干旱区生态环境知识资源中心

It is important for sustainable use of soil water resource to forecast the soil moisture in forestland. In this paper, taking the soil water data as an example, the soil moisture content was divided into a low frequency component and a high frequency component using wavelet analysis, and then NARX neural network was used to build model I and model II. For model I, low frequency component was the input variable, and for model II, low frequency component and high frequency component was predicted. The results showed the average relative error for model I is 3.52% and for model II is 0.31%. The average relative error of predicted soil moisture at 100cm layer using model II is 0.81%, then soil water content in 40 cm and 200 cm soil layers is selected and the forecast errors are 4.94 % and 0.41 %, showing that model II well predict soil water.