基于贝叶斯网络的梅雨季节降水诊断分析

Diagnostic Analysis of Precipitation in Plum Rain Season Based on Bayesian Network

摘要: 江南地区由于地理气候因素，每年夏季都会进入梅雨季节，降水频繁。为了能够更加准确的对梅雨季节的降水情况进行分析，从而能够更好的预测梅雨季节可能到来的洪涝灾害，以防造成较大的经济损失，本文以南京市梅雨季节的降水状况为例，利用南京市2011~2022年每年6~7月的气象观测资料，选取发生降水天气的要素资料，基于贝叶斯网络模型和决策树模型，对南京市降水发生情况进行诊断分析。结果表明：贝叶斯网络模型准确率为83.61%，决策树模型的准确率为81.97%，可见，贝叶斯网络模型的准确率更高，效果也更好。

Abstract: Due to geographical and climatic factors, the Jiangnan region will enter the plum rain season every summer, with frequent precipitation. In order to more accurately analyze the precipitation in the plum rain season, it can better predict the possible flood disaster in the plum rain season, so as to prevent greater economic losses, in this paper, the precipitation in Nanjing during the plum rain season was taken as an example, the meteorological observation data from June to July of 2011 to 2022 were used to select the element data of precipitation weather, and the diagnosis and analysis of the precipitation occurrence in Nanjing was carried out based on Bayesian network model and decision tree model. The results show that the accuracy of Bayesian network model is 83.61%, and that of decision tree model is 81.97%. It can be seen that the accuracy of Bayesian network model is higher and the effect is better.