大数据驱动的暴雨灾害预测模型  

A Rainstorm Disaster Prediction Model Driven by Big Data

作　　者：庞畅 

机构地区：[1]杭州师范大学经亨颐教育学院,浙江 杭州

出　　处：《可持续发展》2025年第2期270-279,共10页Sustainable Development

摘　　要：随着全球气候变化的加剧和极端天气事件的频发,有效预防和应对自然灾害成为当务之急。利用大数据技术对历史气候数据进行深入分析,可以识别潜在的暴雨灾害风险,并制定相应的应急预案。本文旨在通过剖析中国地理环境时空演化特征,结合LSTM和XGBoost等算法建立暴雨灾害预测模型。首先,对数据进行可视化,分析中国降雨量的时空演化模式;采用Spearman相关性分析多种特征对暴雨形成的影响。其次,基于XGBoost算法对极端暴雨天气进行临界条件分析。最后,使用LSTM和CNN网络捕捉降雨量与地理位置间的复杂非线性关系,并结合时空集成树方法进行优化,构建暴雨灾害预测模型。研究结果表明,该模型能够有效降低预测误差,实现对不同地区的极端暴雨天气进行精准预测,为我国制定有效的防灾减灾策略提供了有力支持,有助于提升各地区的抗灾能力。With the intensification of global climate change and the frequent occurrence of extreme weather events, effective prevention and response to natural disasters have become a top priority. Using big data technology to conduct an in-depth analysis of historical climate data, potential rainstorm disaster risks can be identified, and corresponding emergency plans can be formulated. The purpose of this paper is to analyze the spatiotemporal evolution characteristics of China’s geographical environment and establish a rainstorm disaster prediction model by combining LSTM and XGBoost algorithms. Firstly, the data were visualized to analyze the spatiotemporal evolution pattern of rainfall in China. The Spearman correlation was used to analyze the influence of multiple characteristics on the formation of heavy rainfall. Secondly, the critical condition analysis of extreme rainstorm weather was carried out based on XGBoost algorithm. Finally, the LSTM and CNN networks are used to capture the complex nonlinear relationship between rainfall and geographical loc

关 键 词：暴雨灾害预测 Spearman相关性分析 XGBoost LSTM CNN 

分 类 号：P45[天文地球—大气科学及气象学]