论文部分内容阅读
在长期田间试验基础上,分别利用数值模拟方法(Numerical Simulation,NS)和人工神经网络(Artificial Neural Network,ANN)模型构建江南平原土地整治区典型林地的土壤水分运动模型,并对土壤贮水量进行预测。NS模型校正结果表明,该模型虽能较好地预测林地土壤含水量动态变化,但是NS模型对训练期和验证期0~60cm土层贮水量预测的均方根误差(Root Mean Square Error,RMSE)分别为11.09和8.29mm,而ANN预测的RMSE分别为4.17和4.08mm,说明ANN的预测效果好于NS模型。最后,敏感性分析结果表明ANN预测精度对输入参数的敏感程度由高到低依次为:前期土壤贮水量>降水量>最高气温>最低气温。
Based on the long-term field experiments, the soil moisture movement models of typical woodlands in the land remediation area of Jiangnan Plain were constructed by numerical simulation (NS) and Artificial Neural Network (ANN) models, respectively. prediction. The result of NS model calibration shows that although this model can predict the dynamic change of soil water content in forest land, the Root Mean Square Error (RMSE) of NS model for prediction of water storage in 0 ~ 60cm soil layer during training and validation period ) Were 11.09 and 8.29mm, respectively. The RMSE of ANN prediction was 4.17 and 4.08mm respectively, which showed that the ANN prediction effect was better than the NS model. Finally, the results of sensitivity analysis show that the sensitivity of ANN prediction accuracy to input parameters from high to low are as follows: Pre-soil water storage> Precipitation> Maximum temperature> Minimum temperature.