Pangu-Weather and FengWu edge out physics models on 2023 typhoons

In slides for the Typhoon Committee, HKO reported that Pangu-Weather and FengWu, run in a trial from mid-2023, had an edge over traditional models on typhoon tracks.

  • 2 an organization

    HKO found an edge for the AI models in track forecasts for eight cyclones in 2023, and an AI model correctly foresaw that a predicted cyclone would not form.

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

ECMWF, JMA, UKMO and NCEP global models for the same 2023 cyclones. Measured.

The system’s part
HKO compared the two models' own track forecasts with those of four traditional global models for the same cyclones, which isolates the models' skill. HKO describes them as machine learning-based models. HKO ran them itself in a trial, and the makers are outside HKO.
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
Pangu-Weather and FengWu, run by HKO in an internal real-time trial
Made by
Huawei (Pangu-Weather) and Shanghai Artificial Intelligence Laboratory (FengWu)
Used by
Hong Kong Observatory (internal trial to assist forecasting operations)
Where
Hong Kong, China (western North Pacific and South China Sea)
Stage
In a pilot
Date
, when the source was published
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.

    Hong Kong Observatory. Assessment and Verification of Artificial Intelligence Models on Tropical Cyclone Forecasting in 2023. Presentation by LUI Yuk-sing, Scientific Officer, to the 56th session of the ESCAP/WMO Typhoon Committee, 27 February 2024, slide 11.

    Although AI models have an edge over traditional NWPs in TC track forecast, TC intensity prediction remains a big challenge.

Details

AI system

Workflow

The AI models started from the operational analyses of traditional global models. HKO verified track and intensity errors for the cyclones of 2023.

Follow-up

Conditions for replication

Forecasters rely on traditional models for intensity and wind structure. The verification covers one season of eight cyclones.

Entry details

Sector

Weather forecasting / tropical cyclones.

Purpose

Huawei and the Shanghai Artificial Intelligence Laboratory built Pangu-Weather and FengWu to forecast global weather in the medium range with models trained on past weather data.

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). Pangu-Weather and FengWu edge out physics models on 2023 typhoons (AIBD-0754). Artificial Intelligence Benefits Database, edition of 31 October 2026. https://petervartanian.xyz/aibd/records/AIBD-0754.html (archived at https://doi.org/10.7910/DVN/AADSII, version 4.0)

BibTeX Data Field definitions