Skip to content

X Experiments

Jump to bottom
rtodling edited this page Nov 12, 2024 · 189 revisions

List of X-Experiments

Exp Name Period Tag/Version Storage Strategy Blurb
x0016 Summer 2015; Winter 2015-16 GEOSadas-5_13_1_p2 N/A Full ADAS Baseline
x0027_ctl Winter 2016-17 EnADAS-5_17_0p5 /archive/u/mfuente2/ Full ADAS Baseline
x0027A Winter 2016-17 EnADAS-5_17_0p5 /archive/u/rtodling/ Full ADAS Hyb-4DEnVar
x0035_Ana Winter 2017-18; Summer 2018 GEOSadas-5_22_0 /archive/u/dao_it Full ADAS Baseline
x0039_p5 Winter 2019-20 rt-GEOSadas-5_25_1p5 /archive/u/dao_it Full ADAS Baseline
x0039_p6 Summer 2019 rt-GEOSadas-5_25_1p6 /archive/u/dao_it Full ADAS Baseline
x0041 Summer 2019; Winter 2019-20 GEOSadas-5_26_0_p3 PRJ/dadev/dao_it/archive Full ADAS Baseline
x0042 Summer 2019; Winter 2019-20 GEOSadas-5_27_1 PRJ/dadev/dao_it/archive Full ADAS Baseline
x0042c2 Summer 2019; Winter 2019-20 GEOSadas-5_27_1 PRJ/dadev/dao_it/archive Full ADAS COSMIC-2
x0043 Summer 2019; Winter 2019-20 GEOSadas-5_27_1 PRJ/dadev/dao_it/archive Full ADAS Baseline
--- -------- HERE ON: ALL GIT TAGS -------- -------- --------
x0044 S2019; W2019-20; W2020-21 rt-GEOSadas-5_27_1_p3 PRJ/dadev/rtodling/archive Full ADAS Baseline
x0045 Winter 2021 rt-GEOSadas-5_28_0_p3 PRJ/dadev/dao_it/archive Full ADAS Baseline
x0045z Winter 2021 rt-GEOSadas-5_28_0_p5z PRJ/dadev/dao_it/archive Full ADAS Test
x0045a Winter 2021 rt-GEOSadas-5_28_0_p5 PRJ/dadev/dao_it/archive Full ADAS Baseline
x0046 Winter 2021 rt-GEOSadas-5_29_1 PRJ/dadev/dao_it/archive Full ADAS Baseline
x0046aRPY Winter 2021 rt-GEOSadas-5_29_2 PRJ/dadev/dao_it/archive Full ADAS Baseline
x0046a S2019 W2021 W2022 rt-GEOSadas-5_29_3_p2(*) PRJ/dadev/dao_it/archive Full ADAS Baseline
x0046c W2022 GEOSadas-5.29.4.1 PRJ/dadev/rtodling/archive Full ADAS Baseline
x0046d W2022 GEOSadas-5.29.4.1 PRJ/dadev/rtodling/archive Full ADAS Baseline
x0046d S2022 GEOSadas-5.29.5-p1 PRJ/dadev/rtodling/archive Full ADAS Baseline
x0046d S2023 GEOSadas-5.29.5-p2 PRJ/dadev/rtodling/archive Full ADAS Baseline
x0046e S2023 GEOSadas-5.29.5-p4 PRJ/dadev/dao_it/archive Full ADAS Baseline
x0047RPY S2022 rt-GEOSadas-5.30.0-x0047 PRJ/dadev/rtodling/archive Replay ADAS Baseline
x0048 S2022 GEOSadas-5.30.1 PRJ/dadev/dao_it/archive Full ADAS Baseline
x0048 W2021-22 GEOSadas-5.30.2 PRJ/dadev/dao_it/archive Full ADAS Baseline
x0049 S2023 GEOSadas-5.30.3 PRJ/dadev/dao_it/archive Full ADAS Baseline
x0050RPY S2023 GEOSadas-5.41.0 PRJ/dadev/dao_it/archive Full ADAS Baseline
x0050 S2023 GEOSadas-5.41.1 PRJ/dadev/dao_it/archive Full ADAS Baseline

Notes:

  • PRJ stands for /discover/nobackup/projects/gmao
  • Removed refers to experiments removed from archive and no longer available

Key Forward Processing Upgrades and Corresponding X-Experiments

FP Tag/Version date X-Exp Baseline Main Feature
e513_fp GEOSadas-5_13_1_p2 ?? x0016 Hybrid 3DEnVar
f516_fp GEOSadas-5_16_0_p5 01 Nov 2016 xxx 12.5 km system
f517_fp GEOSadas-5_17_0_p5 24 Jan 2017 x0027A Hybrid 4DEnVar
f521_fp GEOSadas-5_21_2 11 Jul 2018 x0034 see
f522_fp GEOSadas-5_22_0_p9 13 Mar 2019 x0035_Ana see
f525_fp GEOSadas-5_25_1_p6 30 Jan 2020 x0039(p5/p6) see
f5271_fp GEOSadas-5_27_1_p2 25 Feb 2021 x0044 see, TM
f5293_fpp rt-GEOSadas-5_29_3_p2 1 Nov 2021 x0046a see, TM
f5293_fpp GEOSadas-5_29_4 3 Feb 2022 x0046a see, TM
f5294_fp GEOSadas-5_29_4 1 Mar 2022 x0046a see, TM
f5295_fpp GEOSadas-5.29.5-p2 Mar 2022 x0046d see, TM
f5295_fp GEOSadas-5.29.5-p3 Feb 2023 x0046d see, TM
f5295_fp GEOSadas-5.29.5-p4 Mar 2024 x0046e see, TM
f5295_fp GEOSadas-5.29.5-p6 May 2024 x0046e see, TM

Note: Additional information on GEOS Forward Processing (FP) updates and other GEOS Atmospheric Assimilation Products is available by following this link.

Blurb on Experiments

x0035_Ana

GCM Jason + CrIS (FSR) and ATMS NOAA-20, retune the polar AMV’s, NEXRAD winds off, OMPS ozone mapper on, and AIRS and IASI correlated errors.

x0039

Mainly changes to AGCM: see the GEOSgcm_App ChangeLog file for GCM updates for Jason-3_3 & Jason-3_4

  • RRTMG updates for version 4.10 from Eli
  • Correction to GF diagnostics.
  • Updates to produce stratospheric forecast statistics
  • Correction to gcm_run.j & gcm_forecast.tmpl related to REPLAY with GAAS aod files
  • Updated Aerosol Tuning and Transport with GF.
  • Implement new regrid.pl for proper use with LAND (fixes long-standing errors with regridding)
  • many others; see ChangeLog

x0039_p6 (p5)

  • Correction to RRTMG Shortwave to remove/significantly_reduce stratospheric cold bias.

x0041

  • Merge with NCEP
  • All-sky MHS, AMSR-2
  • Aircraft bias correction
  • MetOp-C radiances (no IASI)
  • Ozone-sensitive hyper-IR radiances
  • Fix for passive thinning

x0042

  • Update to Atmos-Ocean Skin Layer (though same formulation as in x0041)
  • Revise GNSSRO foward operator (code change to reduce bias)

x0042c2

  • As x0042 plus COSMIC-2 GPSRO

x0042t1

  • Replay to x0042 ensemble, with Deterministic ADAS removing the Dry Mass Constraint

x0043

  • Variational bias correction of aircraft observations
  • Modified observation operator for GNSS RO
  • All-sky MHS and AMSR-2 radiances
  • Add Metop-C: AMSU-A and MHS radiances, as well as GRAS GNSS RO
  • Add COSMIC-2 GNSS RO
  • Revised CrIS radiance assimilation: (i) CrIS-FSR for SNPP; (ii) revised FSR channels for both SNPP and NOAA-20; (iii) account for inter-channel observation error correlations

x0044

As x0043 but:

  • Ozone-sensitive radiances for all hyperspectral IR
  • CrIS-FSR over ocean waters only
  • x0044 is also a skip-ahead experiment that moves the baseline to Winter 2020/21

x0045

Looks at the consequences of:

  • Additional sources of RO observations (#28): PAZ (kx=44), KOMPSAT(kx=825)
  • Additional commercial RO observations (#28): GeoOptics(kx=265) & Spire (kx=269)
    Note period of availability: 20201218_00z-20210114_18z
  • Additional IASI from Metop-C (kx=877; treated as IASI on Metop-A and -B)
  • Replace version of CRTM to CRTM-2.3.0
  • Correction for high-latitude buoy handling (#23)

Note 1: An error was found in the settings of this exp that cause RO data to be exactly doubled at read. As it turns out, given how GSI handles superob/RO weighting this turns out to be minor, but the experiment is being re-run (see x0045a).

Note 2: A series of independently-run OSE's has identified assimilation of SatWind observations to degrade ADAS results. Further and closer investigation identifies the issue to be with how GOES-R is not being thinned and it is having its errors cut down to half, causing the fits to be overwhelmingly force toward GOES-R at the detriment of fits to other data.

x0045z (never actually ran)

As in x0044 and:

  • CRTM update to CRTM-2.3.0
  • Correction for high-latitude buoy handling (#23)
  • Revised handling of GOES-R: thinning applied to kx=245 and 246, and ob-error as in prep-error-table.

x0045a

As in x0045, and:

  • Additional sources of RO observations (#28): PAZ (kx=44), KOMPSAT(kx=825)
  • Additional commercial RO observations (#28): GeoOptics(kx=265) & Spire (kx=269)
    Note period of availability: 20201218_00z-20210114_18z
  • Additional IASI from Metop-C (kx=877; treated as IASI on Metop-A and -B)
  • Replace version of CRTM to CRTM-2.3.0
  • Correction for high-latitude buoy handling (#23)

but unlike x0045 this now includes:

  • Correction for double RO data error in original x0045, and
  • Revised handling of GOES-R: thinning applied to kx=245 and 246, and ob-error as in prep-error-table (bypass halving of ob error wired by NCEP).

x0046

  • From the analysis perspective the following has been changed:

    1. Removal of GeoOptics & Spire
    2. Mild revision to RO operator (see Revision from Lidia )
    3. Also includes Jing changes to support trace gas analysis (zero-diff for meteorology see)
    4. Windowed-out AIRS data for 1 Dec 2020; data is corrupt in its default location (see obsys.rc)

  • Updates MAPL to MAPL-2 and incorporates version v1.12.3 of the model (this having been shown to give largely neutral results as compared viz. z0044 vs x0044).

  • Tag for this experiment has not been versionized (but it will be to form a release for FPP)

  • REMARK: This run failed: CO2 from model was found to have a problem (it was being removed by deposition and affecting the radiance assimilation in an undesirable fashion). See x0046aRPY and x0046a for replacement.

x0046aRPY

  • This is identical to x0046, expect:
    1. CO2 bug is now fixed (no deposition of carbon species)
    2. Replay ensemble to x0045a: this is for a "quick" check to see that all is well with the fix (see x0046a for full run)

x0046a (under dao_it)

  • This is nearly identical to x0046aRPY, expect that:

    1. Turns off ASCAT from Metop-A (leaving that from Metop-B on) see full explanation
    2. Runs full ADAS (on Skylake nodes)
    3. Adds capability to run on Skylake and Cascade

  • Summer case actually uses: GEOSadas-5_29_3_p3; relevant changes from tag rt-GEOSadas-5_29_3_p2 used for Winter case are:

    1. Trim of aircraft bias correction file (to have reproducibility in Winter case, use ``setenv CLEANUP_BIAS 0'').
    2. Version 5 (as opposed to V4) MLS-T data; these data are presently only used passively for verification; so results are not really affected; reversal to v4 is possible by simply editing obsys.rc.
    3. Fix to auto-launch of monthly means script (zero diff).
    4. Fix for (SLURM) constraint in script controlling ASENS jobs (zero-diff).
    5. There also numerous other changes but those related to procedures not involved in x-exps or FP-like applications.

  • Second Winter covering Nov 2021 to Jan 2022 uses: FP tag GEOSadas-5_29_4

    1. Incorporates all changes in both Winter and Summer above.
    2. Moves control period ahead in time (see x0046b).

Milestone: Upgrade from 5.27.1 to 5.29.4

  • Additional sources of RO observations (#28): PAZ (kx=44), KOMPSAT(kx=825)
  • Additional IASI from Metop-C (kx=877; treated as IASI on Metop-A and -B)
  • Upgrade version of CRTM to CRTM-2.3.0
  • Correction for high-latitude buoy handling (#23)
  • Revised handling of GOES-R: thinning applied to kx=245 and 246, and ob-error as in prep-error-table
    Mild revision to RO operator (see Revision from Lidia )
  • Changes to support trace gas analysis (zero-diff for meteorology see)
  • MAPL-2.8.0 - patched
  • Integrated RadMon
  • On the model side:
    (i) no deposition of carbon species;
    (ii) revised Henry's Law constant for NH3 following https://doi.org/10.5194/acp-17-12911-2017
  • Running on Skylake
  • And a host of minor changes/fixes/expanded scripting workflow capabilities

x0046c (under rtodling)

  • As in x0046a, but running over Dec 2021 and Jan 2022 testing addition of Spire RO: commgpstop=45.,spiregpserrinf=2.
  • The follow parameters associated with RO have been adjusted: nsig_ext=18,gpstop=55 The justification for the top adjustment is to follow recommendation above others (Rick Anthes & U.K. Met Office) in thinking that RO above 55 km has error that are not properly handled in our current systems.

x0046d (under rtodling)

  • Winter 2021-22: Spire RO test in the configuration of NCEP's operations; also lower top of RO use to 55km.
  • Summer 2022: Similar RO configuration as from Winter case.
  • Winter and Summer based on mildly different tags (see table above); the difference being the model solar parameters.
  • Summer 2023 - run with slightly modified sat-db to match FP's choice (since running very two-weeks behind real-time).

Milestone: Upgrade from 5.29.4 to 5.29.5

  • Tag: GEOSadas-5.29.5-p2 (same as GEOSadas-5.29.4-p3 - the GEOS-IT tag)
  • GEOS-IT ready
  • Add Spire RO (NOAA/NCEP operational version)
  • Lower RO top from 60km to 55km
  • Update model solar parameters
  • Revise MAPL regrid options in HISTORY (not for background fields, so zero-diff in this sense)
  • Introduce convinfo database as a means to control in/out data considered conventional (zero-diff)

x0047RPY (under rtodling)

  • Summer 2022: Similar x0046d, but updating model to v10.23.0 (including MAPL and related repos).
  • The main addition here is GOCART-2G.
  • This exp replays the ensemble to that from x0046d (its control).

x0048

  • As x0047PRY, but running full ADAS (including ensemble).
  • Main purpose: test various fixes related to assimilation of hyperspectral IR observations: revide cloud detection for CrIS-FSR; adjust QC; adjust varBC.
  • This experiment also introduces: CRTM v2.4.0 (in NCEP_Shared)
  • Supports tracer analysis with reference to minimal background information (no need for sfc and upper air meteorological fields).
  • Model version v10.25.1 (zero-diff to v10.25.0, but aided with R21C output needs).

x0048

  • Mostly similar to x0048-Summer, but set up using revised restarts for the satellite biases; both IR and MW.
  • Details of changes can be found in the release notes.

x0049

  • Main modifications are addition of Planet IQ (RO) and revised surface marine data handling.
  • Running over Summer 2023 (mid-July onward).
  • Baseline for comparison is x0046d.
  • Further details are found under: release notes.

Milestone: Upgrade from 5.29.5-p3 to 5.29.5-p4

  • Improved handling of ship observations (should see FSOI for ship going in right direction).
  • Enable Planet-IQ (RO)
  • Capable of handling OMPS-LP
  • Capable of handling NOAA-21 (ATMS)

x0046e

x0050RPY

  • Summer 2023 - Combines the following:

  • Model: 11.5.2 (with a change to GEOS_mkiau to allow for its handling of cubed background and analysis files)

  • Analysis:

    • CRTM-2.4.0 as in the 5.30.x system (notice this points to the CRTM as in M21C - i.e., with proper Cloud Coeff used in All-sky)
    • brings in changes from 5.29.5-p7
    • brings in changes x0049 (5.30.3)

  • Adds GridComp for LDAS - but inactive in any of the cases considered in 5.41.0

  • Restart for summer case differs from restarts used in x50R12rt experiment in that the satbias-pc are refreshed for MW.

x0050

Clone this wiki locally