Scholarlink

This study presents an integrated approach to drought assessment in Bangladesh by combining mathematical evapotranspiration modeling with unsupervised machine learning. Daily satellite-driven environmental data from NASA POWER (2014–2024) were used to compute reference evapotranspiration ( \(ET_0\) ) via the FAO-56 Penman-Monteith equation across all 64 districts. K-Means clustering identified three distinct drought regimes, strongly correlated with \(ET_0\) patterns and climatic variables. Results demonstrate that this method effectively distinguishes regions by drought severity, offering improved interpretability and spatial resolution over traditional precipitation-based indices. The findings support the development of more robust early warning and water management systems in climate-vulnerable regions like Bangladesh.

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

Enhanced Unsupervised Machine Learning Analysis on Evapotranspiration and Drought Based on Satellite-Driven Data

  • 会议论文

  • 发表日期：

    2026年

查看全文
  • Pritam Sarkar,
  • Tonmoy Paul

摘要

This study presents an integrated approach to drought assessment in Bangladesh by combining mathematical evapotranspiration modeling with unsupervised machine learning. Daily satellite-driven environmental data from NASA POWER (2014–2024) were used to compute reference evapotranspiration ( \(ET_0\) ) via the FAO-56 Penman-Monteith equation across all 64 districts. K-Means clustering identified three distinct drought regimes, strongly correlated with \(ET_0\) patterns and climatic variables. Results demonstrate that this method effectively distinguishes regions by drought severity, offering improved interpretability and spatial resolution over traditional precipitation-based indices. The findings support the development of more robust early warning and water management systems in climate-vulnerable regions like Bangladesh.