Mesoscale Parallel Experiment Change Log

Experiment Name:  North American WRF-NMM with various analysis and/or physics changes

When Who What
2007/06/06/18 Wu, Rogers
  • Turned off assimilation of mesonet moisture data due to degraded 600-300mb temperature fits in the GSI and big temperature adjustments due to convection in the first 1-2 hours of the model forecast
  • 2007/05/07/18 Wu
  • Fixed a bug in the GSI analysis so that it would correctly thin satellite data if the analysis domain extended beyond the North Pole. This bug caused the expanded NAM analysis to use about one order of magnitude less satellite data than the operational NAM analysis domain running in NAMX
  • 2007/04/25/18 Wu, Parrish
  • Changes to the GSI analysis:
    1. Newest GSI version, works with either WRF-NMM 3-d array indexing (ijk or ikj I/O) and strong balance constraint
    2. use AIRS radiance data
    3. use GOES 1X1 radiance data
    4. QuikScat ocean surface winds
    5. SATWND (eumetsat and modis)
    6. MESONET q and uv (winds with uselist)
  • 2007/02/15/18 Parrish, Wu
  • Use new version of GSI with strong balance constraint
  • NDASX cycle restarted from ops NDAS
  • 2007/02/05/18 Rogers
  • Started running the tm12 and tm09 portions of the offtime (06z/18z) parallel NDASX to mimic the operational NDAS cycling configuration.
  • NDAS surface update script was changed so that NDASX snow cover/depth, sea-ice, and SST are updated with the latest analyses at the start of the 06z NDASX run instead of the 12z run. This mimics what is done in operations.
  • 2007/01/24/00 Wu, Rogers
  • Revert back to ops BMJ convective parameterization (use ops WRF-NMM model binary)
  • Use ops GSI analysis with both the background error for moisture and the observation error for conventional moisture data reduced by 1/2
  • Parallel moved from mist to dew and NDASX cycle restarted from ops NDAS on mist
  • 2006/12/21/00 Ferrier, Wu
  • Revert back to ops GSI analysis code and input data
  • Use original (19 June - 18 December 2006) version of BMJ convective parameterization in the WRF-NMM model
  • Run with all other WRF-NMM model changes (divergence damping, horizontal diffusion, microphysics) implemented into the parallel ops NAM on mist on 6 December 2006 and the ops NAM on white on 19 December 2006
  • NDASX cycle restarted from ops NDAS on mist
  • 2006/12/12/00 Ferrier
  • Revert back to ops GSI input data, still use Oct 2006 version of the GSI code
  • Changes to convective parameterization from ops NAM code running on mist (the NAMY code from blue):
    1. Time scale of convection increased from 40 to 60 min
    2. Increased stability of deep convective profiles (STABDF,STABDS increased from 0.90 to 1.0). This change increases 700-mb temperatures, where the NAM has had a consistent cool bias.
    3. Slight modification to triggering algorithm to be more consistent when treating parcels that are initially saturated.
    4. Further code optimization to speed up execution (added some F90/95 stuff).
    5. Depth of shallow vs. deep convection was reduced from 0.2*Plow to 0.15*Plow (Plow is the midlayer pressure at the lowest level) based on sensitivity experiments.
    6. At swap points (shallow convection when test for deep convection failed) the shallow cloud top is calculated as a function of virtual temperature exceess and when the ambient relative humidity exceeds a threshold value based on a deficit saturation pressure of -40 hPa at cloud base. The deficit saturation pressure that is used is the one used from the failed deep convection calculation (DSPBK).
  • NDASX cycle restarted from ops NDAS on mist
  • 2006/12/06/00 Rogers
  • Parallel ops NAM run on mist is running with the latest NAM crisis change code (the WRF-NMM model code running in NDASY/NAMY on blue, with changes to BMJ convection and divergence damping) as of 00z 12/5, so NDASX was restarted from the ops NDAS on mist and made the same switch to the NAMY code starting with this cycle.
  • 2006/11/17/00 Wu, Rogers
  • Revert back to running the ops WRF-NMM NAM forecast code in the NDAS and free forecast
  • Use ops (March 2006) version of GSI analysis; turn on AIRS radiance data in GSI analysis
  • Restart cycled NDASX from parallel ops NDAS running on mist
  • 2006/11/07/00 Rogers
  • The 9/5/2006 change to the horizontal diffusion in the ops NAM/NDAS (increase Smagorinsky constant for lateral diffusion from 0.27 to its maximum value of 0.4) was missing in the ops NAM code on mist from which the NAMX binary This oversight was corrected starting with this cycle.
  • The parallel ops NDAS/NAM on mist was restarted from the official ops run on white at 18z 11/6, so NDASX was restarted from the ops NDAS starting with this cycle.
  • 2006/10/24/00 Janjic, Jovic, Rogers
  • Use new divergence damping routine with increased damping of the external mode
  • During NDAS, increase divergence damping of both external and internal mode by 5x over that done during the 84-h forecast
  • Page Last Modified: 23 October 2006