Matthias-Reif-PhD/WRF: v0.3

Le résumé fourni par la source

WRF-VPRM v0.3 (WRF component) This release archives the modified WRF v4.5.2 source used in the manuscript "WRF-VPRM v0.2 (WRF v4.5.2): Sensitivity of CO2 exchange to model resolution and parameter settings over Alpine topography" (Geoscientific Model Development, in review). This is the WRF-P branch — base WRF v4.5.2 extended with: VPRM-based GHG flux module (chem/module_ghg_fluxes.F) Parallel execution of multiple VPRM parameter sets Analytical temperature-derivative implementation CORINE Land Cover (CLC) entries in run/VEGPARM.TBL and run/LANDUSE.TBL LEO5 build script (job_compile.slurm) Companion code WPS fork (modified, WRF-VPRM-CLC branch): https://github.com/Matthias-Reif-PhD/WPS/releases/tag/v0.2 Workflow & post-processing: https://github.com/Matthias-Reif-PhD/WRF_VPRM_inComplexTopo/releases/tag/v0.2 Note on prior Zenodo archives The earlier Zenodo record 10.5281/zenodo.19997546 (versions v0.1 and v2) inadvertently bundled upstream WRF/WPS rather than the modified forks. This v0.2 release on the WRF fork is the first persistent archive of the actual modified WRF source described in the paper. What's Changed in v0.3 Fixes a silent correctness bug in the parallel VPRM channels and adds a five-member Topt-percentile ensemble that runs online in a single simulation. Pairs with WPS v0.2; corrected output is published as Zenodo record 20395680. If you ran v0.2 or earlier with vprm_opt='VPRM_table_MULTI', re-run. The EBIO_GEE, EBIO_GEE_2…_5 and EBIO_GEE_DPDT fields were computed at the wrong optimum temperature; the correction is 7–34 % at midday. Fixed — per-channel Topt In chem/module_ghg_fluxes.F (VPRM and dflux_dT), Topt was assigned with Fortran DATA statements placed inside a runtime SELECT CASE / if (p_ebio_gee_t == …) block. DATA is a compile-time initializer and ignores runtime control flow, so the compiler kept only the first in source order — the VPRM_table_US values 20,20,20,20,20,22,18,0. Every parallel channel ran at that same wrong Topt, though rad0/lambda/alpha/RESP0 did vary correctly. ifort emitted no warning. Each in-branch DATA Topt /…/ is now an executable assignment Topt = (/…/), with a default before the SELECT CASE, in both subroutines. EBIO_GEE_REF, EBIO_GEE_DPDT_REF and all EBIO_RES* are unaffected — their intended Topt equalled the value the bug forced, and respiration has no Topt dependence. Added — online Topt-percentile ensemble Five new channels (9–13) carry the p10/p25/p50/p75/p90 calibrations, so the ensemble comes out of one run instead of five offline recompute passes: EBIO_GEE_P10 EBIO_GEE_P25 EBIO_GEE_P50 EBIO_GEE_P75 EBIO_GEE_P90 EBIO_RES_P10 EBIO_RES_P25 EBIO_RES_P50 EBIO_RES_P75 EBIO_RES_P90 Existing channel names and indices are unchanged — the new fields are appended and are diagnostic only, feeding no co2_bio tracer, so they add no advected species. Validated Cross-checked at 54/9/3/1 km against an independent offline recomputation driven by each run's own archived T2 and SWDOWN: GPP agrees to ≤ 0.028 %, RECO to 0.000 %. At 54 km and 27 km the unaffected control channels reproduce pre-fix output bit-for-bit while the corrected channels change. Upgrading Full rebuild required — the Registry change alters the grid derived type. Run ./clean then ./compile em_real; not ./clean -a, which destroys configure.wrf, Registry/Registry and run/namelist.input. Diffs against older runs also move for two unrelated reasons: WRF is not bit-reproducible across different MPI rank counts, and the 9 km namelist's empty radt = , is now radt = 9. TOPTFIX_STATUS.md in the repo has the full write-up — validation tables, the two traps a future contributor is most likely to hit, and a pre-existing Wscale NaN issue that is not introduced here. Full Changelog: https://github.com/Matthias-Reif-PhD/WRF/compare/v0.2...v0.3

Ce résumé expose les affirmations des auteurs. BNTIC ne l’interprète pas comme une validation indépendante des résultats.