GDMI foresaw Hurricane Lorena's stall off Baja California

NHC found that Google DeepMind's ensemble mean beat its official track forecasts for Hurricane Lorena at all lead times. The model also foresaw that the storm would stall off Baja California.

  • 2 an organization

    GDMI's track errors were lower than NHC's official errors at every lead time, though over only two to eight forecasts.

  • Not rated society

    NHC does not report that the AI guidance changed warnings or the response in Baja California Sur.

Depth runs from 1, a convenience or time saved, to 5, a life or a livelihood preserved.

Status
Reported benefit.
Evidence

foundstudiedreportedclaimed

Only the deployer supports the benefit.

Compared with

NHC official track forecasts (OFCL) for Lorena in the homogeneous comparison. Measured.

Without the system
OFCL errors of 23.3, 46.7, 53.9 and 54.2 n mi at 12, 24, 36 and 48 h
With it
GDMI errors of 19.0, 36.1, 53.0 and 47.4 n mi
The system’s part
NHC verified GDMI's own track forecasts against the best track alongside the official forecast and other models, which isolates the model's skill. NHC gives no account of how forecasters used GDMI for Lorena. NHC helped Google DeepMind develop the model and uses it, so it is not fully outside the case.
Limits
Not yet checked.
Costs and harms
Not yet checked.
Later
Not yet checked.

Download data Cite Suggest a correction

Benefit type
Improved Task Accuracy
System
Google DeepMind tropical cyclone model, ensemble mean (NHC designation GDMI)
Made by
Google DeepMind
Used by
NOAA National Hurricane Center (experimental guidance), with output supplied by Google DeepMind
Where
Baja California Sur, Mexico
Stage
In a pilot
Date
Kind
Aggregate result
Broad benefit
Estimating the Quantifiable Characteristics of Products, Events, or Information
Collections
Cross-border records
Last revised
31 October 2026

Sources

  1. Report, by the organization that used it.

    Jelsema, J., and A. Gibbs. Hurricane Lorena (EP122025), 2–4 September 2025. National Hurricane Center Tropical Cyclone Report, 16 January 2026, p. 5 and Table 4b.

    The Google DeepMind Ensemble Mean (GDMI) was particularly skillful, outperforming the official forecasts at all forecast intervals, and also correctly predicted the cyclone would stall off the coast of Baja California.

  2. News report, by authors with no tie to the system. Context or contrary evidence.

    Michael Lowry, 'This Hurricane Season, Two Forecast Models Stand Out, but for Very Different Reasons', Eye on the Tropics (Substack newsletter), 3 November 2025, with preliminary 2025 track and intensity verification charts for the Atlantic and eastern Pacific by James Franklin, retired chief forecaster of the National Hurricane Center.

    Not only that, DeepMind outpaced even the gold-standard corrected consensus models as well as the National Hurricane Center's official track forecasts this year.

  3. Press release, by the organization that used it. Context or contrary evidence.

    National Weather Service. AI in Hurricane Forecasting at the National Hurricane Center: A Q&A with NHC's Science Operations Officer. NOAA/NWS, undated.

    NHC has partnered directly with Google DeepMind to develop a new AI hurricane forecast model that was used experimentally during the 2025 season.

Details

AI system

Workflow

NHC included GDMI among the guidance models in its homogeneous comparison for Lorena. NHC measured each model's track errors against Lorena's best track.

Follow-up

Conditions for replication

The sample is small, with eight forecasts at 12 hours and two at 48 hours. Forecasters draw on other aids for intensity, since GDMI's intensity forecasts at short range were worse.

Entry details

Sector

Weather forecasting / tropical cyclones.

Purpose

Google DeepMind built the experimental model to forecast the tracks and intensity of tropical cyclones as ensemble guidance for forecasters.

Not yet checked: model, release, provider, role, configuration, inputs and outputs, autonomy, human oversight, beneficiaries, reach, distribution and access, duration, every measurement question, every money question, ownership and control, dependencies, cross-border connections, geopolitical implications, replication attempts, availability, institutional decision, grounds, action and result, success criterion, every evidence question, observed or intended, alternative assessments, related entries, other identifiers, category mappings.

Cite this record

Peter H. Vartanian. (2026). GDMI foresaw Hurricane Lorena's stall off Baja California (AIBD-0752). Artificial Intelligence Benefits Database, edition of 31 October 2026. https://petervartanian.xyz/aibd/records/AIBD-0752.html (archived at https://doi.org/10.7910/DVN/AADSII, version 4.0)

BibTeX Data Field definitions