Hands-on training session at LPS
On this page you’ll find links to the training material for the CoMet tutorial held at ESA’s Living Planet Symposium 2025. The session covers key concepts around uncertainties, error correlation and how to handle these with the CoMet toolkit. There will first be a presentation with some theoretical background, and an introduction to the CoMet toolkit, followed by 3 guided exercises using the CoMet toolkit. Participants will:
- Gain a conceptual overview of uncertainties in Earth Observation data processing.
- Learn how to use the CoMet tools in practical workflows.
- Apply methods through interactive notebooks hosted on Google Colab.
Exercises
Links and descriptions to the three exercises for this training session, hosted on google colab.
Exercise 1: Introduction to Punpy Capabilities
• Get familiar with the punpy tool
• Propagate uncertainties on manually provided input data through a simple measurement functions using punpy
• Explore the various ways uncertainties with different error correlations can be propagated
Exercise 2: Multi-Dimension Datasets
• Store error-correlation information for multi-dimensional measurement datasets using obsarray
• Practice on a multi-dimensional Earth Observation dataset example
• Propagate uncertainties from these datasets through measurement functions using punpy
Exercise 3: HYPERNETS Use Case
• Get familiar with a sample EO data (HYPERNETS)
• Use the previous exercises to add uncertainties to the HYPERNETS data processing chain
• Generate and interpret uncertainty-aware outputs