HSD Releases and Summary of Test Cases

Earth Prediction Innovation Center (EPIC) , through collaboration with our partners or collaborators at NOAA National Weather Service (NWS) Environmental Modeling Center (EMC), NOAA’s Physical Sciences Laboratory, NSF NCAR’s Developmental Testbed Center (DTC), and the University of Michigan, has overseen the release of three test cases as part of a continued effort to enhance the Unified Forecast System (UFS) Hierarchical System Development (HSD) framework. The first HSD release (November 2024) contained the July 2020 Convective Available Potential Energy (CAPE) case (originally developed as a UFS Case Study by the DTC) and baroclinic instability idealized case (developed by Xiaqiong Zhou of NOAA/NWS/EMC). The second HSD release (June 2025) was a collaboration with the NOAA/NWS/EMC/Hurricane Analysis and Forecast System (HAFS) group and focused on the integration of a regional tropical cyclone case. An aquaplanet idealized case was added in February 2026, which was developed in collaboration with NOAA’s PSL. Furthermore, each HSD release also expanded the user documentation and tests-dev framework of the UFS Weather Model (WM). Additional details about each test case can be found below.

• July 2020 CAPE – This case study was used to investigate the low Convective Available Potential Energy (CAPE) bias that exists in the Global Forecast System (GFS) which is attributed to model’s differences in latent and sensible heat fluxes when compared to Global Data Assimilation System (GDAS) observations. This case is an atmosphere-only forecast run at C48 resolution with 127 vertical levels and is set to run 24 hours starting at 2020-07-23 00Z.

  

• Baroclinic Instability Case – This case was investigated and designed to evaluate the accuracy of various atmospheric models in simulating a baroclinic wave. This test aims to assess how well “dry dynamical cores”, the foundational components of weather and climate models that handle air movement and temperature changes, perform in idealized conditions. It also measures their ability to handle the development of baroclinic waves and specifically which individual models are the most accurate in doing so. This case is an atmosphere-only forecast run at C192 resolution with 127 vertical levels and is set to run a dynamics-only 24 hour forecast starting at 2019-12-03 00Z.

  

• Idealized, Regional Tropical Cyclone Case – This case study is derived from the I-HAFS configuration and is designed to support controlled studies of tropical cyclone dynamics and forecast development. The case is configured to run at 4-km resolution, with 81 vertical levels. The forecast is initialized on 24 August 2019, and initial conditions and lateral boundary conditions are provided for up to a five-day forecast duration.

  

• The aquaplanet test case was designed in collaboration with NOAA’s PSL to provide the community with an easy-to-run, pre-configured atmosphere-only configuration in which the Earth is approximated to be covered by ocean and devoid of topography. It follows traditional aquaplanet configurations, which have been utilized historically to study Earth system interactions between the ocean and atmosphere by simplifying the complex interactions which exist in a fully-coupled Earth System Model (ESM). By restricting interactions to ocean-atmosphere components and removing land, sea ice, and other feedbacks, researchers can isolate key physical processes and discern more robust insights into the impact of their code innovations on forecast skill and model performance. Two key features of most aquaplanet experiments are (1) the global replacement of all land and sea-ice with ocean and the removal of topography, thereby approximating the Earth’s surface as an aquaplanet, and (2) the prescription of a fixed, analytically variant sea surface temperature (SST) that is kept constant throughout the model run. The case is available in the “develop“ branch of the UFS-WM in a default C48 resolution (~ 2° × 2° grid). The necessary initial conditions pre-staged on all NOAA Tier-1 platforms and in the HSD S3 bucket. The test case is also available to run via container, available on Docker hub.