考虑人类活动用水的土壤含水量神经网络反演

为在利用神经网络对地表土壤含水量的模拟中,实现对降水等天然要素和人类活动用水的综合考虑,以MPDI(modified perpendicular drought index)作为人类活动作用下的地表干湿状况指标,结合传统的天然要素构建土壤含水量神经网络模型,对河北省2018年地表土壤含水量进行了模拟.结果表明:考虑MP...

Inversion of soil moisture using neural network considering human activities

Taking MPDI(modified permanent drawdown index)as an indicator of surface soil conditions under water utilization for human activities, a soil moisture neural network model with traditional natural factors was constructed to simultaneously consider the influence of natural elements and water use for human activities. The model was used to simulate the surface soil moisture in 2018 in Hebei Province. The results show that the simulation results of soil moisture content considering MPDI index are in good agreement with the measured values, with a correlation coefficient of 0. 7 in training period and 0. 5 in verification period. The soil moisture distribution of the neural network in a single day was analyzed and the correlation coefficient is 0. 67 for the example date, which indicates a good performance in revealing the spatial heterogeneity of soil moisture. The simulated results are consistent with the SMAP(Soil Moisture Active and Passive)products in Hebei Province, with higher values in summer and east plain, but with lower values in spring and northwest mountains.