# Geospatial Foundation Model

Post-disaster building damage mapping is limited by labels: parameter-efficient fine-tuning adapts geospatial foundation models but still assumes large reliable annotation sets. PUMA recasts the task as positive-unlabeled learning, fine-tuning with a few confirmed damaged buildings and many unlabeled ones, while non-contrastive representation refinement and latent-state label disambiguation reinforce each other. On the xBD tsunami, hurricane, tornado and wildfire events, PUMA reaches the highest overall accuracy and F1 among the compared methods on tsunami, hurricane and wildfire, with F1 gains of 5.98 and 4.47 percentage points over the strongest competitor on tsunami and hurricane; on tornado it reaches 95.72% overall accuracy and 92.88% F1, competitive with the best NcPU.

Post-disaster building damage mapping is limited by labels: parameter-efficient fine-tuning adapts geospatial foundation models but still assumes large reliable annotation sets. PUMA recasts the task as positive-unlabeled learning, fine-tuning with a few confirmed damaged buildings and many unlabeled ones, while non-contrastive representation refinement and latent-state label disambiguation reinforce each other. On the xBD tsunami, hurricane, tornado and wildfire events, PUMA reaches the highest overall accuracy and F1 among the compared methods on tsunami, hurricane and wildfire, with F1 gains of 5.98 and 4.47 percentage points over the strongest competitor on tsunami and hurricane; on tornado it reaches 95.72% overall accuracy and 92.88% F1, competitive with the best NcPU.

Earth-observation foundation models such as AlphaEarth and Tessera compress each 10-metre pixel into a dense vector whose individual dimensions often mix several land-surface concepts. Sparse autoencoders expand that vector over an overcomplete dictionary and keep only a few atoms active, yielding sparse, approximately monosemantic features. At matched sparsity, three sparse-autoencoder variants reconstruct better than PCA, ICA, NMF and raw dimensions; yet a control whose decoder is a frozen random dictionary also shows more specific features and finer spatial scales, so those two properties follow from sparsity itself, and only a learned dictionary cuts AlphaEarth's unexplained variance eightfold.

Earth-observation foundation models such as AlphaEarth and Tessera compress each 10-metre pixel into a dense vector whose individual dimensions often mix several land-surface concepts. Sparse autoencoders expand that vector over an overcomplete dictionary and keep only a few atoms active, yielding sparse, approximately monosemantic features. At matched sparsity, three sparse-autoencoder variants reconstruct better than PCA, ICA, NMF and raw dimensions; yet a control whose decoder is a frozen random dictionary also shows more specific features and finer spatial scales, so those two properties follow from sparsity itself, and only a learned dictionary cuts AlphaEarth's unexplained variance eightfold.

PlacePulse compresses 1,055,537 US H3 resolution-7 cells into 256-dimensional unit-norm vectors using only static structural attributes such as demographics, businesses and environmental risk, beating a dimensionality-matched feature-selection baseline on 74 external targets and leading Google's Population Dynamics Foundation Model on health and economic outcomes.

PlacePulse compresses 1,055,537 US H3 resolution-7 cells into 256-dimensional unit-norm vectors using only static structural attributes such as demographics, businesses and environmental risk, beating a dimensionality-matched feature-selection baseline on 74 external targets and leading Google's Population Dynamics Foundation Model on health and economic outcomes.