Scholarlink

<p>This study presents a statistical and deep Q-learning model framework for predicting typhoon genesis and trajectories, utilising kernel density estimation and a generalised additive model (GAM). Insurance and reinsurance firms, policymaking, planning, and decision-making agencies and coastal populations ultimately benefit from the modelling of typhoon activity. Kernel density estimation was employed to analyse the distribution of typhoon genesis using a 50-year dataset (1975–2024) of typhoon track measurements obtained from the International Best Track Archive Climate Stewardship. The tracks are replicated by constructing a GAM and deep Q-learning model. At a 95% confidence level, the GAM (about 73%) and deep Q-learning model demonstrate significant competency (around 76%) in a distance computation approach that compares observed and simulated landfall trajectories.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Modelling of typhoon activities over the Western North Pacific using a generalised additive model and deep Q-learning model

  • 期刊论文

  • 发表日期：

    2025年11月20日

查看全文
  • Md Wahiduzzaman,
  • Xiang Wang

摘要

This study presents a statistical and deep Q-learning model framework for predicting typhoon genesis and trajectories, utilising kernel density estimation and a generalised additive model (GAM). Insurance and reinsurance firms, policymaking, planning, and decision-making agencies and coastal populations ultimately benefit from the modelling of typhoon activity. Kernel density estimation was employed to analyse the distribution of typhoon genesis using a 50-year dataset (1975–2024) of typhoon track measurements obtained from the International Best Track Archive Climate Stewardship. The tracks are replicated by constructing a GAM and deep Q-learning model. At a 95% confidence level, the GAM (about 73%) and deep Q-learning model demonstrate significant competency (around 76%) in a distance computation approach that compares observed and simulated landfall trajectories.