Abstract: The purpose of this paper was to evaluate the performance of 6 models so as to provide theoretical, methodological and parametric basis for model optimization. Six independent diagnostic models of soil moisture were evaluated by using 4 indexes including preferred model ratio, verification method index, verification mode index and outlier index. The results showed that the model priority was as follows: subtractive statistical diagnostic model and interval days statistical diagnostic model > movable statistical diagnostic model > ratio statistical diagnostic model > statistical diagnostic model > balance diagnostic model. For model type, the daily time series models performed better than the time interval models. The model with high qualification rate included 4 characteristics, that is, the number of independent variables was not more than 3, the independent variables were relatively independent, the model could deal with the uncertain monitoring days, and the parameters were not artificially determined. The six independent models established by the monitoring points can be used separately and solve the nonuniversal of the model. The daily diagnosis and prediction of soil moisture can be realized for real-time matching with remote sensing information and crop growth information. Prediction accuracy of the subtractive statistical diagnostic model and the interval days statistical diagnostic model were the highest, the former was a statistical model based on the law of mass conservation, and the latter effectively solved the prediction error caused by irregular interval days.

Key words: rotifer, water environment, community structure., macrophyte