About Project Dataset Model Intercomparison and Evaluation Members and Contacts Meetings Resources

About

The Weather Prediction Model Intercomparison Project (WP-MIP) aims to accelerate the development and integration of the next-generation of operational weather prediction systems.

The project will create a public database of predictions from the full spectrum of artificial intelligence-based (AIWP), physically based (NWP) and hybrid systems using standard production pipelines.  This centralized archive will be used for distributed diagnostics and model evaluation, with the goals of:

  1. Identifying the strengths and weaknesses of the different prediction paradigms in an operational context;
  2. Providing guidance to operational centres on metrics to use for fair evaluation and comparison of AIWP, NWP and hybrid systems; and,
  3. Accelerating the development of the full spectrum of operational models.
Graphic representation of WP-MIP organization and subproject integration.
WP-MIP organization and subproject integration.

As a Pilot Project of the WMO Integrated Processing and Prediction System (WIPPS), WP-MIP will inform the development of AI technical guidance for National Meteorological and Hydrological Centres for short- and medium-term forecasting. This aligns the project with the needs of national weather services for operational decision-making and with WMO priorities under the Early Warnings for All initiative.

The project is divided into two streams based on the initialization data. The Same Initial Conditions (SIC) stream consists of forecasts initialized with ECMWF analysis grids, while forecasts in the Own Initial Conditions (OIC) stream are initialized with each centre's own analyses. The OIC stream reflects operational configurations, and comparisons between the streams provide valuable information on the sensitivity to initial conditions in the different forecasting systems.

In addition to the core SIC and OIC integrations, contributors are encouraged to participate in optional subprojects focusing on specific forecast features. These sub-projects require additional data and offer the benefit of more detailed diagnostics and model evaluation.

More information about the project is contained in the whitepaper.

Project Dataset

Forecasts from global deterministic AIWP, NWP and hybrid models for 2024 are available for diagnostics and evaluation. The forecast archive is being expanded as new contributions are submitted to the project.

 

Model Intercomparison and Evaluation

Diagnostic and evaluation activities are underway now that the WP-MIP forecast archive is taking shape. To maximize the potential for collaboration across teams and to reduce redundancy, coordination of these efforts is guided by the JWGFVR. All contributions to WP-MIP verification research are welcome.
Button "Download data" Button that links to verification dataset

Graphic depiction of a globe with people icons

Members and Contacts

Relevant Meetings

Event Dates and Locations Description
Ver-AI Workshop 23-24 June 2025; Reading, UK Project planning and preliminary diagnostics and evaluations of early OIC-stream contributions
JWGFVR Workshop 20-24 October 2025; Montreal, Canada Diagnostics and evaluation of OIC-stream contributions, and preliminary assessments of initial condition sensitivities and SIC-stream contributions
3rd WP-MIP Project Meeting
 11 December 2025; Virtual
 Updates on project objectives and progress, new tools and support, and coordination of verification activities
Icon depicting papers and a magnifying glass

WP-MIP Resources

Contributors Resources

Project Planning