Scholarlink

We investigate the effectiveness of Machine Learning (ML) as a complementary tool in enhancing the accuracy of precipitation forecasting in the Calabria region (southern Italy), a Mediterranean area frequently affected by extreme rainfall events, with significant material damage and potential loss of human lives. The operational weather forecasting system developed at the University of Calabria, based on the Weather Research and Forecasting (WRF) model, is considered to this end. We test two ML models, namely an artificial neural network and a random forest, to improve the accuracy of WRF forecasting. We provide several WRF output variables to the ML models. After a proper calibration (learning and test phases) over a two-year-long extended dataset, we assess the results to the original WRF outcomes (validation phase). Preliminary experiments resulted in a significant improvement by the WRF-ML coupled systems.

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

Applying Machine Learning to Improve Weather Forecasting in a Mediterranean Area: Preliminary Experiments

  • 会议论文

  • 发表日期：

    2025年

查看全文
  • Umair Yousaf,
  • Alessio De Rango,
  • Luca Furnari,
  • Donato D’Ambrosio,
  • Alfonso Senatore

摘要

We investigate the effectiveness of Machine Learning (ML) as a complementary tool in enhancing the accuracy of precipitation forecasting in the Calabria region (southern Italy), a Mediterranean area frequently affected by extreme rainfall events, with significant material damage and potential loss of human lives. The operational weather forecasting system developed at the University of Calabria, based on the Weather Research and Forecasting (WRF) model, is considered to this end. We test two ML models, namely an artificial neural network and a random forest, to improve the accuracy of WRF forecasting. We provide several WRF output variables to the ML models. After a proper calibration (learning and test phases) over a two-year-long extended dataset, we assess the results to the original WRF outcomes (validation phase). Preliminary experiments resulted in a significant improvement by the WRF-ML coupled systems.