Scholarlink

<p>A deep robust domain adaptation (DRDA) model for heterogeneous weighted information fusion (HWIF), called DRDAHWIF-gram, is proposed to impute long-term missing meteorological data for cross-domain intelligent forecasting. This study’s core contribution lies in assessing forest climate change risks through explainable machine learning, which leverages a multi-criteria computational performance metric to establish a smart early warning index (SEWI). Two real-world case studies, utilizing the Hyrcanian forest dataset and the comparative climatic data, validate the model’s ability to capture predictable meteorological patterns, including temperature, humidity, solar radiation, and wind speed trends, across diverse climatological domains. Evaluated across multiple forecast horizons, the developed model outperforms conventional methods, demonstrating higher accuracy and reliability in weather variation monitoring. The high SEWI confirms the stability of meteorological domains, supporting the deployment of intelligent climatic hazard early warning systems. This approach advances expert forecasting for meteorological threats under forest climate change.</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

An interpretable information fusion approach to impute meteorological missing values toward cross-domain intelligent forecasting

  • 期刊论文

  • 发表日期：

    2025年8月22日

查看全文
  • Milad Zarchi,
  • Zohreh Hashemi Aslani,
  • Kong Fah Tee

摘要

A deep robust domain adaptation (DRDA) model for heterogeneous weighted information fusion (HWIF), called DRDAHWIF-gram, is proposed to impute long-term missing meteorological data for cross-domain intelligent forecasting. This study’s core contribution lies in assessing forest climate change risks through explainable machine learning, which leverages a multi-criteria computational performance metric to establish a smart early warning index (SEWI). Two real-world case studies, utilizing the Hyrcanian forest dataset and the comparative climatic data, validate the model’s ability to capture predictable meteorological patterns, including temperature, humidity, solar radiation, and wind speed trends, across diverse climatological domains. Evaluated across multiple forecast horizons, the developed model outperforms conventional methods, demonstrating higher accuracy and reliability in weather variation monitoring. The high SEWI confirms the stability of meteorological domains, supporting the deployment of intelligent climatic hazard early warning systems. This approach advances expert forecasting for meteorological threats under forest climate change.