Design And Implementation Of Guangxi Monthly Precipitation Climate Prediction System Based On Machine Learning

Monthly-scale climate prediction provides technical support for disaster prevention decision-making and deployment of leading departments at all levels,and is of great significance to disaster prevention and mitigation.In recent years,machine learning methods have been widely used in the field of climate prediction,in order to explore the application effect of machine learning algorithms in monthlyscale precipitation prediction.Based on the BCC＿CSM1.1 coupled climate model circulation prediction and NCEP/NCAR reanalysis data,this paper improves the dynamical-statistical prediction method.Apply the machine learning algorithm to the statistical part of the dynamical-statistical method,and conduct research and application on the monthly precipitation and climate prediction method based on machine learning.The main contents of the study are as follows:(1)Based on the BP neural network,support vector machine,and random forest algorithm,the dynamic-statistical forecasting method was improved,and the dynamical-machine learning forecasting model was established,and the return test was carried out in the historical data of monthly precipitation in Guangxi from 1991 to 2020 from June to August.The experimental results show that the machine learning algorithm has a strong ability to map the nonlinear relationship between the climate impact factor and the time coefficient,which improves the prediction accuracy of the time coefficient and effectively corrects the precipitation prediction directly output by the BCC＿CSM1.1 model.Among them,the BP neural network has the best prediction effect.Compared with the BCC＿CSM1.1 model,the direct prediction Ps score has increased by 7.63 points,and the support vector machine and random forest have increased by 6.83 points and 5.94 points.(2)According to the good improvement effect of the machine learning algorithm forecasting model in monthly precipitation forecasting and the objective needs of Guangxi Climate Center’s monthly precipitation and climate forecasting business,a monthly precipitation and climate forecasting system based on machine learning in Guangxi was developed using Python and Django framework.The system is designed to realize the functions of climate data processing,climate monitoring,machine learning model prediction and prediction verification.Climate data processing is used for data management of climate data,climate monitoring is used for climate analysis,machine learning model prediction function can use machine learning algorithm prediction model for monthly precipitation prediction,and prediction verification is used for forecast verification.In this paper,the demand analysis,design and implementation of the system are studied in detail.After testing,it is found that each functional module of the system operates normally,the output data is accurate,and the performance is stable.The system has a good graphical processing ability of meteorological data,has the advantages of simple operation and high degree of visualization,simplifies the workflow of climate forecasting business,improves work efficiency,and has certain guiding significance and practical value in the practical application of monthly precipitation forecasting business.