Feature Analysis And Forecast Of Precipitation In Anhui Province Based On Wavelet Analysis

Meiyu and typhoon precipitation are important components of flood season precipitation and are the main influencing factors of extreme precipitation events in Anhui Province.The periodic analysis and prediction of Meiyu,typhoon and annual precipitation series were carried out respectively,and the increasing or decreasing trend of precipitation series and its predicted value were obtained,which can provide data support for flood and drought disaster prevention and water resources allocation in Anhui Province.Morlet wavelet was used to analyze the periods of Meiyu,typhoon and annual precipitation in the region in north of the Huai River,between the Yangtze River and Huaihe River and in the south of the Yangtze River from 1957 to 2011.The increasing and decreasing trends of meiyu,typhoon and annual precipitation in three regions were analyzed based on the obtained periods.The original precipitation sequence was decomposed into 4-layer subsequences by db wavelet technology.The ARIMA model was used to predict each subsequences separately,and the predicted values of each subsequence were synthesized to obtain the predicted values of the original precipitation sequence.The ARIMA model was used to analyze and predict the original precipitation sequence directly at the same time.The results of the analysis and prediction of the precipitation sequence by wavelet,ARIMA model and ARIMA model were compared.The main results are as follows:(1)The meiyu precipitation is mainly affected by the periodic fluctuations of 15 a,5a and 10 a,the typhoon precipitation is mainly affected by the periodic fluctuations of 15 a and 7a,and the annual precipitation is mainly affected by the periodic fluctuations of 28 a,15a,10 a and 5a in Anhui Province..(2)Meiyu,typhoon and annual precipitation in the region of north of the Huai River,the region of Yangtze-Huai River and the region of south of the Yangtze River have different trends.Based on the periodic fluctuation,the Meiyu precipitation in the north of Huaihe River maintained a downward trend for a certain period of time,the Meiyu precipitation in the region between the Yangtze River and Huaihe River maintained an increasing trend for a certain period of time,and the Meiyu precipitation in the region in south of the Yangtze River maintained an upward trend for Typhoon precipitation in the region in north of the Huaihe River will maintain a downward trend in a short time,typhoon precipitation in the region between the Yangtze River and Huaihe River is in a downward trend,and typhoon precipitation in the region in south of the Yangtze River will begin to decrease.The annual precipitation in the region in north of the Huaihe River maintained a short-term upward trend.The annual precipitation in the region between the Yangtze River and Huaihe River will decrease in a short time,and then begin to rise,and the annual precipitation in the region in south of the Yangtze River will keep an overall upward trend.The predicted trend of meiyu,typhoon and annual precipitation in the region in north of the Huai River,between the Yangtze River and Huaihe River and in south of the Yangtze River is consistent with the trend of the observation of precipitation,which indicates that the predicted trend of precipitation in Anhui Province based on periodic fluctuation has certain accuracy.(3)ARIMA model and the combined model of wavelet and ARIMA have good prediction effect for the precipitation in Anhui Province,and the combined model of wavelet and ARIMA has better prediction effect than ARIMA model.The mean absolute percentage error of ARIMA model for meiyu precipitation prediction in the region of north of the Huai River,the region of Yangtze-Huai River and the region of south of the Yangtze River from 2012 to 2016 are 15.1%,12.9% and 16.8%.The mean absolute percentage error of typhoon precipitation prediction are 29.5%,13.4% and 11.8%.The mean absolute percentage error of annual precipitation prediction are 8.6%,14.0% and 11.5%.The mean absolute percentage error of the combined model of wavelet and ARIMA for meiyu precipitation prediction in the region of north of the Huai River,the region of Yangtze-Huai River and the region of south of the Yangtze River from 2012 to 2016 are 10.5%,10.9% and 6.3%.The mean absolute percentage error of typhoon precipitation prediction are 9.6%,9.7% and 14.0%.The mean absolute percentage error of annual precipitation prediction are 8.5%,9.9% and 6.9%.The prediction accuracy of ARIMA model for precipitation anomalous years is poor,and there are many anomalous years with large errors.The combination model of wavelet and ARIMA has better prediction accuracy for precipitation anomalies,and the combination model of wavelet and ARIMA has fewer errors.It indicates that the combination model of wavelet and ARIMA can reflect the detailed characteristics of the series and reduce the impact of precipitation anomalies.ARIMA model and combination model of wavelet and ARIMA are data-driven methods.They do not directly consider the physical mechanism of precipitation.As a result,the prediction errors of meiyu,typhoon and annual precipitation anomalous years in the region of north of the Huai River,the region of Yangtze-Huai River and the region of south of the Yangtze River are large.The combination of precipitation physical mechanism(such as ENSO index)and data-driven precipitation analysis and prediction model may have higher accuracy.