Abstract: The principle and modeling method of statistical diagnostic model were introduced. The model is based on precipitation in different periods, soil initial water content and their interaction. Models were established by the data of 87 monitoring sites in 23 counties of 7 provinces during 2012-2014, and were validated by the data of 2015. The results showed that the statistical diagnostic model had high qualification rates (more than 85%) in diagnosis and prediction. The non-fixed time interval of the adjacent soil moisture monitoring date would lead to the emergence of outliers, which was the main reason affecting the qualification rate of model Prediction. Statistical diagnostic model based on daily time series data can predict daily soil water content. In conclusion, statistical diagnostic model can be used alone as a soil moisture diagnosis model.

Key words: black loessial soil, organic carbon, feedback relationships., Loess Plateau, total nitrogen