广西暴雨集中度智能气候预测方法研究

Research on intelligent climate prediction methods of rainstorm concentration in Guangxi

利用1961—2023年广西79个气象观测站逐日降水和国家气候中心大气环流、海温指数资料,构建广西暴雨集中度计算方法,基于逐步回归方法、粒子群-神经网络、随机森林算法,建立暴雨集中度气候预测模型。结果表明,广西存在以桂林和柳州两市北部为中心的桂东北地区、以“东兰、巴马、凤山”为中心的桂西山区和沿海地区三个暴雨集中度高值区,暴雨集中度异常大小基本反映发生洪涝和干旱灾害的严重程度。经过2020—2023年气候预测试验,粒子群-神经网络算法预测效果最好,其次为随机森林算法,第三是逐步回归方法。

In this study,using the daily precipitation of 79 meteorological observation stations in Guangxi and the data of the atmospheric circulation indices and sea surface temperature indices data of the National Climate Center from 1961 to 2023,we constructed a method for calculating the rainstorm concentration in Guangxi,and established a climate prediction model for the concentration of rainstorm based on the stepwise regression method,the particle swarm neural network,and random forest algorithm.The results showed that there are three areas of high rainstorm concentration in Guangxi,namely,the northeastern part of Guangxi centered on the northern part of Guilin and Liuzhou,the mountainous area in the western part of Guangxi centered on“Donglan,Bama,Fengshan”area,as well as the coastal area.The anomaly of rainstorm concentration basically reflects the severity of flooding and drought disasters.The climate prediction model of rainstorm concentration based on the stepwise regression method,particle swarm optimization neural network and random forest algorithm is established.According to the climate prediction experiments in 2020—2023,the most effective prediction was made by the particle swarm-neural network algorithm,followed by the random forest algorithm,and finally by the stepwise regression method.