Scientific framing¶
The forecast couples the nowcast with weather nowcast/forecast and climate scenarios to project hazard susceptibility into the future — from hours/days (operational warning) to seasonal and climate-scenario timescales (scenario exploration). This is the “computational playground” of the problem statement.
State variables & observables¶
inputs: forecast forcing (precipitation, temperature, atmospheric rivers), evolving soil state, scenario parameters;
outputs: time-evolving per-hazard susceptibility with forecast uncertainty and lead time.
Data & forcing¶
(Link to DataHub and weather/climate products: AR index, ACE2, Clima-X, downscaling. See Ocean–Atmosphere Coupling.)
Models¶
(Link to ModelHub: coupled weather→hazard model chains, surrogates / reduced- order models for fast scenario sweeps, downscaling.)
Evaluation & metrics¶
(Link to HazEvalHub: skill vs. persistence/climatology baselines, ROC / precision-recall at decision thresholds, cost-loss / value metrics, lead-time vs. skill.)
Open questions & roadmap¶
Propagating uncertainty from weather forecast through hazard models.
Climate-scenario design for actionable, decision-relevant projections.