Unifying Innovations in Forecasting Capabilities Workshop
A UFS Collaboration Powered by EPIC
Monday, July 20, 2026 – Friday, July 24, 2026
College Park, MD
Join us this July at the College Park Marriott Hotel & Conference Center in College Park, Maryland or online for the Unifying Innovations in Forecasting Capabilities Workshop 2026 (UIFCW26). This event is an excellent opportunity for community partners, across academia, industry, and government, to collaborate in our shared mission to advance forecasting capabilities. The objective of this workshop is to strengthen the understanding of our collaborative efforts to improve the Unified Forecast System (UFS), accelerate meaningful contributions, and measure progress more effectively.
Stay tuned for more details on the workshop trainings and agenda, coming in late spring, 2026.
Registration
Registration is now open! Click the button below to register for UIFCW26, either in person at the College Park Marriott Hotel & Conference Center in College Park, Maryland, or virtually.
If you have ANY issues submitting the form, please reach out to EPIC Support (support.epic@noaa.gov ) to complete your registration.
In-person registration must be completed by Monday, July 6, 2026; there will be no on-site registration for in-person attendance.
UIFCW 2026 will take place from Monday, July 20, 2026, to Friday, July 24, 2026 at 3501 University Boulevard E, Hyattsville, MD, 20783-7998 and online. This event is free to attend. Keep an eye out for additional information regarding after-hours dinners and networking events.
Save the date and check back for more details about #UIFCW2026.
Abstract Details
UFS Application | Short Range Weather (SRW) Application and Rapid Refresh Forecast System (RRFS)
Abstract submissions are open! We are soliciting abstracts from the following topics:
- UFS Applications
- Artificial Intelligence in Numerical Weather Prediction
- Machine Learning
- Data Assimilation
- MPAS Integration
- Cross-Cutting NWP Topics
- High-Performance Computing/Cloud
- R2O2R
- Community Engagement
Abstracts are due by Thursday, April 30, 2026.
Travel and Lodging
We have reserved a block of rooms at the College Park Marriott Hotel & Conference Center and The Hotel at the University of Maryland for attendees who need accommodations.
Each guest is responsible for making their own reservation.
Hotel and travel details are available on the UIFCW 2026 event page. Please refer there for more information for information on hotel options, booking and group rates. Attendees are encouraged to review these details in advance of registrations.
2026 Workshop Schedule
Get ready for five incredible days of inspiration, learning, and connection. Whether you’re joining in-depth technical sessions or exploring broader topics, the schedule is designed to support your experience and interests.
- Want a quick overview? Start with the Schedule at a Glance for a snapshot of each day.
- Looking for details? Browse the Detailed Agenda to explore session topics, speaker lineups, and more.
Pre-event Training
Training 1 – Introduction to Running NWS Global NWP Models in Development Mode
Monday, July 20th
9am – 12pm EDT
College Park Marriott Hotel & Conference Center in College Park, MD, and Online
Description: Learn how to run the National Weather Service (NWS) Global Workflow (GW), the driver behind Global Forecast System (GFS), Global Ensemble Forecast System (GEFS), the upcoming Seasonal Forecast System (SFS), and the Global Chemistry and Aerosol Forecast System (GCAFS), all of which are based on the UFS weather model. Participants will run a low-resolution GFS test case while learning how the workflow code and output are structured, and the capabilities of the recent public release. Participants who wish to participate in the hands-on portion should bring a laptop. Limited Amazon Web Services (AWS) cloud resources will be available to those without another supported High Performance Computing (HPC) resource. Users with accounts on NOAA’s Weather and Climate Operational Supercomputing System 2 (WCOSS2), Ursa, or Gaea C6; MSU’s Hercules; or National Science Foundation(NSF)/ National Center for Atmospheric Research’s (NCAR) Derecho will be asked to use one of those machines to save AWS resources for those without.
Objective: Learn how to run Global Workflow cases spanning many different configurations
Instructor(s): Walter Kolczynsk & Wei Huang
This training is limited to 50 participants.
Training 2 -Adding Idealized Test Cases to the UFS Hierarchical System Development Framework
Monday, July 20th
9am – 12pm EDT
College Park Marriott Hotel & Conference Center in College Park, MD, and Online
Description: This training will introduce users to the Unified Forecast System (UFS) Hierarchical System Development (HSD) framework, with an emphasis on integrating a new test case into the HSD framework. Users will receive an overview of the framework (background, structure, workflow), followed by a walkthrough of the steps required to integrate a new test case. The walkthrough will consist of identifying prerequisite input and fix files, determining appropriate model configurations and compatible physics suites, as well as understanding the hierarchical placement and required modifications to the workflow scripts and files. The training will use the recently added Aquaplanet idealized test case as a basis for adding a new case to the UFS HSD framework, complementing the generalized walkthrough for case integration.
Objective: Provide users with background on how to integrate a test case into the UFS HSD framework
Instructor(s): Natalie Perlin, Ratko Vasic, Priya Pillai, Josh Kublnick
This training is limited to 50 participants.
Training 3 – Project EAGLE Training: Machine Learning Weather Forecasting on Microsoft Azure
Monday, July 20th
1pm – 4pm EDT
College Park Marriott Hotel & Conference Center in College Park, MD, and Online
Description: This training introduces participants to NOAA’s Experimental Artificial Intelligence (AI) Global and Limited-area Ensemble forecast system (EAGLE) running in near real time on the Microsoft Azure cloud platform. The course will cover the workflow and infrastructure supporting machine learning (ML)–based weather forecasting, including preprocessing pipelines, model training procedures, inference execution, and forecast verification techniques.
Participants will also learn how Project EAGLE supports collaborative development and experimentation, including the process for contributing new machine learning models and workflows for operational-style inference and verification. The training is intended for both NOAA and external collaborators interested in developing, evaluating, and deploying ML-based forecasting applications within the EAGLE framework.
Objective: Provide NOAA and non-NOAA participants with hands-on knowledge and practical guidance for using NOAA’s EAGLE forecasting system in a near-real-time cloud environment. Participants will learn the end-to-end workflow for machine learning weather forecasting, including data preprocessing, model training, inference, and verification, while also gaining an understanding of how to contribute new capabilities and models to Project EAGLE.
Instructor(s): Alex Burrows & Mariah Pope
This training is limited to 100 in-person participants and 50 virtual participants.
New User/Student Track Training
CADRE-JEDI-Edu Tutorial: Introducing Data Assimilation Methods in JEDI Using a Simple Two-Layer QG Model
Wednesday, July 22nd
1:45pm – 4:15pm EDT
College Park Marriott Hotel & Conference Center in College Park, MD
This training is scheduled during the new user/student track on Wednesday of the workshop and is gauged towards those new to model code and development.
This tutorial will use the CADRE-JEDI-Edu python-based workflow to introduce the basic data assimilation algorithms in JEDI in the context of a simple two-layer quasi-geostrophic model that you can run from your own laptop. The tutorial will consist of a series of Observation System Simulation Experiments with single-observation and multi-observation settings, to showcase LETKF, 3DVar, 4DVar, 3DEnVar, hybrid 3DEnVar, 4DEnVar, and local particle filter. In addition to introducing technical aspects of the use of these methods in JEDI, the tutorial will compare their differences and discuss the sensitivity of results to select parameters.
Information will be provided to registered attendees about how to install Docker and view the Python notebooks, which should be done before the tutorial.
Instructor(s): Aaron Johnson, Nick Gasperoni, Jon Poterjoy
Abstract Details
This session will showcase developments in the UFS across multiple applications including:
- Short Range Weather (SRW) & Rapid Refresh Forecast System (RRFS)
- Hurricane Applications (HAFS)
- Coastal
- Air Quality
- Land
- Global workflow (includes Subseasonal to Seasonal (S2S), Seasonal Forecast System (SFS), and Atmospheric Rivers (AR))
Topics may also include model coupling, Data Assimilation (DA), physics improvements, and applications for forecasting and ecosystem prediction. Contributions on operational implementation and research advancements are also welcome.
Artificial Intelligence (AI) for Numerical Weather Prediction (NWP)
This session invites developers and scientists working on topics including but not limited to: The development and operationalization of partial/full data-driven models for short, medium, and long range weather forecasting, AI-based parameterization, novel techniques for accuracy improvement using AI in DA and/or post-processing, software/hardware engineering aspects related to model development, data acquisition, scalability on training and inference, etc.
Machine Learning
This session explores the integration of Machine Learning (ML) into NWP to enhance forecast accuracy and computational efficiency. We invite abstracts focusing on data-driven emulators, hybrid modeling, and ML-based post-processing. Key areas of interest include deep learning for high-resolution nowcasting, bias correction of global ensembles, and the development of energy-efficient models capable of processing large volumes of data.
MPAS Integration
This session highlights advancements in the Model for Prediction Across Scales (MPAS) dynamical core and its integration into UFS. Topics covered in this session include model development, data assimilation, physics enhancements, coupling, performance tuning/optimization, code management, workflow, and verification efforts that improve high-impact, short-range weather forecasts. Contributions on system performance, operational readiness, and emerging research are also encouraged.
Cross Cutting NWP Topics
This session features cross-cutting components of the UFS, and presentations highlight cutting-edge research on Physics, Dynamics, System Architecture and Infrastructure, and Verification & Validation which transform multiple applications.
High-Performance Computing (HPC)/cloud
High-Performance Computing (HPC) enables researchers, including those from academia and industry, to run complex simulations without the constraints of local hardware. This session explores the critical evolution of HPC and Cloud infrastructure as the backbone of NOAA’s next generation operational and research applications. This session bridges the gap between atmospheric science and system architecture. Abstracts are encouraged, topics relating to transition to cloud computing, the NOAA RDHPCS program, and success stories.
Data Assimilation
DA is a critical component of Earth System Models (ESMs). In this session, presentations highlight cutting-edge research on DA, DA applications (such as Land-DA), innovative observing techniques, and coupled DA processes. Challenges and opportunities in data assimilation are also welcome.
Research to Operations to Research (R2O2R)
This session focuses on discussions regarding the end-to-end pipeline of the UFS, emphasizing the bidirectional exchange between the research community and operational centers. Presentations highlight successful transitions of innovations into NOAA’s production suite, the use of “Operations to Research (O2R)” feedback loops to identify systematic model errors, and the evolving infrastructure—such as UFS code releases and the UFS Usability Assessment—that supports community contributions. We welcome abstracts on transition metrics, hurdles in the R2O process, and strategies for accelerating the delivery of cutting-edge science to forecasters.
Community Engagement
This session highlights community engagement efforts related to the UFS and showcases opportunities for collaboration across the broader user community, WINGS fellows, and students or new users. Presenters are encouraged to emphasize interdisciplinary partnerships, knowledge sharing, and effective communication of science across various audiences.
Poster session
Share your research, use case, or innovation with other community members during our poster session and networking mixer! Submission of an abstract is required to participate in the poster session.
