Scholarlink

<p>Mapping permafrost distribution is significant for comprehending the impacts of climatic alterations and providing baseline data for delineating permafrost-induced hazard potential areas in high mountainous regions in the Himalayas. This study investigates the spatial distribution of permafrost in the Garhwal Himalaya, Uttarakhand, by integrating topo-climatic variables and rock glacier inventories using three machine learning algorithms: Binary Logistic Regression, Random Forest, and Extreme Gradient Boosting (XGBoost). A total of 268 rock glaciers comprising 247 active and 21 relict forms were mapped using high-resolution imagery from Sentinel-2 and Google Earth. Logistic regression models were developed based on key predictor variables, including mean annual air temperature (MAAT), mean annual ground temperature (MAGT), land surface temperature (LST), snow cover duration, potential incoming solar radiation (PISR), and slope aspect. The Random Forest and XGBoost models incorporated extreme air temperature, elevation, and the mean temperature of the warmest quarter (MTOWQ) to better capture the spatial variability of permafrost. The logistic regression models (LRM-MAAT, LRM-MAGT, LRM-SC, and LRM-LST) demonstrated classification accuracies of 94.8%, 91.8%, 92.05%, and 91.4%, respectively. The Random Forest and XGBoost models outperformed the regression models, achieving testing accuracies of 97.6% and 97.0%, respectively. Validation using the Permafrost Zonation Index map showed strong similarity among all models with adequate accuracy indicating the applicability of these models for permafrost distribution modeling in the Himalayas. (Keywords: Garhwal Himalaya, Logistic Regression Model, Permafrost, Random Forest ML, Rock Glacier, XGBoost ML)</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

Permafrost distribution modeling using remote sensing and machine learning technique in the Garhwal Himalaya, India

  • 期刊论文

  • 发表日期：

    2025年12月27日

查看全文
  • Arvind Chandra Pandey,
  • Anikul Islam,
  • Chandra Shekhar Dwivedi,
  • Bikash Ranjan Parida,
  • Alexey Maslakov,
  • Ekaterina Koroleva

摘要

Mapping permafrost distribution is significant for comprehending the impacts of climatic alterations and providing baseline data for delineating permafrost-induced hazard potential areas in high mountainous regions in the Himalayas. This study investigates the spatial distribution of permafrost in the Garhwal Himalaya, Uttarakhand, by integrating topo-climatic variables and rock glacier inventories using three machine learning algorithms: Binary Logistic Regression, Random Forest, and Extreme Gradient Boosting (XGBoost). A total of 268 rock glaciers comprising 247 active and 21 relict forms were mapped using high-resolution imagery from Sentinel-2 and Google Earth. Logistic regression models were developed based on key predictor variables, including mean annual air temperature (MAAT), mean annual ground temperature (MAGT), land surface temperature (LST), snow cover duration, potential incoming solar radiation (PISR), and slope aspect. The Random Forest and XGBoost models incorporated extreme air temperature, elevation, and the mean temperature of the warmest quarter (MTOWQ) to better capture the spatial variability of permafrost. The logistic regression models (LRM-MAAT, LRM-MAGT, LRM-SC, and LRM-LST) demonstrated classification accuracies of 94.8%, 91.8%, 92.05%, and 91.4%, respectively. The Random Forest and XGBoost models outperformed the regression models, achieving testing accuracies of 97.6% and 97.0%, respectively. Validation using the Permafrost Zonation Index map showed strong similarity among all models with adequate accuracy indicating the applicability of these models for permafrost distribution modeling in the Himalayas. (Keywords: Garhwal Himalaya, Logistic Regression Model, Permafrost, Random Forest ML, Rock Glacier, XGBoost ML)