| 注册
首页|期刊导航|东南大学学报(自然科学版)|基于支持向量机方法的气候变化影响下台风生成预测建模

基于支持向量机方法的气候变化影响下台风生成预测建模

卫苗苗 方根深 葛耀君

东南大学学报(自然科学版)2025,Vol.55Issue(5):1283-1290,8.
东南大学学报(自然科学版)2025,Vol.55Issue(5):1283-1290,8.DOI:10.3969/j.issn.1001-0505.2025.05.008

基于支持向量机方法的气候变化影响下台风生成预测建模

Typhoon genesis prediction model under climate change impact based on support vector machine method

卫苗苗 1方根深 2葛耀君2

作者信息

  • 1. 同济大学土木工程防灾减灾全国重点实验室,上海 200092
  • 2. 同济大学土木工程防灾减灾全国重点实验室,上海 200092||同济大学桥梁结构抗风技术交通行业重点实验室,上海 200092
  • 折叠

摘要

Abstract

Tropical cyclones cause severe damage annually,making accurate predictions of genesis frequency and location crucial for disaster mitigation.This study proposes a support vector machine(SVM)-based data-driven approach for cyclone genesis prediction and evaluates its applicability under climate change sce-narios.Key meteorological parameters,including 850 hPa absolute vorticity(AV),600 hPa relative humidity(RH),vertical velocity(VV),sea surface temperature(SST),and 850-200 hPa wind shear(WS),are se-lected to construct a high-dimensional mapping for Western North Pacific(WNP,0°N-60°N,100°E-180°E)cyclone genesis prediction.The SVM model is trained using MIROC(model for interdisciplinary research cli-mate)historical data(1979-2014),with optimization of positive-negative sample ratios and train-test splits.It is then applied to MIROC6 SSP585 data(2015-2019),and predictions are compared with Japan Meteoro-logical Agency(JMA)observations to validate model performance.Finally,future projections for 2020-2100 under SSP585 are conducted.Results indicate an overall increasing trend in cyclone genesis frequency and a long-term northeastward shift in genesis locations under SSP585.SVM is effective for cyclone genesis prediction under climate change scenarios,providing insights for future risk assessment and disaster mitiga-tion.

关键词

台风生成/支持向量机/机器学习/气候变化

Key words

typhoon genesis/support vector machine/machine learning/climate change

分类

天文与地球科学

引用本文复制引用

卫苗苗,方根深,葛耀君..基于支持向量机方法的气候变化影响下台风生成预测建模[J].东南大学学报(自然科学版),2025,55(5):1283-1290,8.

基金项目

国家自然科学基金资助项目(52108469,52278520) (52108469,52278520)

中国科协青年人才托举工程资助项目(2023QNRC001) (2023QNRC001)

上海市教育委员会晨光计划资助项目(22CGA21) (22CGA21)

中央高校基本科研业务费专项资金资助项目(22120220577). (22120220577)

东南大学学报(自然科学版)

OA北大核心

1001-0505

访问量4
|
下载量0
段落导航相关论文