Application of Rainfall Forecasting Base on Improved Time Series Model
Bai Yu-jie · Jisuanji fangzhen · 2011
Rainfall forecast problem is studied.Rainfall is the result of mutual influence of natural factors which have the characteristics if non-stationary and time-varying,and the traditional forecasting methods cannot reflect the law of non-stationary and time-changing and forecasting accuracy is low.In order to improve the rainfall forecast precision,a rainfall forecast method is proposed combining wavelet transform with time series prediction model(ARIMA) in this paper.Firstly,rainfall raw data are normalized in this method,then non-stationary data are processed into smooth data by wavelet transform,then using ARIMA model to analyze the time series,the rainfall optimal rainfall forecast model is established,and finally,the model is tested based on actual rainfall.Simulation results show that the proposed method has higher forecast precision than the traditional forecasting methods,and it can reflect the law of changing of rainfall very well and provide a new prediction way for rainfall forecasting.