| Rainstorm disaster is one of the most destructive natural disasters in China.It is usually accompanied by a series of secondary disasters such as mudslides and landslides.Guangxi is a major area in southern China with frequent torrential rains.During the first flood season(April-June),the precipitation intensity is high,precipitation is high,and it is affected by complex geographical environment.It has the characteristics of locality,suddenness and short duration.It is one of the main areas of frequent rainstorms in South China.The paper will use the EOF analysis,wavelet analysis,M-K test and sliding T test to discuss the influence of topographical factors on the strong precipitation of 57 a storms in Guangxi from 1961 to 2017.The paper focuses on the spatial distribution characteristics of the rainstorm precipitation in the first flood season in Guangxi.Using the weather diagnosis method,the development rules,formation mechanism and interannual variation characteristics of the heavy rain in the flood season from 1961 to 2017 in Guangxi were summarized.According to the topographical distribution and precipitation characteristics of Guangxi,Guangxi is divided into three different regions,and the KPCA feature extraction method and the random forest algorithm set model are established to predict and forecast the heavy rain in Guangxi during the first flood season.The following conclusions were obtained:(1)Under the influence of topography,the spatial distribution pattern of heavy rain in the first flood season in Guangxi is more in the northeast and less in the southwest.There are three high-value centers and one low-value center.The high-value areas are Rongbei,Yongfu and other areas in northern Guangxi.In the northern regions of Jinxiu,Mengshan and other coastal areas such as Dongxing,the low-value area is Ningming.(2)The interannual variation of heavy rainfall in Guangxi during the first flood season is significant,with obvious short-period changes of 1-2a,4-6a,and long-period changes of around 24 a.In the long-term trend,the overall change of heavy rainfall in the first flood season in Guangxi is relatively flat,and the mutation is not obvious.(3)Using the EOF method to analyze the spatial characteristics of the heavy rainfall in Guangxi during the first flood season,the first mode is the consistency of the whole region and is distributed from east to west.The high value area is located in Lingui,Yongfu and Laibin.The value area is located in the northwestern part of Guangxi,and the variance contribution rate is 30.14%.The second mode is the spatial distribution of the northwest-southeast reverse distribution.The high value area is located in Donglan,Tiandong and other places.The large area in southeastern Guangxi is a negative area,and the variance contribution rate is 12.21%.The third mode is a spatial pattern with north-south reversal and decreasing distribution from north to south.The high-value area is located in Yongfu and Xing'an,and the low-value center is located in southern Guangxi.The variance contribution rate is 9.4%.(4)Using the time coefficient corresponding to the eigenvector obtained by EOF decomposition,the time variation characteristics of the heavy rain in the first flood season in Guangxi are analyzed.The time coefficient of the first mode is between 20 and-40,and there is an oscillation period of 3-4a.The type of overall favoritism.The time coefficient of the second mode is between 6 and-6,showing a downward trend,and there is an oscillation period of about 12 a,which is in the overall type of drought.The time coefficient of the third mode is between 15 and-15,showing an upward trend,which is in the northern part of the country and the type of drought in the south.(5)A statistical analysis of large-scale continuous heavy rains in Guangxi during the first flood season revealed that there were 41 consecutive heavy rainstorms in Guangxi,with an average annual average of 0.73.The frequency of occurrence in April was the least,followed by May,and the frequency appeared in June.The interannual and continuous changes in the continuous heavy rain in Guangxi are more obvious.The linear trend analysis found that there was a slight decrease in April,while in May and June it was gradually increasing,and the trend of increase in May was more obvious.(6)The analysis shows that the impact mechanism of large-scale continuous heavy rain in different months is different,which is represented by two troughs and two ridges in April and short-wave troughs in low latitudes affecting Guangxi;May is the situation of two ridges and one trough;in June,one trough and one ridge cooperate with the middle and low latitudes of the East Asian trough.These circulation conditions are conducive to the accumulation of cold air and southward impact,while Guangxi is affected by the southwesterly flow of the subtropical high edge in May and June;the lower layer is the abnormal convergence of the airflow,which is conducive to the continuous rise of low-level water vapor.(7)Analysis of water vapor and dynamic conditions shows that the monthly changes correspond to different main sources of water vapor.Among them,the water vapor sources in April are mainly in the South China Sea and the Bay of Bengal.In May,the water vapor in the South China Sea,the Indian Ocean and the Bay of Bengal is mainly over Guangxi.Transportation;the source of water vapor in June is mainly in the Indian Ocean and the Bay of Bengal.From April to June,there was a clear upward movement over Guangxi,which corresponds to strong unstable energy,which provided a favorable trigger mechanism for the occurrence of heavy rain in Guangxi.(8)Using the KPCA method and the random forest algorithm to study the predictor data excavator,establish a new nonlinear artificial intelligence forecasting model,and study the model of the rainstorm precipitation forecasting model in Guangxi.The forecast results are obtained and tested.The new model based on the independent model of 2012-2016 has a TS score of 0.14,and the European central numerical forecast product has a TS score of 0.07.From the partition analysis,in the first district,the new model TS score is 0.16,the European fine grid is 0.12;in the second zone,the new model TS score is 0.10,the European fine grid is 0.01;in the third zone,the new model TS score is 0.14,and the European fine grid is 0.02,both are superior to ECMWF.Aggregate forecast results.The new method has stable prediction results and high precision.It has certain scientific and practical significance for the actual prediction of heavy rain in Guangxi during the flood season. |
PDF Full Text Request