Scholarlink

<p>This paper presents a machine learning–based framework for forecasting southwest Indian summer monsoon rainfall (ISMR). Ten climate predictors are preprocessed to remove anomalies, and rainfall is modeled using support vector regression (SVR), multiple linear regression (MLR), and ridge regression within an n-length sliding window approach. The analysis covers ISMR deviations from the seasonal mean during 1991–2019 across 36 meteorological subdivisions, seven homogeneous regions, and the national level. Results show that SVR and MLR models outperform ridge regression, achieving Pearson correlation coefficients up to 0.82 at the national level and 0.83 in regional analysis, with performance improvements of 5.7–12.7%. The proposed methodology demonstrates the effectiveness of integrating multiple climate parameters with sliding window strategies, contributing to more accurate regional and national monsoon forecasts and supporting agricultural planning and disaster preparedness.</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

Forecasting of Southwest Indian Summer Monsoon Rainfall Using Machine Learning Technique

  • 期刊论文

  • 发表日期：

    2026年3月30日

查看全文
  • Aman Kumar,
  • Samayveer Singh,
  • Aruna Malik,
  • Arun Kumar Rai

摘要

This paper presents a machine learning–based framework for forecasting southwest Indian summer monsoon rainfall (ISMR). Ten climate predictors are preprocessed to remove anomalies, and rainfall is modeled using support vector regression (SVR), multiple linear regression (MLR), and ridge regression within an n-length sliding window approach. The analysis covers ISMR deviations from the seasonal mean during 1991–2019 across 36 meteorological subdivisions, seven homogeneous regions, and the national level. Results show that SVR and MLR models outperform ridge regression, achieving Pearson correlation coefficients up to 0.82 at the national level and 0.83 in regional analysis, with performance improvements of 5.7–12.7%. The proposed methodology demonstrates the effectiveness of integrating multiple climate parameters with sliding window strategies, contributing to more accurate regional and national monsoon forecasts and supporting agricultural planning and disaster preparedness.