---
title: 'Google Research Flood Hub: AI forecasts for ungauged basins'
slug: google-flood-hub
stable_id: 73cebfbedc998854
company: Google Research Flood Hub
function_code: engineering
pattern_codes:
  - continuous_decisioning
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.
collections: []
bundle_version: 1.0.0
bundle_fingerprint: sha256:c23c6cc2b88153d008ea8fda928f632ce0011fc2d4c5036672a16e5d895bab93
canonical_url: https://brianletort.ai/transformations/google-flood-hub
---

# Google Research Flood Hub: AI forecasts for ungauged basins

A global model can transfer hydrology into ungauged basins.

Function: Engineering design. Patterns: Continuous decisioning. Evidence: verified.

Freshness: current. Reviewed: 2026-08-23. Updated: 2026-08-22.


Source quality: 1 peer reviewed, 1 primary; publication outcomes are verified.

## Before

1. **Hydrologic forecasting system** — Uses gauge-calibrated or global physical models to produce river forecasts. (control: Ungauged basins lack long local records; GloFAS day-zero nowcast is the study comparator.)
2. **Government or humanitarian authority** — Interprets available forecast and decides whether to warn or mobilize. (control: Official warning and emergency powers remain human.)

## After

1. **Google AI hydrology and inundation models** — Transfer learning from open data to produce river-flow and inundation forecasts up to seven days. (control: Published model skill and confidence; no autonomous emergency command.)
2. **Government, NGO, or user** — Interprets the forecast and decides protective action. (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.

## Outcomes

- **Reliable lead time** (verified): GloFAS day-zero nowcast → Similar or better reliability at up to five days. Model skill, not fatalities avoided.

## 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?

## Sources

- [Global prediction of extreme floods in ungauged watersheds](https://www.nature.com/articles/s41586-024-07145-1) — Nature
- [How Google uses AI for global flood forecasting](https://blog.google/innovation-and-ai/products/google-ai-global-flood-forecasting/) — Google