Scholarlink

<p>Urban green infrastructure plays a crucial role in sustainable city planning by enhancing permeability, mitigating urban heat islands, and improving stormwater management. This study assesses the potential of existing green infrastructure within the Kolkata Municipal Corporation with a special emphasis on estimating green and permeable surfaces using an integrated remote sensing-machine learning approach. The research identifies and quantifies permeable surfaces, including grass/low vegetation fields, barren fields, semi-impervious low to moderate vegetation areas, urban forests, and tree pits. High-resolution satellite imagery from SuperDove PSB SD (3-m resolution) is utilized for accurate identification and classification of green and permeable surfaces. The best model for assessing green infrastructure, permeable surfaces, and impermeable surfaces was selected through a comparative analysis of three methods: AutoML, XGBoost, and Random Forest–refined Linear Spectral Unmixing (RF-refined LSU). After classification, the areas of permeable and impermeable surfaces, as well as the overall green infrastructure in the study area, were quantified. The ratio between permeable surface and impermeable surface and ward-wise statistics were also computed. The results indicate that the AutoML method outperforms the others. The cumulative sum of areas labelled as Green Infrastructure (GI) was estimated at 66.52 km<sup>2</sup>, whilst the total permeable surface area was estimated to be approximately 71.28 km<sup>2</sup>. The largest area of GI is found in Ward 108 with 8.39 km<sup>2</sup>. It is essential that policymakers, urban planners, and residents act now to mitigate these environmental challenges, otherwise water-logging and urban flooding will continue to occur with more frequency and severity.</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

Mapping green and permeable surfaces in Kolkata municipal corporation using high-resolution multispectral imagery and comparative machine learning

  • 期刊论文

  • 出版模式：

    开放获取

  • 发表日期：

    2026年5月20日

查看全文
  • Suman Chatterjee

摘要

Urban green infrastructure plays a crucial role in sustainable city planning by enhancing permeability, mitigating urban heat islands, and improving stormwater management. This study assesses the potential of existing green infrastructure within the Kolkata Municipal Corporation with a special emphasis on estimating green and permeable surfaces using an integrated remote sensing-machine learning approach. The research identifies and quantifies permeable surfaces, including grass/low vegetation fields, barren fields, semi-impervious low to moderate vegetation areas, urban forests, and tree pits. High-resolution satellite imagery from SuperDove PSB SD (3-m resolution) is utilized for accurate identification and classification of green and permeable surfaces. The best model for assessing green infrastructure, permeable surfaces, and impermeable surfaces was selected through a comparative analysis of three methods: AutoML, XGBoost, and Random Forest–refined Linear Spectral Unmixing (RF-refined LSU). After classification, the areas of permeable and impermeable surfaces, as well as the overall green infrastructure in the study area, were quantified. The ratio between permeable surface and impermeable surface and ward-wise statistics were also computed. The results indicate that the AutoML method outperforms the others. The cumulative sum of areas labelled as Green Infrastructure (GI) was estimated at 66.52 km2, whilst the total permeable surface area was estimated to be approximately 71.28 km2. The largest area of GI is found in Ward 108 with 8.39 km2. It is essential that policymakers, urban planners, and residents act now to mitigate these environmental challenges, otherwise water-logging and urban flooding will continue to occur with more frequency and severity.