Temperature And Rainfall Prediction Models Based On EEMD And Theirs Application

Meteorological factor play a key role in social progress and sustainable economic development.The impermanence of weather changes will bring irreversible disasters to human beings.Therefore,scientifically and effectively predicting weather changes is of great significance for accelerating social progress and economic development.Temperature and rainfall are two important meteorological indicators.Abnormal temperature will have a negative impact on human life and work,and will bring huge losses to agricultural and industrial production.Storms and droughts have always been two of the main threats to human property and life.Therefore,accurate predictions of temperature and rainfall are particularly important.Then,apply the proposed temperature and rainfall prediction model to the weather option valuation.The time series forecasting of meteorological elements based on EEMD and the multi-factor option estimation of weather index based on EEMD are based on the multi-scale combination model of meteorological elements and the multi-factor option of weather index,comparing the option valuations of other single temperature and rainfall prediction models,the model option valuation based on this paper is the most accurate.This paper proposes two multi-scale temperature prediction models and rainfall prediction models based on EEMD.For the temperature prediction model,the EEMD decomposition method is first used to decompose the temperature into multiple IMFs with different frequencies.Then calculate their Hurst exponent,fractal characteristics between 0.5 and 1 according to the Hurst index value,then extract sequences with fractal features;and then combine the correlation analysis to reconstruct these IMF into high frequency term,medium frequency term and fractal term,and reconstruct the sequence.Prediction of reconstructed sequences using neural network methods and improved fractal interpolation methods.Finally,the generalized regression neural network(GRNN)is used to integrate the predicted values of these items to obtain the final temperature prediction value.And through the application of the monthly average temperature forecast in Nanjing,the prediction results are compared with the prediction results of other temperature prediction models.For the rainfall prediction model,in order to predict the rainfall more accurately,amulti-scale combined prediction model based on EEMD-SE is proposed.According to the nonlinear and non-stationary characteristics of rainfall time series,based on the decomposition-reconstruction-prediction-integrated idea,the ensemble empirical mode decomposition(EEMD),sample entropy(SE),BP neural network,Elman neural network and Support vector machine regression(SVR)predict the monthly average rainfall in Nanjing,and compared with the prediction results of BP neural network and Elman neural network,the results of the prediction model are better,and a new prediction model for rainfall is provided.