# An experimental drought forecast reaches thousands of streamgages.

USGS reports that its machine-learning models forecast streamflow drought one to thirteen weeks ahead at 3,219 streamgages in the contiguous United States, and support a new experimental forecasting tool.

Canonical: https://brightaifuture.com/signals/signal-usgs-river-droughtcast-2026
Format: source-report
Source publication: 2026-01-06
Bright publication: Not established
Substantive update: None recorded
Evidence and review: in-review; source report. Research candidate. Source page describes an experimental tool; its performance differs by drought intensity and forecast horizon.

## AI’s role

Machine-learning models forecast streamflow drought.

## What is documented

Forecasts at 3,219 streamgages for one to thirteen weeks ahead.

## Limitations

Experimental forecast product.

Performance varies by drought intensity and forecast horizon.

## Why follow this?

Water managers may gain earlier warning of constrained river flows. This is an experimental forecast product, not evidence that a particular allocation decision improved.

## Provenance and history

{
  "dates": {
    "eventDate": null,
    "publicationDate": "2026-01-06",
    "captureDate": "2026-09-06",
    "lastReviewedDate": null
  },
  "provenance": {
    "origin": "editorial",
    "externalId": "10.3389/frwa.2025.1709138"
  },
  "revisions": [],
  "corrections": []
}

## Original sources

- [Machine learning generated streamflow drought forecasts for the conterminous United States](https://www.usgs.gov/publications/machine-learning-generated-streamflow-drought-forecasts-conterminous-united-states-0)

## Continue exploring

- [Can AI help communities understand and use water more wisely?](https://brightaifuture.com/threads/water)