Simulating the human dimension in Destination Earth. An EO-Informed digital twin application for climate-adaptive policy planning

As Destination Earth (DestinE) matures, the capability to simulate not just natural phenomena but also the "related human activities" becomes critical for delivering actionable insights on sustainable development. This work presents TRANSITION, an operational Digital Twin application designed to model the complex socio-environmental dynamics of land-use change, renewable energy integration, and agricultural sustainability within the DestinE ecosystem.

While traditional Earth system digital twins excel at forecasting physical variables (e.g., crop yields or solar irradiance), they often lack the behavioral fidelity to predict how human actors will respond to these changes. TRANSITION bridges this gap by integrating Earth Observation (EO) data with a Multi-Level Agent-Based Modelling (ML-ABM) system driven by Reinforcement Learning (RL). In this framework, autonomous agents—representing farmers, landowners, and policymakers—make spatially explicit decisions based on environmental suitability, economic incentives, and social factors (PECS framework).

We demonstrate the application of this digital twin through three core stakeholder-co-designed use cases:

• Climate Change Adaptation Strategies: Simulating long-term land-use shifts under various CMIP climate scenarios to identify regions at risk of agricultural abandonment or suitable for crop diversification.
• Green Credit & Policy Simulation: allowing policymakers to "stress-test" interventions—such as subsidies for photovoltaics (PV) or green credits—in a risk-free virtual environment to assess adoption rates and potential conflicts between food and energy production.
• Renewable Energy Optimization: Utilizing Sentinel-derived analytics and high-resolution Digital Elevation Models (DEMs) to identify optimal deployment zones for renewable infrastructure while accounting for socio-economic acceptance.

 

To ensure these insights are scalable and actionable, the application is architected to eventually run on Destine’s Platform, with the potential to utilize High-Performance Computing (HPC) for heavy agent training and the DestinE Data Lake for seamless access to Sentinel and ERA5 datasets. By coupling high-precision physical modeling with realistic human behavior, TRANSITION offers a robust decision-support tool for evidence-based policymaking, directly contributing to the European Green Deal’s vision of a resilient and adaptive society.