Brand new weather model debuts today and more on August 31, 2026.

The EURO model and the EURO AI model got an upgrade last night and some models are being replaced by other models on August 31, 2026. 

National Weather Service Headquarters Silver Spring MD Service Change Notice 26-48 

Here is the AI summary of the changes:

Effective August 31, 2026, the National Weather Service (NWS) in Silver Spring, MD, will implement the Rapid Refresh Forecast System (RRFS) and RRFS Ensemble Forecast System (REFS), replacing the legacy NAM, HREF, SREF, and HiresW models. Other 2026 changes include updates to climate data assimilation (CORe), Day 4-5 excessive rainfall discussions, and the transition of GFS model products. [1, 2, 3, 4]


Key 2026 Model Changes and Updates
  • RRFS and REFS Implementation (Aug 31, 2026): NCEP will launch these new modeling systems, marking a major shift in convective-allowing modeling.
  • Retirement of Legacy Models: The NAM, HREF, SREF, and HiresW models will be retired on Aug 31, 2026.
  • CORe Implementation (Mar 2026): The Climate Object-oriented Reanalysis (CORe) replaces the Climate Data Assimilation System (CDAS) for atmospheric reanalysis, offering improved data from 1950 to present.
  • ERO Enhancements (May 2026): Day 4 and 5 Excessive Rainfall Outlook (ERO) discussions are being integrated into the QPFERD product.
  • GFS Product Changes: Public comments are being solicited for the proposed termination of various Global Forecast System (GFS) products. [1, 2, 3, 4]
Operational Changes & Notices
  • NOMADS Upgrade (Feb 23, 2026): OpenDAP access is being retired, requiring users to switch to grib filter or HTTPS downloads for NCEP model data.
  • Weather Prediction Models: Recent updates include improvements to the National Blend of Models (NBMv4.2/v5.0) and the experimental Wind Tropical Cyclone forecast Message (WTCM).
  • Notification System: Official notices are available through the NWS Service Change Notices page. [2, 5, 6, 7, 8]
For the full list of retired products and detailed technical changes, refer to SCN 26-47. [1]



AI responses may include mistakes.