Extraction of glacial lakes from high mountains of Karakorum using SAR data and machine learning algorithms The Hunza Basin Case study

The sentinel-1 mission is expected to deliver a wealth of data and images.As the first memeber of the sequence of two satellites. The Sentinel-1A with C- band was lunched in 3,Aprial, 2014.It has dual polrization capacity (VV+VH) which can provide much more ground surface information.This

2025-06-28 16:32:29 - Adil Khan

Project Title

Extraction of glacial lakes from high mountains of Karakorum using SAR data and machine learning algorithms The Hunza Basin Case study

Project Area of Specialization Artificial IntelligenceProject Summary

The sentinel-1 mission is expected to deliver a wealth of data and images.As the first memeber of the sequence of two satellites. The Sentinel-1A with C- band was lunched in 3,Aprial, 2014.It has dual polrization capacity (VV+VH) which can provide much more ground surface information.This study aims to analyse sentinel-1 Synthetic Aperture Radar of Ground Range Detected (SAR)- GRD data for extraction of glacial lakes over high mountains of Karokoram.

In Asia, high altitude mountain range is Karakoram.This mountain range has  key components that are glacial lakes. As a result of global warming they endanger aquatic life, natural ecosystem and public infrastructure in a short space of time through outburst floods. While most attempt were made to obtain glacial lake outlines and identify their modifications via remote sensing imagery, the temporal and spatial resolution of glacial lake data set are typically not sufficient fine to represent specific glacial lake  dynamics functions, particularly for highly hazardous glacial lakes of higher frequency heterogeneity. Through using fulltime-series sentinel-1 A  synthetic aperture radar (SAR) images for a year (2018) and machine learning algorithms, our  research introduce a comparative approach for extracting outlines of the glacial lakes that  show a rapid variations in every year or in ablation period. Our proposed approach was centered on a level set segmentation together with a median pixel classification of synthetic aperture radar (SAR) backscattering coefficients mounted as a regularization interval in attempt to rigorously determine the lake extant over the time range observed. However, this study findings showes that the Ground Range Detected (GRD) SAR data is very useful in every weather condition it may be cloudy and day/night. On other hand, the comparative analysis of  machine learning algorithms and Geographic Information Techniques (GIS)  that are backscatter analysis  indicates that machine learning algorithms are more strong and successful techinques for extraction of glacial lakes or water bodies from high mountains of  Karakoram  in contrast with manually digitization technique and backscatter analysis.

Project Objectives Project Implementation Method

Over all research implementation method shown by methodology steps in simple and easy way:

  1. extraction of water body
  2. data acquistion
  3. sentinel-1 images(10m), Vector data(shape files of study area,water sheds etc ) and DEM-ASTER(30m)
  4. Re-sampling of sentinel-1 and DEM 
  5. Clip study area
  6. Data Pre-processing(Terrain correction and speckle filtering
  7. Data post-processing(Backscatter analysis(Mean backscatter,standard deviation backscatter, maximum-minimum ratio and saturation)
  8. glacial lake detection
  9. machine learning alorithms
  10. Accurarcy assessment
  11. validation (using google earth manual digitization)
  12. Final water bodies

Benefits of the Project Technical Details of Final Deliverable

It will deliver soon.

Final Deliverable of the Project Software SystemCore Industry EducationOther 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) 60000
camera Equipment23000060000

More Posts