Machine-Learning Algorithms for Mapping Debris-Covered Glaciers: The Hunza shishper Case Study
As global warming is one of the most significant challenges of recent times. The glaciers melt faster than expected resulting in a rise in global mean sea level and increased flood risk. The advent of modern remote sensing technologies allowed images to be obtained more often than ever before.
2025-06-28 16:34:04 - Adil Khan
Machine-Learning Algorithms for Mapping Debris-Covered Glaciers: The Hunza shishper Case Study
Project Area of Specialization Artificial IntelligenceProject Summary Project Summary As global warming is one of the most significant challenges of recent times. The glaciers melt faster than expected resulting in a rise in global mean sea level and increased flood risk. The advent of modern remote sensing technologies allowed images to be obtained more often than ever before. On the other hand, the development of high_performance computing equipment and proces ng techniques has allowed a cost_effective solution to track temporary changes in glaciers on a large scale. In this analysis, supervised machine learning methods are investigated using texture, topographical, and spectral data to automatically identify glacier coverings from multi-temporal Sentinel-2 imagery. Three most widely used techniques of supervised machine learning were investigated: vector supporting machine (SVM), artificial neural network (ANN) and random forest (RF). The proposed approach was used on data obtained from Shishper watershed in Hunza Basin, situated in husanabad Hunza. This considered three major classes: glaciers, debris-covered glaciers, and no glaciated areas. The data was split into training (70%) and evaluation datasets (30%). Eventually, an areabased accuracy evaluation was performed by comparing the results obtained with the reference data for each classifier. Experiments showed that the results obtained for all classifiers were highly accurate and visually more compatible with the portrayal of glacier cover types. For all tests, random forest performed the best on all three groups compared to ANN and SVM. The high classification accuracy obtained from distinguished debris-covered glaciers using our method will be useful in determining the actual water resources available which can be further useful for the management of hazards and water resources.
- The main objective of this paper is to evaluate the efficiency of the ML algorithms for their classification of glaciers especially glacier floods.
- To establish and understand the dynamics of glaciers sensitivity to climate change through remote sensed data from satellite.
We will implement our method in three steps.
- In the first step a set of features will be extracted, which include our required band values , spectral, textural features. The spectral features include reflectance information from each band, Normalized Difference Vegetation Index (NDVI), Normalized Difference Snow Index (NDSI), Normalized Difference Water Index (NDWI), and New Band Ratio (NBR),Nbareness and moisture index. The textural features include mean, variance, homogeneity, contrast, dissimilarity, entropy, energy, correlation and ASM.
- In the second step, we will train and test three widely used machine learning classifiers (SVM, RF, and ANN) on our data. In each classifier, the feature vectors obtained in the previous step were fed which generated classification maps for each class, i.e. glaciers, debris-covered glaciers and non-glacier areas
- In the final step, an area-based accuracy assessment will be carried out by comparing the output generated with the reference data to assess the performance of the proposed method for each classifier.
Benefits of the project
- The advantages are that we can distinguish glacier surfaces covered with and without debris using ML approaches, as well as using remote sensing data.
- The effectiveness and accuracy of the machine learning algorithms can also be compared.
- we can effectively map glaciera and debris covered glaciers.
A methodology which will be highly accurate and precise to detect debris and glaciers' floods.
Final Deliverable of the Project Software SystemCore Industry ITOther Industries IT Core Technology Artificial Intelligence(AI)Other Technologies Artificial Intelligence(AI)Sustainable Development Goals Climate ActionRequired Resources| Item Name | Type | No. of Units | Per Unit Cost (in Rs) | Total (in Rs) |
|---|---|---|---|---|
| Total in (Rs) | 50000 | |||
| camera | Equipment | 1 | 50000 | 50000 |