When |
What |
2019/06/02/12 |
Parallel is temporarily shut off, it will be resumed when the lateral boundary imbalances
seen in the Fv3SAR DA run are fixed.
|
2019/05/17/00 |
Changed nsst setting from "nstf_name = 2,1,1,0,5", to "nstf_name = 2,0,0,0,0"
Changing the second flag from 1 to 0 means the NSSTs will be warm started off of the global
model instead of cold started / spun up.
|
2019/04/12/12 |
Switch to using the hybrid EDMF PBL scheme that is
being used in the current operational GFS.
|
2019/03/13/00 |
High resolution (~3-km) tiled land-surface data is now used instead of the ~13 km Gaussian
grids.
|
2019/03/05/00 |
Two model changes:
- In the current FV3 model, the snow flag is set to 1 when any of the
snow, ice and graupel produced from GFDLMP is greater than zero. To correct
for excessive forecast snowfall, the snow/rain flag will be set to be the
ratio of sum of snow, ice, graupel and convective snow to total precipitation.
In the land surface code, the sfc_drv.f routine is modified to pass fractional
srflag to sflx.f; in sflx.f, the precip amount is separated into rain and
snow based on srflag and the difference of specific heat capacity for ice and
water is accounted for based on srflag. The sfc_sice.f routine is also modified to
separate rain and snow, and the snow/rain difference is accounted for potential
evaporation.
- In the current GFDL microphysics, all hydrometeor species (liquid cloud, ice,
rain, snow, graupel) are combined and re-partitioned and grouped based solely by
temperature into ice and liquid before they are used in the radiation calculation.
An alternate and more physically based method is implemented to use hydrometeor
information from the GFDL microphysics. The individual hydrometeor species and
their corresponding radii from the GFDL MP are fed into the radiation.
|
2019/02/13/12 |
The GSI analysis was not using GOES radiances because it was not getting the correct
GOES file from the RAP. The scripts were changed to copy the correct file.
run.
|
2019/02/11/12 |
Changed full_zs_filter from true to false in input.nml_firstguess_writecomp (namelist
file used in 6-h forecast off FV3GFS which coldstarts DA cycle) so it is consisent with Ben's FV3REG
run.
|
2018/12/11/12 |
Now running w/k_split, n_split=6 in the FV3SAR namelist, as the 12z 12/10 run
failed at 58-h with k_split=5 and n_split=6.
|
2018/11/20/00 |
Using the latest version of the FV3SAR and CHGRES code which puts MP tracers
from the Fv3GFS microphysics into the boundary condition files read into the FV3SAR
Running with k_split=5 and n_split=6 (changed from 4 and 5, respectively) in the
FV3SAR namelist.
Run moved to the Dell
|
2018/10/24/00 |
The central latitude of the domain was changed from 35.5N to 38.5N. This change will
allow the 3 km HRRR output grid to have no undefined grid points outside the FV3 integration
domain.
|
'
2018/10/12/00 |
Turn on aircraft data QC in GSI analysis
|
2018/10/07/00 |
Official start of FV3 standalone regional (FV3-SAR) 3km CONUS run initialized with a 6-h DA cycle
(currently running on WCOSS Phase 2)
Forecast model details
- 3 km standalone regional FV3, running same physics as FV3GFS (w/GFDL microphysics) but
with no convective parameterization
- Same vertical layer structure as FV3GFS
- 60-h forecast from 00z/12z initial conditions with hourly output on the 3 km HRRR grid
- Using 3-hourly FV3GFS Gaussian grid files from the same cycle for lateral boundary
conditions
Data Assmilation details
- Regional FV3-GSI analysis
- Hourly DA cycle:
- Run 6-h FV3SAR forecast from tm12 to tm06 from FV3GFS IC's
- Use 6-h FV3SAR forecast valid at tm06 for the first guess to the tm06 FV3GSI analysis
- Run 1-h FV3SAR forecast and hourly analyses from tm06-->tm00 to create initial conditions
for the 60-h forecast
- FV3GSI uses RAP observations and FV3GDAS ENKF members for hybrid analysis
- No cycling of FV3SAR soil states
|