Training on Application of Machine Learning and Deep Learning in Mountain Cryosphere Research

About the training

The International Centre for Integrated Mountain Development in collaboration with College of Natural Resources, Royal University of Bhutan is organising a training on Machine learning (ML) and Deep Learning applications in the cryosphere studies.
The training program will introduce participants to the use of ML and DL tools for analysing glaciers, permafrost, snow and glacial lakes. Participants will learn how to prepare data, training models and apply these techniques using open datasets and computational tools to support cryosphere research and climate change assessments.

Training objectives

To build foundational knowledge on the state-of-the-art machine learning and deep learning models and its applications in mountain cryosphere studies.
To provide hands-on experience in applying ML model (e.g. scikit learn) and DL (e.g. PyTorch, DeepLabv3+) to mountain cryosphere datasets.
To foster collaboration and project-based learning among participants.

Expected participants

The training is designed for early-career researchers, graduate students, and professionals working in the fields of mountain cryosphere science, geoinformatics, environmental science, and climate change studies. Around 20 professionals from partners, academic and research institutions will be invited to attend the training.

Eligibility

Applicants with comprehensive understanding of mountain cryosphere dynamics and processes.

Applicants are familiar with geospatial tools and remote sensing techniques for high-altitude data acquisition.

The training is organised by ICIMOD’s Cryosphere and Water under Climate and Environment Risks, supported by Government of Norway and Swiss agency for development and cooperation.

Training structure

The program is structured over six days, with each day focusing on a specific aspect of mountain cryosphere research and machine learning and deep learning:

Day	Topic	Key Activities
1	Introduction to mountain cryosphere and geospatial tools (crash-course)	– Overview of mountain cryosphere components (glaciers, rock glaciers, permafrost, snow cover) – Climate change impacts on high-altitude frozen environments – Introduction to geospatial data and tools (GIS, remote sensing) – Overview of open mountain cryosphere datasets (Sentinel, Landsat, ASTER, MODIS) – Introduction to Google Earth Engine (GEE) for mountain cryosphere data analysis
2	Fundamentals of machine learning (ML) for mountain cryosphere	– Introduction to machine learning concepts (supervised, unsupervised learning) – Key algorithms for mountain cryosphere data (classification, regression, clustering, decision trees) – Data preprocessing and feature engineering for mountain cryosphere applications
3	Fundamentals of Deep Learning (DL) Setting up the computational environments	– Convolution Neural Network Architecture and Frameworks – Transfer learning, Vision Transformers – Configuring a Linux workstation using WSL in Windows – Command line basics for mountain cryosphere data processing – Setting up Python programming environment (Miniconda3, Jupyter Notebooks) – Python crash course: numpy, pandas, matplotlib, scikit-learn (ML), PyTorch (DL) – Practical: Data manipulation and visualization in Python
4	Hands-on machine learning for mountain cryosphere	– Integrating mountain cryosphere data with ML/DL models (scikit-learn (ML), PyTorch (DL), DeepLabv3+) – Case studies: glacier mass balance, glacial lake mapping, permafrost distribution modeling, snow cover analysis – Practical: End-to-end ML/DL project using GEE and mountain cryosphere datasets
5	Group projects	– Participants form groups to work on selected mountain cryosphere-related ML/DL projects – Mentor guidance and peer collaboration – Preparation of project presentations
6	Project presentation and closing	– Group presentations and feedback sessions – including certificate distribution

Resource persons

Internal resource person: Sonam Wangchuk (PhD), ICIMOD
External resource person: China/India
Technical Training Support: Sunwi Maskey, ICIMOD

Expected outcomes

Practical skills in geospatial data processing, machine learning and deep learning in cryosphere applications.
Collaborative project experience.
Networking opportunities with peers and experts.

Conclusion

This training program is a unique opportunity to bridge the gap between mountain cryosphere science and cutting-edge data analytics. It aims to empower researchers to tackle pressing environmental challenges in high-altitude regions using innovative data-driven approaches.

Application of Machine Learning and Deep Learning in Mountain Cryosphere Research

Programmes

Venue

Date & Time

Contact

About the training

Training objectives

Expected participants

Eligibility

Training structure

Resource persons

Expected outcomes

Conclusion

Application of Machine Learning and Deep Learning in Mountain Cryosphere Research

Programmes

Venue

Date & Time

Contact

About the training

Training objectives

Expected participants

Eligibility

Training structure

Resource persons

Expected outcomes

Conclusion

Gender equality & social inclusion (GESI) reflections and best practices in cryosphere science from the Arctic and Hindu Kush Himalayan region

Developing a National Clean Air Action Plan for Bhutan

Promoting gender and social inclusive crop residue supply chain to reduce open burning in Nepal