AI system
- System
-
ECMWF Artificial Intelligence Forecasting System (AIFS)
- Model
-
AIFS Single v1.
- Release
-
AIFS Single v1.
- Application developer
-
ECMWF - Role
-
AIFS generated weather forecasts.
- Configuration
-
The reported forecasts used AIFS Single v1, operational from 25 February 2025.
- Inputs and outputs
-
Not assessed
- Autonomy
-
Not assessed
- Workflow
-
Operational twice-daily AIFS Single runs initialized from ECMWF analyses. ECMWF's standard headline-score and tropical cyclone verification against the IFS.
- Human oversight
-
Not assessed
- Relation to AI
-
ECMWF's verification compares AIFS directly with the IFS. The only non-ECMWF cited source (NWS Q&A) confirms NHC evaluated ECMWF's AI system in 2025 but states no performance result. ECMWF is both the developer and the deployer, so its evaluations are the body reporting on its own system.
- AI contribution
-
ECMWF's verification compares AIFS directly with the IFS. The only non-ECMWF cited source (NWS Q&A) confirms NHC evaluated ECMWF's AI system in 2025 but states no performance result. ECMWF is both the developer and the deployer, so its evaluations are the body reporting on its own system.
Benefits
- Outcome identifiers
-
Not assessed
- Benefit
-
5–15% lower errors than IFS
- Category
-
Improved Task Accuracy
- Status
-
Not assessed
- Beneficiary groups
-
People. Organizations. Society.
- Beneficiaries
-
Not assessed
- Reach
-
Not assessed
- Distribution and access
-
Not assessed
- Duration
-
Not assessed
- Expected?
-
Yes
- Costs and harms
-
Not assessed
Measurement
- Studies
-
Not assessed
- Study design
-
Not assessed
- Measurement
-
Not assessed
- Setting and sample
-
Not assessed
- Measures
-
5–15% lower errors than IFS
- Comparison
-
ECMWF's physics-based Integrated Forecasting System (IFS)
- Results
-
Result: errors 5 to 15% lower in the medium range (out to 15 days).
- Additional value from AI
-
Not assessed
- Limitations
-
Not assessed
- Denominators
-
Not assessed
- Materials for replication
-
Not assessed
Money
- Kind of value
-
Not assessed
- Amount
-
Not assessed
- Currency and prices
-
Not assessed
- Period and unit
-
Not assessed
- Realization
-
Not assessed
- Recipients and payers
-
Not assessed
- Gross or net
-
Not assessed
- Included costs
-
Not assessed
- Comparison
-
Not assessed
- Method
-
Not assessed
- Return and payback
-
Not assessed
People and places
- Participants
-
Developer:
ECMWF. Operator: ECMWF. - Jurisdictions
-
Benefit: Reading,
United Kingdom. Developer: United Kingdom. Operator: United Kingdom. - Ownership and control
-
Not assessed
- Dependencies
-
Not assessed
- Cross-border connections
-
Not assessed
- Geopolitical implications
-
Not assessed
Follow-up
- Conditions for replication
-
Intensity under-prediction persists per ECMWF and the AIFS-TC preprint. No ECMWF-documented named event where AIFS beat the IFS was found. Event-level superiority remains unverified.
- Replication attempts
-
2 follow-up entries. Benefit recurred: 2.
- Later findings
-
1 July 2025. ECMWF launched the ensemble version, AIFS-ENS, operationally on 1 July 2025, and its 2025 evaluation (May 2026) reports both AIFS Single and AIFS ENS beating the IFS on most parameters.
- Availability
-
Not assessed
- Institutional decision
-
Not assessed
- Grounds
-
Not assessed
- Action and result
-
Not assessed
- Success criterion
-
Not assessed
Evidence
- Source numbers
-
AIBD-0143-source-0001–0006
- Citations
-
Numbered references below.
- Language
-
Not assessed
- Source relationships
-
Developer, Independent
- Original evidence
-
Not assessed
- Access and versions
-
Not assessed
- Supported findings
-
Not assessed
- Evidence links
-
Not assessed
- Evidence strength
-
Not assessed
- Missing evidence
-
Not assessed
Grades
- Assessed benefits
-
People. Organizations. Society.
- Depth
-
Not assessed
- Reasons
-
People: Aggregate skill statistics, no discrete event. Organizations: Aggregate skill statistics, no discrete event. Society: Aggregate skill statistics, no discrete event.
- Observed or intended
-
Not assessed
- Rubric
-
Not assessed
- Assessor and date
-
Not assessed
- Alternative assessments
-
Not assessed
- Fill
-
Independently studied
Entry details
- Kind
-
Not assessed
- Related entries
-
Not assessed
- Field
-
Weather forecasting. Public-sector meteorological service.
- Purpose
-
Operational data-driven global medium-range weather forecasting at ECMWF, run alongside the physics-based IFS.
- Dates
-
Date basis unassessed: 2025 (precision unassessed). Updated: 21 October 2026.
- Stage
-
Not assessed
- Version
-
SHA-256: 9e2e95a24742…
References
- Other identifiers
-
Not assessed
- Category mappings
-
Not assessed
- Brief
-
Not assessed
- Revisions
-
5 revisions. Latest: 21 October 2026.
Sources
-
Report Developer
Haiden, T., Chevallier, M. (2026). Forecast performance 2025. ECMWF Newsletter No. 187, May 2026.
Tropical cyclone (TC) position errors are reduced substantially in the AIFS compared to the IFS, whereas TC intensity errors are generally larger due to a weak-intensity bias.
-
Report Independent author
National Weather Service (n.d.). AI in Hurricane Forecasting at the National Hurricane Center: A Q&A with NHC Science Operation Officer Wallace Hogsett. NOAA/NWS.
the European Center for Medium-Range Weather Forecasts (ECMWF) have developed AI-driven global forecast systems, both of which were also available for evaluation during the 2025 season
-
Preprint Developer
Allen, A., Bruinsma, W. P., Maier-Gerber, M., Cook, H., Chantry, M., Turner, R. E. (2026). AIFS-TC: A simple correction competitive with the operational frontier for tropical cyclone intensity forecasting. arXiv:2608.09959.
While these models have been shown to achieve superior performance to physics-based NWP in forecasting tropical cyclone (TC) tracks, they tend to dramatically underestimate intensity.
-
Press release Developer
ECMWF (2025). ECMWF's AI forecasts become operational. ECMWF news, 25 February 2025.
The AIFS outperforms state-of-the-art physics-based models for many measures
-
Press release Developer
ECMWF (2026). 2025 forecast evaluation met with positive feedback. ECMWF news, 8 May 2026.
the Artificial Intelligence Forecasting System (AIFS) is outperforming the IFS for most surface and upper-air variables, with errors 5 to 15% lower in the medium range
-
Other Developer
ECMWF Forecast User wiki (2025). Severe Event Catalogue: 202501 - Windstorm - Eowyn.
Wind gusts not available for AIFS