Scholarlink

<p>Topographic complexity and illumination gradients in the Himalayas frequently cause systematic underestimation in snow products. This study evaluates a 25-year Landsat archive (2000–2024) of the Miyar basin to compare standard fixed-threshold indices Standard Normalized Difference Snow Index (NDSI) (STD-NDSI), Double Threshold Sentinel-2 (DT-S2), Double Threshold Moderate Resolution Imaging Spectroradiometer (MODIS) (DT-MOD) against a terrain optimized workflow-corrected C-corrected NDSI (C-NDSI) and C-corrected Dynamic Threshold NDSI (C-DYN-NDSI). By integrating band specific C-correction with Otsu based scene adaptive thresholding, the C-DYN-NDSI method reclaims spectral signals suppressed by terrain shadows, yielding an annual mean Snow Covered Area (SCA) of 483 sq km, a 27.6% increase relative to fixed threshold approaches. This divergence is most acute during the ablation season where shadow recovery identifies more snow than static products. While all methods exhibit high temporal correlation (<i>r</i> &gt; 0.94), fixed thresholds introduce a systemic negative bias that masks the true cryospheric extent. It is important to note that these findings are based on methodological intercomparison and have not been validated against independent ground-truth observations. The results highlight the importance of terrain-aware processing in reducing bias in optical snow mapping and suggest that such frameworks are potentially transferable across optical sensors, although this study is limited to Landsat data.</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

A Terrain-Corrected Dynamic NDSI Framework Integrating C-Correction and Adaptive Thresholding for Improved Snow Cover Mapping in the Himalayas

  • 期刊论文

  • 发表日期：

    2026年7月22日

查看全文
  • Deva Jefflin Aruldhas,
  • Chandre Gowda Cheluvegowda,
  • Geetha Priya Murugesan,
  • Charu Prabha Rajasekaran Palanikumar,
  • Sushil Kumar Singh,
  • Anil V. Kulkarni

摘要

Topographic complexity and illumination gradients in the Himalayas frequently cause systematic underestimation in snow products. This study evaluates a 25-year Landsat archive (2000–2024) of the Miyar basin to compare standard fixed-threshold indices Standard Normalized Difference Snow Index (NDSI) (STD-NDSI), Double Threshold Sentinel-2 (DT-S2), Double Threshold Moderate Resolution Imaging Spectroradiometer (MODIS) (DT-MOD) against a terrain optimized workflow-corrected C-corrected NDSI (C-NDSI) and C-corrected Dynamic Threshold NDSI (C-DYN-NDSI). By integrating band specific C-correction with Otsu based scene adaptive thresholding, the C-DYN-NDSI method reclaims spectral signals suppressed by terrain shadows, yielding an annual mean Snow Covered Area (SCA) of 483 sq km, a 27.6% increase relative to fixed threshold approaches. This divergence is most acute during the ablation season where shadow recovery identifies more snow than static products. While all methods exhibit high temporal correlation (r > 0.94), fixed thresholds introduce a systemic negative bias that masks the true cryospheric extent. It is important to note that these findings are based on methodological intercomparison and have not been validated against independent ground-truth observations. The results highlight the importance of terrain-aware processing in reducing bias in optical snow mapping and suggest that such frameworks are potentially transferable across optical sensors, although this study is limited to Landsat data.