Research On Correction Algorithm Of Precipitation Data In Flood Season Based On Machine Learning

Prediction of precipitation during the flood season has long been a key task of China's climate prediction,which is related to national economy and people's livelihood,and is also one of the difficult problems in climatology.At present,there are some models for forecasting the precipitation during the flood season,but there are certain deviations,which makes it difficult to forecast more accurately.This paper corrects the precipitation forecast data in the flood season in C hina,mainly uses the existing historical data to establish a model to predict precipitation,and compares with the actual observation data to determine the best correction model by using relevant evaluation index analysis.This paper summarizes the characteristics and current status of the flood season precipitation forecast data in the national region,introduces and studies the application of machine learning in the revision of the flood season precipitation forecast data in the Jianghuai region,summarizes a relatively ideal algorithm model,and extends to the national region Revision of precipitation forecast data in flood season.The main research conte nts include:(1)The definition and difference between weather forecast and climate prediction are elaborated in detail.The main research object of this paper is the revision of the precipitation forecast data in the flood season in the climate.The difficulties of climate prediction are shown in three aspects: climate change,causes of climate change,and difficulties in climate prediction,which reflects the necessity of the algorithm proposed in this paper.(2)The revised algorithm of flood season precipitation forecast data based on machine learning is studied.Combined with the characteristics of flood season precipitation data,different machine learning algorithms are introduced into the calculation of flood season precipitation forecast data.Mainly include: neural network back propagation,random forest Random Forests.Considering that the Yangtze-Huaihe River Basin is one of the important production areas of industry and agriculture in C hina,it is of great significance for preventing and reducing catastrophic climate.Therefore,based on the application background and characteristics of the precipitation forecast in the flood season,an improved algorithm is adopted to determine the best model algorithm to improve the accuracy of the precipitation forecast in the flood season.(3)Analysis of the correction results of the precipitation forecast data in the flood season based on the commonly used evaluation indicators mean absolute error,mean square error,and temporal correlation coefficient.The performance of the CWRF precipitation forecast of the National Meteorological Center of the China Meteorological Administration based on the algorithm in this region is evaluated.The above-mentioned algorithm model with relatively good forecasting performance in the flood season is extended to all regions of the country to verify the forecasting performance and generalizatio n ability of the region based on this algorithm.According to the climate characteristics of different regions,C hina is divided into eight regions.Finally,the historical data of the eight regions of C hina are pre-processed by distance data,and the modified algorithm of flood forecast precipitation data based on machine learning is used to model different regions Forecast.The results verify the effectiveness and feasibility of the algorithm applied in the actual business.