Scholarlink

<p>Rainfall forecasting plays a crucial role in both urban and rural planning and has become increasingly challenging with vast changes in global weather patterns and climate conditions. This study focuses on rainfall forecasting in specific districts of Tamil Nadu within the Amaravathi basin, where accurate predictions are essential for agricultural planning. Data was collected for three specific districts—Erode, Karur, and Dindigul—from 2000 to 2023. This study develops an ensemble modeling framework for rainfall prediction in the Amaravathi basin, Tamil Nadu, addressing limitations of traditional meteorological forecasting approaches.Several regression models, including Random Forest Regression, Gradient Boosting, k-Nearest Neighbors, AutoRegressive Integrated Moving Average, Long Short-Term Memory, and Exponential Smoothing, were evaluated to identify the most effective models for rainfall forecasting in the region. Random Forest and Gradient Boosting were the best performers, showing consistently high prediction accuracy. Gradient Boosting scored an extremely high R-squared value of 0.9955, accounting for close to 99.55% of rainfall variance, with low Mean Absolute Error (MAE), Mean Squared Error (MSE), and Root Mean Square Error (RMSE) scores while also being computationally efficient (30% time reduction compared to Random Forest). Random Forest achieved similar results, with an R-squared of 0.9903 and competitive MAE and RMSE values. The findings highlight the need for data-driven, adaptive methods of rainfall forecasting in semi-arid areas where conventional techniques tend to be inadequate because of non-stationarity and spatial heterogeneity in rainfall patterns. The results highlight the need to include a combination of methods in order to extract intrinsic characteristics pertaining to hydrometeorological relationships, thus enhancing the ability to forecast rainfall accurately.</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

Rainfall Forecasting Model for Amaravathi Basin Using Machine Learning Approach

  • 期刊论文

  • 发表日期：

    2025年8月12日

查看全文
  • R. Nirranjana,
  • R. Aishwarya,
  • S. Tejshree,
  • K. S. Gayathri,
  • Surendar Natarajan,
  • P. Thirunavukkarasu

摘要

Rainfall forecasting plays a crucial role in both urban and rural planning and has become increasingly challenging with vast changes in global weather patterns and climate conditions. This study focuses on rainfall forecasting in specific districts of Tamil Nadu within the Amaravathi basin, where accurate predictions are essential for agricultural planning. Data was collected for three specific districts—Erode, Karur, and Dindigul—from 2000 to 2023. This study develops an ensemble modeling framework for rainfall prediction in the Amaravathi basin, Tamil Nadu, addressing limitations of traditional meteorological forecasting approaches.Several regression models, including Random Forest Regression, Gradient Boosting, k-Nearest Neighbors, AutoRegressive Integrated Moving Average, Long Short-Term Memory, and Exponential Smoothing, were evaluated to identify the most effective models for rainfall forecasting in the region. Random Forest and Gradient Boosting were the best performers, showing consistently high prediction accuracy. Gradient Boosting scored an extremely high R-squared value of 0.9955, accounting for close to 99.55% of rainfall variance, with low Mean Absolute Error (MAE), Mean Squared Error (MSE), and Root Mean Square Error (RMSE) scores while also being computationally efficient (30% time reduction compared to Random Forest). Random Forest achieved similar results, with an R-squared of 0.9903 and competitive MAE and RMSE values. The findings highlight the need for data-driven, adaptive methods of rainfall forecasting in semi-arid areas where conventional techniques tend to be inadequate because of non-stationarity and spatial heterogeneity in rainfall patterns. The results highlight the need to include a combination of methods in order to extract intrinsic characteristics pertaining to hydrometeorological relationships, thus enhancing the ability to forecast rainfall accurately.