Scholarlink

<p>The open dumping of waste poses severe environmental and public health hazards when exposed to the atmosphere, yet systematic identification of dumpsites remains challenging for environmental agencies due to data accessibility constraints. This study presents an integrated geospatial framework combining Random Forest (RF) classification and logistic regression analysis using Sentinel-2 and Landsat-8 imagery to detect and characterize waste disposal sites across Madhya Pradesh, India. The RF algorithm achieved 86.49% overall accuracy in automated dumpsite detection, successfully identifying 199 sites, including 60 previously undocumented locations. Spectral-thermal analysis of 36 validated sites revealed distinctive signatures: Land Surface Temperature (LST) ranging 35.47–39.58&#xa0;°C (mean: 37.30 ± 0.85&#xa0;°C), Normalized Difference Vegetation Index (NDVI) of 0.04–0.25 (mean: 0.1539 ± 0.0378), and Normalized Difference Built-up Index (NDBI) of -0.06 to 0.12 (mean: 0.0282 ± 0.0511). Logistic regression modelling demonstrated LST’s positive influence on dumpsite classification (88% accuracy), while NDVI exhibited a negative association, reflecting reduced vegetation cover at waste sites. Correlation analysis revealed significant inverse relationships between NDBI-NDVI and NDVI-LST. Multi-criteria threshold classification (NDBI &gt; 0.028, NDVI &lt; 0.15, LST &gt; 37.0&#xa0;°C) effectively delineated high-risk zones requiring prioritized monitoring. This scalable methodology provides a cost-effective alternative to ground surveys, enabling proactive waste management interventions and environmental protection strategies. However, it lags in analyzing the morphological and compositional attributes of such areas. The further studies will focus on integrating the field investigations with high-resolution remote sensing data to assess characteristics of the identified dumpsite locations.</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

Ground truthing of dumpsites using remote sensing and machine learning approaches in peri-urban settings

  • 期刊论文

  • 发表日期：

    2026年5月18日

查看全文
  • Veena N Bhajantri,
  • Ashootosh Mandpe

摘要

The open dumping of waste poses severe environmental and public health hazards when exposed to the atmosphere, yet systematic identification of dumpsites remains challenging for environmental agencies due to data accessibility constraints. This study presents an integrated geospatial framework combining Random Forest (RF) classification and logistic regression analysis using Sentinel-2 and Landsat-8 imagery to detect and characterize waste disposal sites across Madhya Pradesh, India. The RF algorithm achieved 86.49% overall accuracy in automated dumpsite detection, successfully identifying 199 sites, including 60 previously undocumented locations. Spectral-thermal analysis of 36 validated sites revealed distinctive signatures: Land Surface Temperature (LST) ranging 35.47–39.58 °C (mean: 37.30 ± 0.85 °C), Normalized Difference Vegetation Index (NDVI) of 0.04–0.25 (mean: 0.1539 ± 0.0378), and Normalized Difference Built-up Index (NDBI) of -0.06 to 0.12 (mean: 0.0282 ± 0.0511). Logistic regression modelling demonstrated LST’s positive influence on dumpsite classification (88% accuracy), while NDVI exhibited a negative association, reflecting reduced vegetation cover at waste sites. Correlation analysis revealed significant inverse relationships between NDBI-NDVI and NDVI-LST. Multi-criteria threshold classification (NDBI > 0.028, NDVI < 0.15, LST > 37.0 °C) effectively delineated high-risk zones requiring prioritized monitoring. This scalable methodology provides a cost-effective alternative to ground surveys, enabling proactive waste management interventions and environmental protection strategies. However, it lags in analyzing the morphological and compositional attributes of such areas. The further studies will focus on integrating the field investigations with high-resolution remote sensing data to assess characteristics of the identified dumpsite locations.