Application Of Time Series Analysis In Rainfall Prediction Of Changsha City

Time series analysis is an important branch of mathematical statistics,which is usually used to study and process dynamic data with randomness.With the development of science technology and computer,the time series analysis method has gained more development space and become an indispensable data processing method in the field of natural science and social science research.Rainfall is an important indicator to reflect the climate status of a region.The change of climate in a region can be directly reflected in the change of rainfall in the region.The amount of rainfall will also affect the social production and ecological water demand.In recent years,with the intensification of climate change,the fluctuation of rainfall is large,and extreme rainfall events begin to occur frequently in urban areas,and the regional floods disaster and other phenomena caused by extreme rainfall events are becoming increasingly prominent.Therefore,it is of great significance to study the accurate prediction of urban rainfall in the future.This paper mainly uses the basic theory and method of time series analysis and neural network to study and analyze the rainfall time series data of Changsha City,to explore the regularity and characteristics of daily rainfall in Changsha City,and to predict it.Based on the dailyrainfall data of Mapoling area in Changsha from 2015 to 2017,BP,RBF neural network and ARMA-EGARCH are used to simulate and predict the rainfall time series data by using EVIEWS and MATLAB software,in order to provide some reference for rational allocation of water resources and reduction of economic losses caused by meteorological disasters.The empirical results show that:(1)All the three models can well predict the daily rainfall of Changsha,but the prediction accuracy of ARMA-EGARCH and RBF models is significantly higher than that of BP neural network models,which also verifies the feasibility of time series analysis in rainfall prediction.(2)The three models have their own advantages,ARMA-EGARCH model has higher fitting and prediction accuracy,BP neural network model is more accurate in predicting micro-rainfall weather,while RBF neural network model has better global prediction effect.Therefore,three methods can be combined to predict the rainfall in Changsha City and improve the accuracy of the prediction.(3)The ARMA-EGARCH model is based on the past rainfall data for modeling and prediction,while the BP and RBF neural network model is based on the data of six meteorological factors for modeling and prediction of rainfall.Based on different data,it provides two effective methods for rainfall forecasting in Changsha.The main research content of this paper is the important application of time series analysis and neural network in the rainfall forecast of Changsha City.This study provides an effective prediction method for the future rainfall of Changsha City,and provides some reference for relevant government departments to take effective measures to deal with natural disasters,rational planning of urban water resources.