In this study, by using the former data of precipitation that belong to Shouguang city we investigated the advantages and
disadvantages of SARIMA and the regression model with seasonal latent variable in forecasting precipitation demand. It can
be basically concluded that the SARIMA model has good model fitting degree in decision-making for agricultural irrigation.
In spite of an appropriate model for the time series defined as the Box–Jenkins methods, there are problems about this model
because of its failure in forecasting especially if in the past the sequence of time series has abnormal changes.
If we consider the advantages of the forecasting we can make full use of natural rainfall in corresponding areas, and
underground water resources. These would provide a quantitative index and theoretical basis for the region to set a
reasonable irrigation system. It has the important significance in that it supports agricultural and long-term stability of
groundwater resources for sustainable development.