Challenge 23 - COMPEXT: Tracking Compound Extreme Events for Energy Security in a Changing Climate

[DiegoDomenici]

Stream 2 - Machine Learning

Mentors

• Alberto Troccoli
• Penny Boorman
• Kristian Nielsen
• Letizia Lusito
• Stefano Cordeddu
• Christoph Rudiger
• Gianpaolo Balsamo
• Joaquin Munoz-Sabater

Skill Required

• Strong Python programming skills
• Experience with ML frameworks (e.g. PyTorch)
• Proficiency in handling large meteorological datasets
• Understanding of environmental science principles
• UI design experience beneficial
• Background in environmental sciences or related fields
• Interest in renewable energy systems
• Passion for applying ML to environmental challenges

Goal

Design a Machine Learning-based CEI (Compound Event Index) that:
• Identifies co-occurring climate extremes that threaten renewable energy production (e.g. low wind and solar output, heatwaves, drought).
• Uses a combination of climate data (e.g. ERA5, nextGEMS) and ideally also energy indicators (e.g. wind/solar capacity factors, demand indicators from e.g. the C3S Energy service).
• Offers a flexible and scalable framework for tracking, analysing, and communicating compound event risks.

---

Note: The funding source for this challenge is EU funding under the Destination Earth Initiative. For details on the eligibility, please refer to Article 3 of the Terms and Conditions.

---

Description of the Challenge

As we shift to a net-zero energy system, the reliability of renewables faces a growing challenge: compound weather events—instances when multiple adverse climate conditions happen at the same time. For example, imagine a week of unusually low wind and solar radiation, paired with high temperatures and drought. These situations can severely strain energy supply systems and grid stability.

This challenge invites you to develop an innovative Compound Event Index (CEI) to detect, analyse, and predict these critical events, with a focus on impacts to the energy sector.

Key Focus Areas

1. Compound Event Detection
   • Identify co-occurrence of multiple adverse weather variables.
   • Explore trends in frequency, duration, and intensity of such events.
   • Map regional vulnerability hotspots for energy systems.

2. Machine Learning Innovation
   • Use techniques like Self-Organizing Maps (SOMs), clustering, or anomaly detection.
   • Integrate climate and/or energy indicators to anticipate future events.

3. Energy-Relevant Metrics
   • Translate compound weather signals into actionable risk indicators for energy planning.
   • Optionally incorporate variables soil-atmosphere coupling effects on energy generation (incl. analysing land-atmosphere feedback mechanisms such as evapotranspiration).

4. Usability & Impact
   • Build an interactive tool or visualisation to explore CEI outputs.
   • Consider regional customisation using the Climate Data Tool (CDT) or other frameworks.
   • Emphasise real-world applicability for energy planners, system operators, or grid risk assessments.

Evaluation criteria

• Feasibility
• Innovative approach
• Easy to maintain / Future-proof approach
• Transferability