VegSim: A Geospatial World Model for Scenario-Conditioned Vegetation Simulation

arXiv 2026 64 method, application

TLDR

VegSim is a geospatial world model for scenario-conditioned vegetation simulation using latent dynamics and weather forcing, evaluated on real data.

Reasoning

The paper presents a novel approach for scenario-conditioned vegetation simulation, with strong empirical results on real-world data and spatial-temporal shifts. However, it is domain-specific to vegetation and does not address interactive or reinforcement learning settings, limiting its generality as a world model.

Read-first score

Read-first score 64, weighted from topical fit, citation, graph, method, reproducibility, and recency signals. Original total remains 39.

Recency 6%

100

Uses a gentle age decay so recent papers surface without erasing older foundations. 2026

Citation impact 18%

91.7

Uses OpenAlex-shaped citation metadata as a bibliometric attention signal, separate from paper quality. citation_normalized_percentile=0.9167586

Reproducibility 18%

Screens links and visible text for paper, code, dataset, artifact, and repository signals. pdf=True; code=True; dataset=False; markers=code,dataset,github

Methodology quality 18%

Screens visible abstract and analysis fields for experiment, dataset, baseline, metric, and limitation evidence. markers=baseline,dataset

Topical relevance 29%

55.7

Uses existing LLM keyword relevance scores normalized to 0-100. world model,world simulator,generative world model,interactive world model,video world model,world dynamics prediction,model-based reinforcement learning world model

Citation velocity 12%

Citation velocity estimates citations per publication-year to reduce old-paper bias. velocity=0.00

Field roles

FoundationFrontierBridgeReproducibility anchor

Rank sensitivity

Stability: volatile; rank range: 479.

Keyword Scores

world model

world dynamics prediction

generative world model

world simulator

interactive world model

video world model

model-based reinforcement learning world model

Deep Analysis

Innovations

Scenario-conditioned vegetation simulation using a geospatial world model
Inference of latent vegetation state from sparse satellite-derived NDVI histories, past meteorological covariates, and static spatial context
Recurrent latent dynamics that propagate the latent state forward under controllable future weather forcing
Probabilistic forecasting and conditional simulation without supervision on scenario responses

Methodology

VegSim infers a latent vegetation state from sparse satellite-derived NDVI histories, past meteorological covariates, and static spatial context, then propagates it forward through recurrent latent dynamics under future weather forcing, and decodes predictive NDVI quantiles at each lead time. The model is trained on the GreenEarthNet dataset and evaluated under in-distribution and spatial, temporal, and joint spatial-temporal shifts against time series and Earth observation forecasting baselines.

Key Results

VegSim achieves strong point and probabilistic accuracy across all shift conditions while using a compact architecture. In a France summer 2022 case study, it produces spatially coherent vegetation response patterns consistent with known sensitivity to temperature and precipitation.

Resources

Paper Code