{"stable_id":"73cebfbedc998854","slug":"google-flood-hub","company":"Google Research Flood Hub","workflow_name":"AI forecasts for ungauged basins","function_code":"engineering","pattern_codes":["continuous_decisioning"],"changed_assumption":"A global model can transfer hydrology into ungauged basins.","evidence_strength":"verified","publication_tier":"showcase","freshness":"current","reviewed_at":"2026-08-23","updated_at":"2026-08-22","source_quality_summary":"1 peer reviewed, 1 primary; publication outcomes are verified.","caveat_summary":"Model skill, not fatalities avoided.","freshness_caveat":null,"workflow_summary":{"before":[{"actor":"Hydrologic forecasting system","action":"Uses gauge-calibrated or global physical models to produce river forecasts.","actor_type":"control"},{"actor":"Government or humanitarian authority","action":"Interprets available forecast and decides whether to warn or mobilize.","actor_type":"control"}],"hinge":"A global model can transfer hydrology into ungauged basins.","after":[{"actor":"Google AI hydrology and inundation models","action":"Transfer learning from open data to produce river-flow and inundation forecasts up to seven days.","actor_type":"ai"},{"actor":"Government, NGO, or user","action":"Interprets the forecast and decides protective action.","actor_type":"system"}],"decision_mode":"moved","decision_marker":"Selection moves from a fixed rule to the model"},"before":[{"order":1,"actor":"Hydrologic forecasting system","action":"Uses gauge-calibrated or global physical models to produce river forecasts.","handoff_to":"Warning authorities","control":"Ungauged basins lack long local records; GloFAS day-zero nowcast is the study comparator."},{"order":2,"actor":"Government or humanitarian authority","action":"Interprets available forecast and decides whether to warn or mobilize.","handoff_to":"Communities","control":"Official warning and emergency powers remain human."}],"after":[{"order":1,"actor":"Google AI hydrology and inundation models","action":"Transfer learning from open data to produce river-flow and inundation forecasts up to seven days.","handoff_to":"Flood Hub, Search/Maps, and partner feeds","control":"Published model skill and confidence; no autonomous emergency command."},{"order":2,"actor":"Government, NGO, or user","action":"Interprets the forecast and decides protective action.","handoff_to":"Community response","control":"Official channels remain authoritative; low-confidence or uncovered areas use existing systems."}],"decision_rights":"Model publishes forecasts; humans own warnings and protective action.","exception_path":"Low-confidence locations use official alternative channels.","removed_work":["Per-basin local calibration requirement"],"outcomes":[{"metric":"Reliable lead time","baseline":"GloFAS day-zero nowcast","result":"Similar or better reliability at up to five days","period":"Published 2024","scale":"80+ countries; areas covering 460m people","attribution_caveat":"Model skill, not fatalities avoided.","evidence_label":"verified"}],"executive_lesson":"The change is useful lead time where gauges are absent.","anti_pattern":"Do not claim lives saved from skill alone.","questions_for_leaders":["Where is the operating threshold set and who can override it?","What measured result would trigger rollback or retraining?","Which residual decisions must remain human-owned?"],"collections":[],"bundle_version":"1.0.0","bundle_fingerprint":"sha256:c23c6cc2b88153d008ea8fda928f632ce0011fc2d4c5036672a16e5d895bab93","canonical_url":"https://brianletort.ai/transformations/google-flood-hub"}