PiCo is a post‑hoc framework that explains Geo‑foundation model predictions by identifying concept vectors in the model’s embedding space. It generates synthetic multispectral satellite data using a 3D radiative transfer model and a game engine to reveal physical sensitivities, then projects embeddings onto these vectors to quantify concept importance. Evaluations on three GFMs show that PiCo can recover most of the full‑embedding accuracy and uncover how different physical concepts influence predictions for tasks such as ground‑cover estimation and forest classification.
Geo-foundation models (GFMs) are typically evaluated by downstream task predictive performance, revealing little about which physical features they are sensitive to or use in prediction. This limits d...
arXiv:2607. 27217v1 Announce Type: cross Abstract: Forest aboveground biomass (AGB) is a critical indicator of ecosystem productivity and terrestrial carbon storage, yet regional carbon monitoring remains constrained by the sparse spatial and temporal availability of field inventories and airborne structural measurements.
The paper introduces FOLIAGE, a geometry‑centered latent world model designed for systems that grow and change shape over time. FOLIAGE allocates higher‑resolution state to regions expected to grow while keeping mature areas compact, fusing RGB, point‑cloud, and mesh data into a deployable geometric state and conditioning dynamics on material controls. The authors evaluate the model on new benchmarks (SURF‑GARDEN and SURF‑BENCH) and demonstrate significant reductions in inverse‑material error, mesh forecasting error, and improved cross‑modal retrieval, with robust performance under sensor loss and correspondence corruption.
arXiv:2607. 05207v1 Announce Type: cross Abstract: Self-supervised learning (SSL) is designed to learn generic, transferable representations rather than representations optimized for a single task.
The paper introduces a composition‑aware pretraining framework for geospatial foundation models that explicitly encodes fractional land‑cover mixtures as histogram targets for each satellite image cell. By using Earth Mover’s Distance to distill these composition targets into a 36.8 M‑parameter backbone, the authors demonstrate significant improvements on region‑level tasks such as zero‑shot image retrieval and scene classification, while maintaining competitive performance on fine‑grained tasks like segmentation and object detection. The method outperforms larger models (SatMAE and Prithvi‑EO‑2.0) and achieves a 55.6 % relative boost on the ForestNet‑12 dataset, evidencing the benefit of explicit composition modeling.
arXiv:2608.30392v1 Announce Type: new Abstract: Global food security and sustainable climate action increasingly rely on robust, scalable agricultural monitoring. Earth observation foundation models...
Self-supervised learning (SSL) is designed to learn generic, transferable representations rather than representations optimized for a single task. Most geospatial benchmarks evaluate representations solely through downstream tasks, providing limited insight into the information encoded within the representation itself.
arXiv:2609.24787v1 Announce Type: new Abstract: Forest inventories increasingly rely on artificial intelligence (AI) models to derive forest attributes from large-scale 3D point clouds. Current model...
arXiv:2606. 32023v1 Announce Type: cross Abstract: Forest attributes are essential for national-scale resource monitoring.
arXiv:2609.13868v1 Announce Type: cross Abstract: Earth-observation (EO) foundation models have become exceptionally effective at learning se mantic, high-dimensional geospatial embeddings, while mod...
arXiv:2608. 04792v1 Announce Type: new Abstract: Accurate estimation of Above-Ground Biomass (AGB) from satellite imagery is essential for the large-scale monitoring of carbon stocks, yet it remains a challenging regression task at global scale.
arXiv:2607. 07758v1 Announce Type: new Abstract: Foundation models (FMs) have transformed machine learning from isolated task-specific model development toward general-purpose models pretrained on broad data and adapted to multiple downstream tasks.