. The paper took 4 weather factors including relative humidity, air temperature, solar radiation and wind speed as the in- put variable, and established the forecast model of soil temperature with multiple linear regression and BP artificial neural network.The maximum relative error of forecast of BP artificial neural network was less than 1.24%, but the smallest relative error of forecast of multiple linear regression reached 1.93%. So BP artificial neural network model which had very high precision can be well used to forecast soil temperature.