Support for GaeaC6 Added to UFS Short-Range Weather Application

[gspetro-NOAA]

As part of ongoing efforts to expand multi-platform compatibility of the Unified Forecast System (UFS) Short-Range Weather (SRW) Application, a set of targeted updates was introduced to support the GaeaC6 partition as a new National Oceanic and Atmospheric Administration (NOAA) Research and Development High-Performance Computing (RDHPC) system. These changes ensure that the SRW App can run seamlessly on GaeaC6 for both development and production workflows, particularly for testing, forecast execution, and post-processing.

Summary of Key Changes

1. New Modulefiles for GaeaC6 Tasks

   New Lua-based modulefiles were added to the modulefiles/tasks/gaeac6/ directory to configure the environment for common workflow tasks:

   • run_vx.local.lua: Loads appropriate versions of MET and METplus for verification steps and sets necessary environment variables.
   • python_srw.lua: Prepares a Python/Conda environment for SRW tasks, ensuring compatibility with system libraries.
   • plot_allvars.local.lua and get_obs.local.lua: Provides streamlined access to plotting and observation ingestion environments.

   These files help isolate GaeaC6-specific setups from other machine configurations, improving maintainability and clarity.

2. Generalized Workflow Module Support

   • wflow_gaeac6.lua: A new top-level modulefile was added to define the base environment for the SRW App on GaeaC6. It updates the Math Kernel Library (MKL) path, sets up Conda, and adds a message reminder to activate the appropriate environment (srw_app).
   • Updates to wflow_gaea.lua refine the definition for the older C5 partition and provide naming consistency.

3. New Machine Configuration YAML

   A comprehensive YAML configuration file (ush/machine/gaeac6.yaml) defines scheduler settings, path mappings to observational and model data, and commands for job execution on the GaeaC6 cluster. This file ensures that both Workflow End-to-End (WE2E) testing and real workflows can be run without manual intervention.

4. Workflow Testing Enhancements

   • Added GaeaC6 entries to test scripts (tests/WE2E/setup_WE2E_tests.sh, tests/build.sh) to enable WE2E testing.
   • New machine suite test lists were created (i.e., comprehensive.gaeac6, coverage.gaeac6 in tests/WE2E/machine_suites/) for systematic testing across configurations.

5. Scheduler Configuration for Slurm

   Modifications to sbatch job card templates (e.g., get_ics.sbatch, get_lbcs.sbatch in ush/wrappers/job_cards/sbatch) include comments specifying correct Slurm workload manager flags for GaeaC6 (e.g., --clusters=es, --partition=eslogin_c6, --export=NONE) to help developers adapt scripts as needed.

6. Parameter Validation Updates

   The parameter validation file was updated to include "GAEAC6" as a valid machine identifier, allowing the configuration validation logic to recognize and correctly handle the new platform.

Why These Changes Matter

These updates reflect a strategic step toward enabling UFS applications to run robustly across all NOAA RDHPC systems, enhancing resilience, scalability, and flexibility for developers and researchers. GaeaC6 offers newer architecture and more cores per node (128), allowing faster execution and broader test coverage for ensemble and deterministic forecasts. By codifying these changes in modular and well-scoped updates, the system remains maintainable while supporting new hardware as it comes online.