Action aware emulators improve earth ecosystem simulations
Earth System World Model for What-If Simulations: A Case Study for Terrestrial Ecosystems
Machine LearningArtificial Intelligence
Summary
Simulating Earth’s ecosystems accurately can take a lot of time and computing power. The authors developed a new kind of model that not only imitates how ecosystems change naturally but also lets users test 'what if' scenarios by making changes during the simulation. This model learns from data without needing specially labeled examples showing interventions. Their tests show the model predicts long-term ecosystem changes well while allowing interactive changes, which could help scientists explore different environmental outcomes more flexibly.
What this means in practice
- •For earth system modelers: Run interactive ecosystem simulations that let users modify states mid-run to test hypothetical environmental changes.
- •For environmental data platform developers: Incorporate action-conditioned surrogate models to build more responsive digital twins of terrestrial ecosystems for research and planning.
Authors
Zhihao Wang, Ruichen Wang, Ruohan Li, Lei Ma, George Hurtt, Xiaowei Jia, Gengchen Mai, Shaowen Wang, Yiqun Xie
Abstract
Machine learning emulators have become essential for accelerating expensive Earth-system simulations, but most existing approaches remain passive forecasters: they reproduce simulator trajectories under prescribed forcings without an explicit interaction mechanism for user-specified interventions. This limits their use in interactive scientific workflows and Earth-system digital twins, where users often need to explore how a system would respond if selected state components were changed. We propose an action-conditioned world-modeling framework for Earth-system emulation that reformulates simulator trajectories as supervision for controllable state-transition learning. The key idea is transition-action pretraining: naturally observed state changes are treated as label-free action supervision, allowing the model to learn both prescribed dynamics and action-conditioned responses without manually annotated interventions. We further introduce masked response learning to infer unobserved variables under partial state edits and learn coupled system dependencies. We test this framework on ecosystem dynamics across six global regions and multiple stand ages. Experiments show that the model preserves competitive long-horizon emulation accuracy while enabling controllable structural interventions and coherent responses in coupled ecosystem-cycle variables. These results suggest a practical route from passive Earth-system emulators toward interactive, intervention-aware scientific surrogates.