Awesome GAIA

A curated list of software around climate-compounded geohazards built by GAIA HazLab — predicting, in real time, the susceptibility of landslides, liquefaction/ground failures, and floods in the critical zone.

The repositories below are organized to mirror the GAIA digital-twin framework: a DataHub that stages observations, a ModelHub of susceptibility and surrogate models, hazard detection and sensing, and the agentic-AI and knowledge tooling that ties them together. See the DataHub Integration Guide for how these pieces are meant to fit together.

• DataHub — staging & catalogs
• ModelHub — susceptibility, surrogates & digital twins
• Forecasting & AI weather
• Hazard detection & sensing
• Agentic AI, knowledge & translation
• Compute & infrastructure
• Use cases
• Project site & meta

The data backbone: a s3://cresst object store (COG/Zarr/Parquet), static STAC catalogs on GitHub, and a staging CLI.

• gaia-cli — Prototype data-staging CLI: gaia stage prism|hrrr|synoptic|all clips an AOI/time range from multiple providers, harmonizes to standardized names/units/CRS, and writes a Zarr DataTree.
• catalog — Interactive Leaflet web map of Washington-state geohazard sensors and context layers, published at gaia-hazlab.github.io/catalog.
• solus-stac — STAC catalog wrapping USDA SOLUS100 soil properties (18 variables × 7 depths, 100 m) as odc-stac-loadable COGs.
• prism-stac — STAC catalog for PRISM daily precipitation and mean temperature (800 m, CONUS).
• precip-stac — STAC catalog for UW atmospheric-river precipitation products (4 km). Work in progress.
• landlab-stac — STAC catalog of co-registered 10 m surface inputs (soil, terrain, geotech, burn) for Landlab modeling.
• gaia-data-downloaders — Reproducible download notebooks for CONUS404 (incl. SMOIS/TSLB), HRRR, WRF-CMIP6, PRISM, Stage IV, USGS gauges, and more; documents the shared s3://cresst integration.
• usgs-gauge-utils — Utilities for querying USGS river-gauge stations. Superseded — see gaia-cli / gaia-data-downloaders.

• landslide-digital-twin — MVP cloud-native, daily-updating landslide-susceptibility digital twin coupling Landlab slope stability, PRISM weather, POLARIS soil, and AI weather forecasts.
• landlab-debrisflow — Landlab pipeline for post-fire terrain, ecohydrology, and landslide/debris-flow probability (LandslideProbability Monte Carlo).
• fire-debrisflow-ml — Post-fire raster data preparation and ML prediction of debris-flow / sediment change from soil, terrain, burn-severity, and land-cover layers.
• da-seis-groundfailure — Combining hydrology, seismology, and geotechnical engineering to predict earthquake liquefaction and ground failure.
• gaia-nhsm-deagg — USGS National Seismic Hazard Model disaggregation client; assembles gridded (M, R, ε, PGA) lookups for the downstream liquefaction surrogate.
• shred-landlab-prototypes — Prototypes combining PySHRED and Landlab for spatiotemporal forecasting and reconstruction of geophysical systems.
• seis-hydro-2-sed — Turning river discharge and seismic power into predictions of dynamic bedload sediment transport.

• earth2studio-test — Experiments with NVIDIA Earth2Studio (GraphCast/AIFS/StormCast) for AI-driven weather forecasts that force the hazard models.

• gaia-landslides-detect — Detecting landslides from seismic monitoring.
• mt-rainier-smart-sensing — Scripts to pull multi-sensor data at Mt. Rainier for monitoring multiple geohazards.

• gaia-agentic-ai — The GAIA HazLab agentic-AI project for autonomous data and analysis workflows.
• gaia-skills — Multi-disciplinary agentic instructions ("skills") shared across GAIA agents.
• gaia-literature-kb — Graph-RAG knowledge base over multi-disciplinary geoscience literature (seismology, hydrology, geomorphology, atmospheric/soil sciences, tectonics, geophysical sensing).
• gaia-translate-QA — Evaluation dataset for the GAIA Translator, the cross-disciplinary geoscience Q&A assistant.

• skypilot-hyak — SkyPilot setup for running GAIA workloads on UW Hyak HPC resources.

• gaia-stehekin-postfire-debrisflows — Investigation of the 2025 Stehekin post-fire debris flows.

• gaia-hazlab.github.io — The GAIA HazLab website and documentation book (problem statement, digital-twin framework, hazards, DataHub/ModelHub/HazEvalHub).
• awesome-gaia — This curated index.

New GAIA repository? Add it under the matching section with a one-line description, then place it in the project organization. For data-producing repos, follow the DataHub Integration Guide so the data is discoverable and provenance-tracked.