Bringing partners back
Built with our community
GAIA HazLab is powered by the people and institutions advancing open geohazard science.
We build digital twins of the Earth — fusing open, multimodal data with AI and cloud computing, all grounded in physical models — to monitor and forecast soil, landslides, liquefaction, and floods.
Landslides from atmospheric rivers, liquefaction in saturated soils, debris flows after wildfire, storms fed by the land surface. These cascade across the ocean, atmosphere, and solid Earth. We take a data-driven, physics-grounded approach to monitor, characterize, and predict them — in real time and under future climate.
Three coupled processes where dense sensing and physical models change what we can see.
Exploring memory effects in the critical zone with soil reanalysis products and event catalogs — how antecedent wetting, drying, and disturbance set today's hazard.
Explore →Slope stability under atmospheric-river rainfall, coupling antecedent moisture and transient pore pressure.
Explore →How saturated soils lose strength under shaking — and how antecedent state modulates seismic ground failure.
Explore →Each demo is a working slice of the GAIA stack applied to a real, coupled-hazard use case — real events, real sensors, real time.
During the December-2025 atmospheric-river floods on Mt. Rainier's glacial rivers, a seismic network estimated river discharge in the long reaches between sparse stream gauges, while a Stage IV precipitation mosaic tracked the rainfall driving the floods — the forcing and the response, side by side, toward earlier flood awareness in the Orting–Puyallup lahar corridor.
A live digital twin that fuses real-time rainfall, soil-moisture, and seismic signals with physics-based slope-stability models to anticipate where and when shallow landslides are likely to fail — turning monitoring into early warning.
Open, reproducible building blocks — data, models, evaluation, and the tools that connect them.
Streamlined access to precipitation, streamflow, seismic, and DAS data across Washington State.
Explore data →Physics-informed and surrogate ML models for hazard prediction and pattern discovery.
View models →Benchmarks and skill scores to validate hazard forecasts against what actually happened.
Evaluation tools →The agentic-AI layer — a cross-disciplinary translator, agentic data downloaders, and research-software agents.
Learn more →Open packages for data I/O, AI/ML, visualization, and reproducible hazard workflows.
Browse software →An interactive map of every colocated sensor and event in our regions of interest.
Open dashboard →The GAIA CRESST catalog brings seismic, hydrological, meteorological, and remote-sensing observations into a single interactive view — so researchers across disciplines can spot what's colocated with their region of interest.
Open the full dashboardEvery hazard, the models that predict or detect it, and the data they rely on — one interactive view of the whole system. Hue is the category, brightness the subcategory; hover a node to trace its chain, click for detail.
Open the full graphGAIA HazLab is powered by the people and institutions advancing open geohazard science.
A cross-disciplinary group spanning seismology, hydrology, geotechnics, and machine learning.