jgfs_atmos_atmanlupp.ecf (GFS v17)

Branch / tenant: dev/gfs.v17 (gw_v17) Path: ecf/scripts/gfs/analysis/atmos/jgfs_atmos_analupp.ecf Analysis date: 2026-06-11 Method: On-disk v17 worktree analysis.

---

Summary

jgfs_atmos_atmanlupp.ecf (on disk as jgfs_atmos_analupp.ecf) is the NCO production ecFlow task that runs the Unified Post Processor (UPP) on the atmospheric analysis to produce GRIB2 master files from the analysis fields. It reuses the shared JGLOBAL_ATMOS_UPP J-Job with UPP_RUN='analysis' and FORECAST_HOUR='0'.

---

1. The script itself

Section	Detail
PBS resources	1 node, 120 mpiprocs, 1 OMP thread, 120 cpus, 500 GB, exclusive host, 15-min wall
Modules	load_modules.sh upp (UPP stack)
J-Job	JGLOBAL_ATMOS_UPP (shared, with UPP_RUN=analysis)
Exports	FORECAST_HOUR='0', FHR3='000', UPP_RUN='analysis'
Rocoto breadcrumb	upp.sh

---

2. Execution chain

jgfs_atmos_analupp.ecf
  +-- JGLOBAL_ATMOS_UPP (UPP_RUN=analysis)
       +-- exglobal_atmos_upp.py (Python/pygfs)
            +-- pygfs.task.upp.UPP class
            +-- UPP executable (analysis mode)
            +-- Produces GRIB2 master from analysis fields

---

3. v17-specific notes

• Shared J-Job: same JGLOBAL_ATMOS_UPP used by goesupp_master (UPP_RUN=goes) and product_master (UPP_RUN=atmos). Differentiated by UPP_RUN variable.
• Analysis mode: UPP_RUN='analysis', FORECAST_HOUR='0' — processes f000 analysis.
• Python-based driver via exglobal_atmos_upp.py using pygfs.task.upp.UPP.
• UPP modules: loads upp (not gsi or ufswm).
• 500 GB, 120 cpus — high memory for full-resolution analysis post-processing.
• Renamed from atmanlupp to analupp in v17.

---

4. Change-impact snapshot

Symbol	Risk	Notes
JGLOBAL_ATMOS_UPP	LOW (per graph)	Shared J-Job; changes affect all UPP_RUN variants

---

Sources

• On-disk v17 worktree: ecf/scripts/gfs/analysis/atmos/jgfs_atmos_analupp.ecf.