Mesoscale Parallel Experiment Change Log

Experiment Name:  North American NEMS/NMMB test

When Who What
2016/12/08/12 Rogers
  • Parallel DGEX turned off since DGEX will be discontinued with this NAMv4 implementation.
    • 2016/11/22/00 Rogers, Y. Lin
  • NAMX verification and web pages now pointing at NCO parallel NAMv4 run as of this cycle.
    • 2016/10/27/18 Collard, Carley, Rogers
  • Disabled AMSU-A Metop-B channel 15 which went bad on Oct 17. Since this channel was used to QC other channels from the same instrument, it had the side effect of rejecting data from the entire instrument. So the GSI code had to be changed so that these other channels are not rejected.
    • 2016/09/08/18 Rogers, Collard
  • Turn off HIRS Metop-A satellite radiances due to data degredation.
    • 2016/08/03/18 Rogers, Carley
  • Modified dependency checks to ensure that the tm00 analysis will wait up to 130 minutes after analysis time (instead of 70 minutes) for a prepbufr file to be available. The 70 minute setting led to several tm00 analyses after a prod machine switch on 8/2 to run without prepbufr files.
  • Modified the dependency checks for the tm00 TC relocation prelim job so that it would wait for the TC vitals file to be on the WCOSS dev machine.
    • 2016/07/18/18 S. Liu, Carley, Rogers
  • Made additional modifications to the dependency checks in the catchup cycle's partialcyc jobs to better ensure that the job waits for the previous catchup cycle to be finished.
  • Restarted land-states from NAMRR as of the tm06 analysis step
  • These three changes went in as of the 7/18/16 18z tm06 forecast and the 12z 7/18/16 tm02 forecast:
    1. Turn on use of heating tendencies derived from lightning data over the entire analysis domain (had been restricted to ocean only).
    2. Use a land mask to limit the usage of heating tendency derived from radar reflectivity over the ocean.
    3. During the DFI in the numerical model: Turn off the cloud analysis-derived heating tendency if the heating tendency from the microphysics gradually increases.
    • 2016/07/15/12 Ferrier, Rogers
  • Removed change on 5/26/2016 which gradually increased background diffusion with decreasing pressure.
    • 2016/07/13/18 Rogers, Carley
  • Modified dependency checks in the catchup cycle's partialcyc jobs for the CONUS/Alaska nest so that it is identical to that for the parent domain as described in the 7/2 change. Previously it waited less time than the parent partialcyc job to get the previous catchup cycle's land-states. Thus old land states were being used for these nests if NAMX was running late, as was often the case on weekdays.
    • 2016/07/13/00 Rogers, G. Lou, Carley
  • Resume running tropical cyclone relocation, done at tm06 and tm00 analysis times.
    • 2016/07/05/12 Rogers
  • Resume using NAM obs dumps (T+70 minutes) for the tm00 analyses.
    • 2016/07/04/18 Rogers
  • Restarted NAMX land states and satellite bias corrections from NAMRR as of the tm05 catchup cycle analysis step.
    • 2016/07/02/00 Rogers, Carley
  • Reinstated dependency checks in the catchup cycle's partialcyc jobs to ensure that the previous cycle's catchup cycle is finished before the new one starts to ensure that the current cycled land-states are used. This was mistakenly taken out of NAMX sometime in early June, and led in some instances to old land-states being contiuously used because the catchup cycle for XXZ did not wait for the XX-6Z XX-6Z cycle to have finished before it started.
    • 2016/07/01/12 Carley, Liu
  • Update to cloud analysis : Changed the prescribed latent heating time period for radar-derived temperature tendencies to 10 minutes for the CONUS nest (down from 20 minutes). This will have the effect of introducing larger heating rates during the DFI. This is effective as of the tm02 analysis in the catchup cycle.
    • 2016/06/30/12 Ferrier
  • Post-processing change: revert back to the operational cloud ceiling height algorithm instead of the Stoelinga & Warner (1999) algorthim that has been in place since May 2015.
    • 2016/06/29/18 W.-S. Wu
  • Disable use of high peaking channels from NPP ATMS (12,13,14,22) and METOP-B AMSUA (12,13). Change was effective as of the TM04 analysis associated with the 20160629 18Z cycle.
    • 2016/06/24/12 Ferrier
  • Turned off a term that was proportional to the standard deviation of the subgrid-scale terrain in the MYJPBL, which affected the vertical mixing of momentum and reduced wind speeds too much in areas with complex terrain.
    • 2016/06/21/18 Ferrier, Aligo
  • Turn off shallow convection in the 3 km nests
  • Microphysics changes:
    1. Adjusted assumed number concentrations and mean diameters of ice in stratiform areas to increase the area of light (<20 dBZ) composite reflectivities, based on comparisons to experiments with Thompson microphysics.
    2. Based on tuning against the Thompson microphysics tests, a slight increase in the area of 20-40 dBZ echoes associated with stratiform rainfall below the melting level. This may show up as slightly larger areas of <40 dBZ 1-km AGL reflectivities in stratiform areas away from convection.
    3. Fewer >50 dBZ reflectivity counts to match up better with the observations.
    4. Reduced areas of light (<20 dBZ) noisy radar echoes over the SE US, within warm conveyor belts, and over the adjacent Pacific and Atlantic oceans). The radar echoes are associated with rain that develops in thin clouds that form at the top of moist boundary layers. The new code treats this rain as drizzle, in which the sizes of the drops are reduced (which reduces their radar backscatter) and their number concentrations are increased based on output from the Thompson microphysics and results from a study on drizzle (Westbrook et al. 2010).
    5. A reduction in >1"/3h rainfall rates without diminishing the intensity of the 50 dBZ 1 km AGL reflectivity.
    • 2016/06/21/12 Ferrier
  • Change to the shallow convection for the 3 km nests from the 17 June change:
    1. Turn overshooting back on, so that when the average RH in the convecting cloud layer approaches saturation the cloud top is raised 1 level to mimic overshooting convection and to include air above cloud top into the cloud mixing.
    • 2016/06/20/12 Ferrier
  • Changes to the shallow convection for the 3 km nests from the 17 June change:
    1. Time-scale for shallow convection is 3600 s (originally 2400 s, revised to 4800 s in recent jun07-jun17 versions).
    2. Maximum depth of shallow convection is limited to <= 100 hPa.
    3. If the (older) stratocumulus convection fails, the (newer) congestus convection is tested.
    • 2016/06/17/18 Ferrier
  • Convective parameterization changed to only run the the old, stratocumulus component of shallow convection for the 3 km nests.
    • 2016/06/17/12 Ferrier
  • Changes to the shallow convection for the 3 km nests:
    1. Force the depth of shallow convection not to exceed 200 hPa.
    2. If the old stratocumulus shallow convection is not successful, try the new congestus convection only if the depth is between 150 & 200 hPa, otherwise exit shallow convection.
    3. Skip unnecessary deep convection calculations, which should speed up the code a little bit, using existing logic and code structure.
    • 2016/06/16/12 Ferrier, Janjic
  • Within the 3-km nests, the rate (time scale) of adjustments associated with shallow convection is reduced (increased) by a factor of two.
    • 2016/06/07/12 Carley, S. Liu, Rogers
  • Upgraded the real-time NAMRR catchup cycle with Shun's remaining cloud analysis changes. This includes re-enabling the use of lightning data, re-tuning the impact of lightning data over the ocean, and re-setting the radar derived latent heating time period back to 20 min (was 200 min). Change went into NAMRR parallel at 18z 6/6.
  • Shallow convection is turned back on in NAMX's 3 km nests.
    • 2016/05/26/12 Aligo, Carley, Ferrier, Janjic, Pyle, Rogers
  • Replace 12-h NDAS with 3-h analysis/forecast updates for the NAM parent with a 6-h assimilation cycle with hourly analyses/forecasts for the NAM 12 km parent and 3 km CONUS/Alaska nests (so called NAMRR "catchup" cycle). See the NAMRR parallel run change log for all changes specific to NAMRR over what was running in NAMX prevously. As part of this change, the parallel NAM nests are now run 4 times a day instead of just at 00z/12z.
  • NMMB model changes
    1. Shallow convection turned off in the 3 km nests
    2. Apply temperature and moisture adjustments - i.e. remove unstable and/or supersaturated layers.
    3. Gradually increase background diffusion with decreasing pressure.
    4. Introduce slight off-centering (w1=0.9 rather than 1.0) in subroutine adv2
    5. Calculate radar reflectivity directly when microphysics=fer_hires (refl_10cm array);
    6. Add precipitation rate to the list of maximum hourly fields (pratemax array);
    7. Add diagnostic variables to the 4D mprates array when lmprate=true.
    8. Added conditional checks so that calls to subroutine CLTEND are made only when nprecip>1.
    9. Increased variable FH from 1.01 to 1.10, which makes it consistent with what's in module_SF_JSFC.F90. The impact appears to require slightly more residual turbulence for determining the diagnostic PBL height.
    10. Fixed a small bug affecting cloud water riming onto ice (PIACW) by removing the air resistance correction to ice fall speeds from incorrectly being applied twice.
    11. Set CHKLOWQ=1 flag everywhere rather than having CHKLOWQ=0 when RH>95% at the lowest model level.
    12. Abort model if an inconsistency is detected between the input vegetation table and the ivegsrc configure file option (whether USGS or IGBP are used), as this inconsistency has tripped up multiple users. A similar check was also added for the soil table, though that is less likely to cause problems.
    • 2016/03/29/18 Jovic, Ferrier
  • Fixed code errors in the gravity wave drag (GWD)/mountain blocking module related to how the model (grid-relative) wind components are rotated to earth-relative before being sent to GWD parameterization scheme.
    • 2016/03/25/00 Ferrier, Pyle, Rogers
  • Turned off the filter applied to the vertical velocity used in the vertical advection of tracers that was put into NAMX at 12z 10/1/2015.
    • 2016/03/24/18 Rogers
  • Changed nphs, nprecip (frequency of physics calls) from 4 to 2 for the 12 km parent and all 3 km nests; from 9 to 4 for the 1.5 km fire weather nest.
    • 2016/03/23/12 Ferrier
  • Pulled out 12z 3/18/2016 changes due to failure of the 00z 3/22 run
  • Reinstated 18z 3/12/2016 change with the following modification: an additional check was put into the new, congestus-based shallow convection code when nodeep=T that will cause the scheme to exit and do nothing if the initial temperature corrections exceed 5C at any level in the cloud layer.
    • 2016/03/18/12 Ferrier
  • Changes to the convective parameterization:
    1. One of the changes that has been running in the NAMX parent since June 2015 was to force the code to run through different scenarios without aborting by setting newswap=T in the configure file, while setting newswap=F reverts back to what's currently run in operations. So first the old, stratocumulus type shallow convection is assessed (upward moisture and downward heat transports) and if that aborts then the new cumulus congestus (upward heat & moisture transports). There are multiple scenarios in which this can occur. One of last checks is to make sure the mean RH in the cloud is <95% (rhmean<0.95) otherwise it exits out of the column physics and "aborts" (i.e. convection is not triggered). However, the changes from June 2015 forced the code to proceed when newswap=T. This change removes this newswap=T check, such that if the mean RH is >95% then the code will exit.
    2. The 18z 3/12 change was intended to reduce the shallow cu triggering n the nests and the extent over which it is active by requiring the RH in the cloud not to be <90% at any given level. The 1 March change included a few more criteria to limit the cloud depth, which appeared to lead to the subsequent failures by failing to account for what happens when RH>>100%. So this change includes an extra check in the code that does not exit the cloud top search if rhmean is >95%, because it is possible that limiting the cloud depth too much acts to increase the mean RH in the cloud beyond reasonable values.
    • 2016/03/12/18 Ferrier
  • For the NAM nests running shalllow convection only (nodeep=true), the check on relative humidity (RH) for cloud top was changed from 75% to 90%.
    • 2016/03/07/00 Rogers
  • Pulled out this shalllow convection change in the parallel 3 km nests pending debugging of NAMX model failures at 12z 3/4 and 12z 3/6:
    1. Because the shallow convection appeared to be too active, its triggering was restricted by (1) limiting the cloud depths to <=0.3 times the normalized surface pressure (nominally 300 hPa when the surface pressure is 1013 hPa), and (2) requiring the relative humidity at every level within the cloud layer to be >=90%. To stabilize unstable layers that developed above midlevels, the search for the most unstable parcel (MUP, i.e. the parcel with the highest CAPE) was extended from 0.6 up to 0.2 times the surface pressure. In other words, the search for the MUP is now extended from above the surface up to 200 hPa rather than only up to 600 hPa (assuming the lowest model pressure is 1000 hPa).
    • 2016/03/01/00 Ferrier
  • Changes for the parallel 3 km nests:
    1. The 6 January 2016 changes turned on shallow convection only in the 3-km CONUS nest by setting nodeep=T (no deep convection) and by setting newswap=T (both parameters are in the configure file). The newswap parameter calls the "old" shallow convection first and if that aborts then calls the "new" shallow convection (see the 2015/06/04/18 log entry for further details). Now the shallow convection is also turned on in the other 3-km nests over Alaska, Hawaii, and Puerto Rico.
    2. Because the shallow convection appeared to be too active, its triggering was restricted by (1) limiting the cloud depths to <=0.3 times the normalized surface pressure (nominally 300 hPa when the surface pressure is 1013 hPa), and (2) requiring the relative humidity at every level within the cloud layer to be >=90%. To stabilize unstable layers that developed above midlevels, the search for the most unstable parcel (MUP, i.e. the parcel with the highest CAPE) was extended from 0.6 up to 0.2 times the surface pressure. In other words, the search for the MUP is now extended from above the surface up to 200 hPa rather than only up to 600 hPa (assuming the lowest model pressure is 1000 hPa).
    • 2016/02/04/12 Ferrier, Jovic
  • The forecast model was changed with updates to the handling/advection of tracers arrays, with the passive advection of Q now called every time step.
    • 2016/02/04/00 Rogers
  • Discovered that from 12z 1/28-12z 2/3, the 3 km parallel CONUS nest was not running with the shalloe convection change put in on 1/6. This has been fixed as of the 00z 2/4 cycle.
    • 2016/02/01/18 Collard
  • Turned back on Metop-A HIRS4 radiances.
    • 2016/01/15/00 Collard
  • Turned off Metop-A HIRS4 radiances dues to degraded data during engineering tests.
    • 2016/01/06/12 Ferrier
  • Turn on shallow convective parameterization in the 3 km CONUS nest by setting nodeep=T (shallow convection only) and newswap=T, which calls the "old" shallow convection first and if that aborts then calls the "new" shallow convection (same as in NAMX parent).
    • 2015/12/29/12 Rogers
  • Reduced the time step for the 1.5 km fire weather nest from 3 1/8 sec to 2 1/12 sec to eliminate a CFL instability problem in the 12z 12/17/2015 run.
    • 2015/12/19/00 Rogers
  • The extended-range burned area "accumulation" was changed back to 30 days to comform to the retention period of the burned area data in the production directory (/dcom/us007003/burned_area)
    • 2015/12/08/18 W.-S. Wu
  • GSI analysis fix for the 11/26 failure : a bug in subroutine sonde_ext in file read_prepbufr.f90 was found. The check to prevent array out of bounds was in place but after printing an error message the program was stopped. A fix was made to return to the calling routine instead of stopping execution, which would allow the analysis to continue to run. With this code change, raob level enhancement has been turned back on.
    • 2015/11/26/18 Carley, Rogers
  • Turned off raob level enhancement (by changing “ext_sonde=true” to “ext_sonde=false”) in the GSI analysis; this change fixed the failure of the ops and parallel 12z 11/26/2015 NAM analysis.
    • 2015/11/25/00 Gayno, Pyle, Rogers
  • Reduced smoothing of the parallel NAM nests surface terrain:
    1. Old settings: 6 passes of the peak smoother, 4 passes of the smoother/desmoother
    2. New settings: peak smoother turned off, one pass of the smoother/desmoother
    • 2015/11/24/18 W.S. Wu, Rogers
  • Turned on Himawari satellite wind data
    • 2015/10/09/12 W.S. Wu
  • New observations types added to GSI analysis:
    1. New satelllite winds
      1. MTSAT2 IMAGER WVct AMVs (JMA)
      2. 254 54 M-7 IMAGER WVct AMVs
      3. M-10 IMAGER WVct AMVs
      4. NOAA -15 AVHRR IR AMVs
      5. NOAA -18 AVHRR IR AMVs
      6. NOAA -19 AVHRR IR AMVs
      7. METOP-A AVHRR IR AMVs
      8. METOP-B AVHRR IR AMVs
    2. New GPS Radio Occultation Data
      1. METOP-B 3 (subtype)
    3. New Satellite radiance data
      1. M10 Seviri
      2. metop-b hirs4 (moni), amsua, mhs, iasi
      3. npp atms, cris
      4. f17 ssmis
    • 2015/10/07/12 Rogers
  • Convective parameterization ("BMJ-lite") was turned off in the 3 km CONUS nest as of this cycle. It was turned off in the ops NAM 4 km CONUS nest at 12z 10/7
    • 2015/10/02/12 Rogers
  • Turned back on "BMJ-lite" in the 3 km CONUS nest to address failures due to model instability associated with Hurricane Joaquin. This was also turned back on in the ops NAM 4 km CONUS nest at 18z 10/2
    • 2015/10/01/12 Black
  • A simple filter is applied to the vertical velocity used in the vertical advection of tracers. This is to address issues associated with unphysical moisture profiles which, in addition to unphysicality, may also be problematic elsewhere within the model (e.g. radiation). It is worth oting that the issue is generally uncommon, but has caused stability problems in model (notably, hte prod NAM failures at 18z 9/29/15 and 06z 10/1/15
    • 2015/09/16/18 G. P. Lou, Rogers
  • Turn on tropical cyclone relocation in the NDAS tm06 and NAM tm00 first guess restart files.
    • 2015/09/11/18 Y. Wu
  • A change was made to the Noah LSM to consider the effect of frozen soil on plant transpiration and soil evaporation, specifically:
    1. Change SMCREF to SMCREF = SMCREF + ICE or SMCREF = SMCREF + (SMC(1)-SH2O(1))
    2. Change GX = (SMC (1) - SMCWLT) / (SMCREF - SMCWLT) to GX = (SH2O (1) - SMCWLT) / (SMCREF - SMCWLT)
    3. Change SRATIO = (SMC(1) - SMCDRY) / (SMCMAX - SMCDRY) to SRATIO = (SH2O(1) - SMCDRY) / (SMCMAX - SMCDRY)
    4. Change (a) and (b) will increase canopy resistance and reduce plant transpiration
    5. Change (c) will reduce direct evaporation from soil.
  • The above changes are based on the assumption that ice is not available to plants and can not be evaporated.
    • 2015/08/06/12 Gayno, Y. Wu
  • Begin use of NESDIS burned area data in the parallel NAM fire weather nest. Two "accumulation" burned area files are created: 2-day and 30-day (changed to 45-day on 8/14/2015). The greenness fraction and albedo is adjusted according to the 45-day data and the top layer soil moisture according to the 2-day data.
    • 2015/08/02/00 Rogers
  • Restarted cycled land states from the operational NDAS
    • 2015/07/30/18 Ferrier, H-M. Lin
  • Changes to the radiation / microphysics:
    1. The cloud droplet effective radius is calculated assuming a monodisperse droplet distribution. It is also not allowed to be less than minimum a value of 5 microns, which is reached when cloud water contents are less than 0.1045 g m^-3. The droplet effective radius is no longer forced to be between 10 and 15 microns. The change is anticipated to reduce incoming surface shortwave fluxes under liquid clouds and reduce daytime 2 m warm biases.
    2. The assumed cloud droplet number concentration within radiation is no longer assumed to vary with temperature. Instead it is assumed to be a fixed value of 200 cm^-3, the same as what is assumed in the FA microphysics.
    • 2015/07/30/12 Rogers
  • 1.5 km Fire weather nest turned on in NAMX as of this cycle
    • 2015/07/25/00 Ferrier, Aligo
  • Changed code so that Rhgrd (RH threshold for onset of condensation) is set in the configure file so it can vary for different domains; current settings are RHgrd=0.98 for the 12 km parent domain, RHgrd=1.0 for the 3 km nests
  • Includes a cloud water mixing ratio threshold for the autoconversion of cloud water to rain.
    • 2015/06/26/00 Ek, Y. Wu, Ferrier, Aligo
  • Removed the limit on the top-layer soil moisture to 80% of SMCREF (the soil moisture threshold below which transpiration begins to stress).
    • 2015/06/17/12 Gayno, Y. Wu
  • Land-sea mask for 3 km CONUS and Alaska nest changed to add all lakes resolved by the new fresh water lake (FLAKE) climatology. Water temperatures at "FLAKE" lake points are a blend using a Cressman analysis of the FLAKE climatology and temperatures at nearby water points resolved by the RTG_SST_HR analysis.
    • 2015/06/04/18 Ferrier
  • Convection changes in the NAMX 12 km parent domain:
    1. Deep convection is turned back on, but its triggering has been reduced compared to what is running in production.
    2. As with the 2015/05/22 change, when calculating the initial cloud-top level, it is located at the level where the CAPE is a maximum. The code change also includes a check on relative humidity (RH), such that the cloud top will also be when the RH falls below 75%. It was made to limit the vertical extent of the convection to prevent it from penetrating elevated dry layers.
    3. As with the 2015/05/22 change, the final, revised estimate for the cloud top is made when the mean RH of the cloud layer exceeds a threshold value (rhshmax) and the cloud top is raised. The cloud top is raised only when the parcel is positively buoyant (dtv>0), plus it is raised one more level above (overshoot) if the parcel's CAPE>0 to emulate a slight overshooting of parcels above the level of neutral buoyancy. (The variable rhsmax is equivalent to the mean RH in the cloud column associated with lifting parcels up 25 hPa to reach saturation.)
    4. The new logical variable (newshal) was removed. As with the 2015/05/22 change, several IF tests were changed and added so that the old shallow convection is calculated first (heating/drying below, cooling/moistening aloft). If the old shallow convection aborts, then rather than exiting the new shallow convection is calculated (cooling/drying below, warming/moistening aloft) in an attempt to maintain thermodynamic stability.
    5. The code changes only take effect if newswap=T, otherwise the code should behave like the existing trunk (and production) code.
    • 2015/06/04/12 Rogers
  • 3 km Hawaii and Puerto Rico nests are turned on in NAMX as of this cycle
    • 2015/05/29/18 Rogers
  • Due to NAMX configure files not being mirrored to Tide, the 12z NDASX/NAMX and the 18z NDASX did not run with the changes to NODEEP, NEWALL, NEWSWAP, and NEWUPUP settings described in the 18z 5/22 log entry. This was fixed as of the 18z 5/29 NAMX forecast.
    • 2015/05/28/18 Rogers
  • When the 5/22/2015 convection changes went into NAMX, these Noah LSM changes were inadvertently removed:
    1. Limit top-layer soil moisture to 80% of SMCREF (soil moisture threshold below which transpiration begins to stress)
    2. Threshold snow depth (water equivalent in meters) that implies 100% snow coverage is increased by 4x
  • The Noah LSM changes were restored in NAMX as of this cycle
    • 2015/05/22/18 Ferrier, Aligo
  • Convection changes in the NAMX 12 km parent domain:
    1. Set the following in the model configure file
      1. NODEEP=.true.
      2. NEWALL=.false.
      3. NEWSWAP=.true.
      4. NEWUPUP=.true.
      5. Code changes:
        1. When calculating the initial cloud-top level in the trunk code, it is located at the level where the CAPE is a maximum. The code change also includes a check on relative humidity (RH), such that the cloud top will also be when the RH falls below 80%. It was made to limit the vertical extent of the convection to prevent it from penetrating elevated dry layers.
        2. The final, revised estimate for the cloud top is made when the mean RH of the cloud layer exceeds a threshold value (rhshmax) and the cloud top is raised. The cloud top is raised only when the parcel is positively buoyant (dtv>0), plus it is raised one more level above (overshoot) if the parcel's CAPE>0 to emulate a slight overshooting of parcels above the level of neutral buoyancy. (The variable rhsmax is equivalent to the mean RH in the cloud column associated with lifting parcels up 25 hPa to reach saturation.)
        3. A new logical variable (newshal) is true if configure file variables nodeep=T and newupup=T. Several IF tests were changed and added so that the old shallow convection is calculated first (heating/drying below, cooling/moistening aloft). If the old shallow convection aborts, then rather than exiting the new shallow convection is calculated (cooling/drying below, warming/moistening aloft) in an attempt to maintain thermodynamic stability.
        4. The code changes only take effect if nodeep=T. If nodeep=F, the code should behave like the existing trunk code.
    • 2015/05/01/00 Ferrier
  • Post-processing change: the cloud ceiling height algorithm was changed to just use the value computed in the Stoelinga & Warner (1999) algorthim rather than the lower of the Stoelinga & Warner value and the lowest level at which a 5/8 cloud fraction is found using the modified GSD algorithm.
    • 2015/04/28/00 Ferrier
  • Post-processing changes:
    1. A modification to the cloud ceiling heights to go from AGL to MSL.
    2. The cloud ceiling height algorithm was refined so that the lower of two height estimates are used.
      1. The vertical visibility following Stoelinga & Warner (1999) or
      2. The lowest level at which a 5/8 cloud fraction is found using the modified GSD algorithm.
    3. Use a modified version of the GSD algorthim to compute low, middle, high, and total (instantaneous) cloud fractions, which is more consistent with a "celestial dome" approach.

      The GSD algorithm looks over a 30-km radius, but really it is a box that is 60 km on a side that is centered over each grid point. Each model grid box is either clear (0 cloud fraction) or cloudy (1 cloud fraction), and it is cloudy if the sum of cloud water + cloud ice mixing ratios exceed 1.e-7 kg/kg. The cloud fraction becomes the ratio of cloudy boxes to the total number of boxes within the (60 km)**2 area.

      The two modifications to the GSD algorthim:

      1. The threshold described above is made grid-resolution dependent, so the 1.e-7 kg/kg threshold is used for a 12-km grid box (like the parent), whereas it is increased for higher resolution grids as a function of (12 km/DX)**2, where DX is the resolution for the other domains.
      2. The cloud fractions for each grid box are not 0 or 1, but instead they are a simple linear function of the mixing ratio such that a cloud fraction of 50% is associated with the 1.e-7 kg/kg threshold.
    • 2015/04/08/12 Carley
  • The relative weight assigned to the static and ensemble background covariance in the hybrid GSI was changed from 50% each to 25% static/75% ensemble
    • 2015/04/07/12 Rogers, Ferrier
  • The fundamental model timestep was changed:
    1. 12 km parent domain : from 26.66 sec to 25 sec
    2. 3 km nests : from 6.666 sec to 6.25 sec
    3. 1.5 km nests : from 3.333 sec to 3.125 sec
    • 2015/04/06/12 Aligo, Carley, Ferrier, Janjic, H-M. Lin, Y. Wu
  • Model/Analysis Changes:
    1. The increase in the threshold snow depth (water equivalent in meters) that implies 100% snow coverage is reduced from 7x over ops NAM to 4x
    2. The PBL physics were changed so that there is enhanced mixing at the top of the PBL under moist conditions where (1) the theta-e decreases with height and (2) the air is close to saturation.
    3. GSI code fixed so that it does not abort reading in the ensemble members for the hybrid analysis if the NMMB grid is square (IM==JM).
    • 2015/03/27/18 Liu, Carley, Wu, Y. Lin, Rogers
  • The assimilation of cloud/radar data was not being done because the parallel NAM GSI code was not compiled with the GSD cloud analysis library. The following configuration has been set up to run with the use of radar-derived temperature tendencies as was intended:
    1. The parallel GSI code is compiled with the GSD cloud analysis library (gsdcloud4nmmb)
    2. At NDAS tm12 time, with the first guess always from the GDAS, we do not use reflectivity in the GSI and we do not do a cloud analysis. The radar_init parameter is set to 0 in the NMMB model configure file. The filt.var.txt file at tm12 has CW set to "R" (for restore)
    3. For tm09 through tm00 times, the analysis checks to see if the radar data is available. If it is, then use_reflectivity=true in the GSI namelist and the radar-derived temoerature tendencies are computed and used in the NMMB forecast's digital filter, and the model uses a filt.var.txt file has CW set to "F" (filter)
    4. For tm09 through tm00 times, if the radar data is not available, then the USEREF variable is false and no cloud/radar assimilation is done, and a filt.var.txt file that has CW set to "R" is used in the model.
    • 2015/03/27/00 Rogers
  • Began running a 3 km CONUS and 3 km Alaska nest in the 00z NAMX run
    • 2015/03/18/18 Ek, Y. Wu, Ferrier, Aligo
  • Changes to the Noah LSM code:
    1. Limit top-layer soil moisture to 80% of SMCREF (soil moisture threshold below which transpiration begins to stress)
    2. Threshold snow depth (water equivalent in meters) that implies 100% snow coverage is increased by 7x
    • 2015/03/16/06 Gayno, Rogers
  • Turn on use of AFWA snow depth analysis using the "envelope" adjustment method, which is done by setting the merge_coeff_snow_depth paramter to a negative number (whole number or fractional). For merge_coeff_snow_depth=-2 (the setting used in the parallel), the envelope bounds will be one-half to twice the depth from the AFWA analysis. If the first guess depth is within the envelope, the snow is cycled. If it is outside the envelope, the first guess depth is adjusted to the edge of the envelope.
    • 2015/03/11/18 Ferrier
  • Post-processor changed to use a new ceiling height computation based on Stoelinga & Warner (1999, JAM) computing ceiling heights as a vertical visibility
    • 2015/03/03/12 X. Wang, Rogers
  • Item #1 in the 1/25/2015 entry below (digital filter fix to apply the filter to all 61 PINT values instead of at levels 1-60 only (excluding the surface) in the NDAS forecast) was implemented in the operational NDAS as of this date.
    • 2015/02/27/12 Rogers
  • Changes to the frequency of physics calls in NAMX
    1. In the parent 12 km domain, call the radiation every 20 minutes instead of every hour
    2. For the parent 12 km domain and nests, call the physics (nphs/nprecip) every 4 model timesteps instead of every 6 timesteps.
    • 2015/02/25/12 Rogers
  • Began running a 3 km CONUS and 3 km Alaska nest in the 12z NAMX run
    • 2015/01/28/18 Wu
  • Switched to using the top of the GSI analysis trunk code (r50280), which has changes in it which allows the code to use the full T574 EnKF members, instead of having to truncate them to T254 as is done for the ops NAM GSI.
    • 2015/01/27/12 Rogers
  • Discovered that in the ops NDAS, the GEFS is used instead of the global EnKF for all NDAS analyses valid at 03/09/15/21z, because in the ops EnKF job only the 6-h ensemble forecasts (valid at 00/06/12/18z) are being truncated to T254. To remedy this, prior to each parallel NDAS tm09 and tm03 analysis, a job is run to truncate the ops T574 EnKF members to T254.
    • 2015/01/26/12 Rogers, Janjic, Ferrier, Aligo, Pyle, H-M Lin, Carley
  • First run of NAM parallel (12 km domain only) with the following changes
    1. Digital filter fix to apply the filter to all 61 PINT values instead of at levels 1-60 only (excluding the surface) in the NDAS forecast
    2. Turn off 5x increase in divergence damping in the 3-h NDAS forecasts
    3. The XMIexp fix (change from INTEGER to REAL declaration) in module_DIAGNOSE
    4. The "mixing length" bug fix in subroutine MIXLEN in phys/module_BL_MYJPBL.F90
    5. Fix to read in radar-derived temperature tendencies for the diabatic digital filter if USE_RADAR=true
    6. Fix to the computation of max 1-h updraft helicity in module_DIAGNOSE.F90
    7. Run digital filter w/radar-derived temperature tendencies for the diabatic digital filter in the NAM forecast (currently only done in the NDAS forecast)
    8. New shallow convection
    9. Post-processor fix to computation of instantaneous updraft helicity