Related Experiment Video
Updated: Jun 10, 2025
Author Spotlight: Efficient Image Recognition Using Directional Gradient Histogram Technique and Support Vector Machines
Published on: January 5, 2024
Effective identification of debris-covered glaciers in Western China using multiple machine-learning algorithms
Rui He1, Donghui Shangguan2, Qiudong Zhao1
1Key Laboratory of Ecological Safety and Sustainable Development in Arid Lands, Northwest Institute of Eco-Environment and Resources, Chinese Academy of Sciences, Lanzhou 730000, PR China; State Key Laboratory of Cryospheric Science and Frozen Soil Engineering, Northwest Institute of Eco-Environment and Resources, Chinese Academy of Sciences, Lanzhou 730000, PR China; University of Chinese Academy of Sciences, Beijing 100049, PR China.
This study introduces a robust machine learning approach for automatically identifying debris-covered glaciers (DCGs) using environmental variables. The Random Forests algorithm demonstrated superior performance, enhancing glacier inventory accuracy.
Area of Science:
- Glaciology
- Remote Sensing
- Machine Learning
Background:
- Debris-covered glaciers (DCGs) are significant globally but challenging to identify automatically.
- Variability in environmental conditions and glacier types complicates automated mapping.
Purpose of the Study:
- To develop and validate a machine learning framework for accurate, automated identification of DCGs.
- To compare the performance of four machine learning algorithms for DCG classification.
Main Methods:
- Utilized Random Forests (RF), XGBoost, LightGBM, and SVM algorithms.
- Integrated surface reflectance, normalized indices, surface temperature, and topography data.
- Employed a three-process approach for distinguishing ice/snow, ice/debris, and debris/land features.
Main Results:
- Random Forests (RF) algorithm showed superior performance, achieving a high average Matthews correlation coefficient (MCC) of 0.819 for DCG identification.
- Regional modeling improved accuracy, especially for continental glaciers.
- Independent binary-class models outperformed integrated multi-class models.
Conclusions:
- The proposed machine learning method offers a reliable and generalizable tool for automated DCG identification.
- This approach significantly aids in compiling and monitoring glacier inventories.
- The method shows high consistency with existing glacier inventories, despite some uncertainties in maritime regions.
