论文部分内容阅读
对影响土壤墒情的主要气象要素,平均气温、相对湿度、日照时数、平均风速、蒸降差和前一旬土壤墒情进行分析合并,建立BP-ANN土壤墒情预报6因子模型;通过缺省因子检验法,判断土壤墒情对6个因子敏感程度,简化冗余因子,构建BP-ANN的3因子(相对湿度、日照时数、前一旬土壤相对湿度)墒情预报模型。结果表明:3因子模型均方根误差3.55,具有数据收集和处理量小的优点,基本能够达到所需精度和拟合度。在北京市山区和平原区2个典型站点的模拟检验表明,3因子模型实测值与预测值的拟合关系均达到极显著相关水平,可操作性强的特点。
The main meteorological factors influencing soil moisture content, average temperature, relative humidity, sunshine hours, average wind speed, steam drop difference and the previous ten days of soil moisture were analyzed and combined to establish a 6-factor model of BP-ANN soil moisture forecasting. By default factor Test method to determine the degree of soil moisture sensitivity to six factors, simplifying the redundancy factor, building BP-ANN 3 factors (relative humidity, sunshine hours, the previous ten days of soil relative humidity) moisture forecasting model. The results show that the root mean square error of the 3-factor model is 3.55, which has the advantages of small data collection and processing capacity and can basically achieve the required precision and fitting degree. The simulation test of two typical sites in the mountainous area and plain area of Beijing shows that the fitting relationship between the measured and predicted values of the three-factor model reaches the extremely significant correlation level and the operability is strong.