This change log starts from October 2003. The starting point for the RUC13 was the code used in the dev RUC20. It runs out to only 12 h as of Feb 2003.
HYBCST, HYBPOST, HYBCST_PRE, Interp
Started running 13km RUC forecasts initialized
with 20km dev2 analyses interpolated to 13km resolution
via interp program (written by Tanya Smirnova)
HYBFRONT - First 13km version
Started running analysis from 13km 3h
forecast background. However, we are still not
cycling yet while we check out performance of the
13km hybfront via web products and verification.
HYBFRONT - revised version
RUC cycle
Began cycling 13km RUC on a 3 hour cycle.
RUC cycle
Began cycling 13km RUC on a 1 hour cycle.
RUC cycle
Delayed cycle time for 13km RUC to T+:45 for non synoptic times
and T+1:03 at 00Z and 12Z to capture additional observations.
HYBCST - coord.F
Added negative check for moisture variables.
Also, the snow (model glacier) in British Columbia is zeroed out
to fit the Canadian analysis at this time of the year.
HYBFRONT - readdrop.f, gtdrop.f, hybing.f, insertd.f, gtpirp.f,
readacars.f, gtprof.f, hgt_index_sort_hq.f, dropin.f, readgoes.f,
getgoes.f, rd_cld.f, profin.f, getcdw.f, hybcloud.f, MAPSCON, getvad.f,
readvad.f, gtrass.f, gtraob.f
Data ingest update- added dropsonde data, check in ACARS for
data source (to screen TAMDAR data), increased bounds in profiler
ingest, updated satellite PW ingest for new format, fixed names of
satellite data to all reflect west and east
instead of satellite number, fixed station names in VAD and RASS,
and quality control in gtrass, added go back an hour to gtraob.f.
Fixed hybcloud to use 2 instead of 1 for minimum layer- have do loops
that subtract one from index, so can't start lt 2. Also changed
script to look for PW in new directory and cleaned up satellite
soft links to just use E and W instead of satellite numbers (got rid
of all references to GOES 8, it's GOES 12 now!)
HYBFRONT - hybanl_qc.f, mk_pbl_obs.f, exsfcobs.f,gtresids_pw.f,
hybcloud.f$
Moved in all hybfront improvements from dev and dev2 runs earlier
in spring and summer, designed for improved moisture analysis,
PBL-based assimilation of sfc obs, and
QC for precipitable water obs.
HYBCST - cup*.F,exmoisg.F,paramr.F,physics_trial.F
Included new version of Grell/Devenyi cumulus parameterization
also implemented into the 20km dev and dev2 RUC model versions
earlier this summer.
HYBFRONT - hybhqc.f, gtdrop.f
Changed QC so that dropsondes are QC'd (were not before
due to previous experimental version). Also, fixed gtdrop.f
so that no drops with sfc pressure less than 950 will be used
(considered to be below the representativeness of the analysis).
HYBPOST - calvis.f, hybint.f, hb2mw.f
Updated visibility calculation to latest dev2 version,
including fixes for nighttime overforecasting of diagnosed fog.
Also, visibility diagnostic is not smoothed as of this point.
HYBFRONT - sfcin.f, hybhqc.f
Use only the first observation from any given surface station
each hour. This forces the analysis to use only one
ob per station, also the observation closest to the analysis
valid time.
Also, increase QC threshold by factor of 2x (3x for 21z run)
for dropsonde data in buddy check.
HYBCST, HYBCST_PRE -
Increased number of time steps for initialization from
80 in each direction to 120 (40 min, instead of 26 min).
Code is changed so that line in HYBCSTIN is no longer used.
The code now figures out the number of time steps needed to
cover the same number of minutes are previously used in the
40km and 20km versions of the RUC.
scripts for HYBFRONT (ruc13env)
Remove access to radar reflectivity and lightning data
to match data to be available at NCEP.
HYBPOST - hb2p, mapsout_post.f
Increase smoothing factors for 13km to match effect
from smoothing at 20km. For MSLP and all isobaric level fields.
HYBFRONT - gtresids_pw.f, hybanl_qc.f, hybcloud.f, sfcin.f
Incorporate all changes in dev2 version of hybfront since
14 Sept 04.
HYBFRONT - hybanl_qc.f, hybcloud.f
Allow degree of supersaturation in background (usually 1h) forecast
to exist also in analysis, conserving the moisture from the 1h forecast
in precipitating areas.
This supersaturation is due to the call to output
2-min after the last call to the cloud physics routine.
Also, BR and low vis are allowed to cause building of
cloud water near the ground, even if there is precipitation
(previously not done in presence of precipitation).
HYBFRONT - hybcloud.f, gtresids_pw.f
Begin to allow clearing down to ground again over oceans and
over land if zenith angle is greater than 0.10. This reverses last
winter's decision to suspect GOES cloud/no-cloud quality below ~620 mb
in all conditions.
In gtresids_pw, put in a small fix to avoid dividing zero
in calculating statistics at the end of the routine.
HYBFRONT - hybcloud.f
Extends fix from 2 days ago to allow building over oceans
from GOES data and
over land if zenith angle is greater than 0.10.
HYBFRONT - hybcloud.f
Clears from METAR-clear obs now only up to
3650m (3658m = 12,000 ft, ceilometer limit)
instead of 4000 m. This gets rid of some of the
holes showing up in cloud analyses.
HYBFRONT - sfcin.f
Set obs (including OCLD and OWX) to missing values,
now in separate loop ahead of the 1050 loop,
instead of inside 1050 loop.
This avoided mesonet data erroneously showing
up with CLR cloud observations.
HYBFRONT - hybcloud.f
Now puts in ceiling at constant height from
METARs instead of constant height above ground level.
HYBCST - inithybv.F
Increase diffusion constant in upper-level sponge
from 0.0015 to 0.0025.
HYBPOST - tropwnd.f
Within diagnosis of PV fields including tropopause
levels, do not allow negative absolute vorticity.
topomini13.dat
Created a new 13km topomini file, using 2-pass
analysis with METAR obs, similar to that done previously
for the 20km RUC. Now, eliminated multiple obs
within hourly file.
Previously, the 13km topomini was equivalent to the
13km topoenvl (main terrain file), since the WRF had not
allowed the sub-resolution hi-res terrain (previously 10km)
from which the initial topomini was derived via trncmp.
Therefore, the previous 13km topoenvl file did not fit the
METAR elevations as closely as the 20km topomini file.
At least now the METAR elevations are fit a little more
closely.
HYBCST (cup_forcing_ens.F)
Change in the convective scheme to reduce the
coastal precipitation. Now uses mean vertical motion
from level 2 to level of free convection (rather than
using just using 3 discrete levels).
HYBPOST (mapsout_post.f)
Re-integrate heights/Montg vertically to ensure
that height, pressure, and theta-v are consistent.
They were previously not, due to writing of output
without a final re-integration of heights since the
last variation of theta-v in coord.F.
HYBFRONT (hybanl_qc.f, hb2p.f, linint.f,
gtresids_p.f, MAPSCON)
Introduce use of isobaric background for z/u/v
instead of native coordinate data. Now z is smoothed,
as well as u/v. Only applied above 700 mb.
HYBPOST (hybint.f)
Restore pass for smoothing of forecast
visibility grids.
HYBFRONT (sfcin.f)
Fixed bug in sfcin.f with the wrong index
for assigning vis obs. [Also, inadvertently
eliminated changes from 10 Nov 00z
with isobaric background for z/u/v resids.]
HYBFRONT
Now hybfront includes changes from both
12 Nov and 10 Nov.
HYBFRONT
Soil moisture and temperature are replaced by the ones
from back-up RUC. Hopefully this will help to improve surface
verifications.
(The goal was to replace them from operational RUC, but the
*grib files from NCEP cannot be read by our wgrib).
HYBPOST (calvis.f)
Returned to using average of lowest 4 levels
for precip hydrometeors only (rain, snow,
graupel). Gives much more reasonable vis
diagnostics. Continue to use only lowest
2 levels for cloud water and ice, which remains
an improvement over the current oper version
of calvis.f with its excessive low vis.
HYBCST - inithybv.F, PARAME
Increase diffusion constant in upper-level sponge
from 0.0025 to 0.005. KHMAX is increased from 1/32 to 1/75.
NSPGK in PARAME is increased from 6 to 7.
HYBCST - swrad.F
Changed weights of hydrometeors in the computation of
the water path. The intent is to increase the downward
short wave radiation in the cloudy areas and areas with
precipitation.
HYBFRONT - SETOFEQS
Increased raob obs errors - height up by 20%,
wind up by 10%(at 700 mb) to 60% (at 150 mb).
HYBCST - PARAME
Number of levels for upper-level sponge decreased from
7 back to 6 levels.
HYBCST - smolar2.F
Smolar2 is replaced with a more computationally
efficient version provided by Jacques Middlecoff.
Possible speed-up - ~2%.
HYBPOST - calvis.f
Now uses max value of cloud water and ice mixing
ratio over lowest 2 levels
instead of mean value over lowest 2 levels.
HYBCST - swrad.f
Change to same coefficients for longwave upward
and downward fluxes, as RRTM but unlike the previous
lwrad values (now 0.144 for both, previously
0.130 and 0.158 for the two values).
HYBPOST - topomini.dat
Completed using new topomini based on 3.3km
fine topography file. The new file uses min of 4x4
file topo values for each 13km value.
HYBFRONT - MAPSCON, hybanl_qc.f
Set up obs error factor as a function of
obs sub-types (e.g., Mesonet - 515, VAD - 210).
The factor for both of these ob types only
were set as 1.5. All other obs have errors
set as a function of NSRC ob types in SETOFEQS.
HYBCST - swrad.F
Changed weights of rain and graupel from 0.05
to 0.02 - the same weight as for snow.
HYBCST - swrad.F
Changed weight of ice in the optical path from 0.1 to 0.05
HYBFRONT - hybanl_qc.f
Fixed coefficients for BAL_NP and moisture 3dvar variable
which had been set for 20km, not 13km.
HYBCST - soilvegin1.F
Frost will be computed based on saturation over ice
(TBQ is change for temperatures below freezing).
HYBFRONT - hybcloud.f -
Fixed a bug with index incorrectly defined.
Now clearing above GOES cloud top should work correctly.
HYBFRONT - hybcloud.f -
Fixed another bug with index incorrectly defined.
Now the stability check should work more properly.
HYBPOST - getsta.f, profout.f
Improved check for BUFR surface data to ensure that
land stations are using land-use grid points, and water
stations (buoys) are using water grid points.
HYBFRONT - hybcloud.f -
Change lowest zenith angle to consider 'daytime'
treatment of GOES cloud data from 0.10 to 0.20.
Added variable to control this value throughout routine,
rather than having to change hard-wired values
found in many lines of code.
Removed unused code and writing of interim cloud
arrays used for debugging.
The RUC-13 cycle is restarted from dev2 RUC-20.
The model was crashing due to a bad first guess
field caused by problems with boundary conditions.
HYBCST_PRE - inithybv_pre.f
The print of SST/TSO through the Great Lakes
region was correct to use grid points relative
to 13km grid.
HYBFRONT - hybcloud.
Correct to use stability thresholds for adding
cloud with more conservative limit (3K/100mb) near
surface instead of 2K/100 mb).
HYBCST - PMICRPH, paramr.F, exmoisg.F
New version of exmoisg.F from G. Thompson.
HYBFRONT - sfcin.f, hybcloud.f
Add 'unknown' precipitation type to
list of current weather precip types
in which deeper cloud is built in hybcloud.
HYBPOST - writeg_coarse.f
Change to a vertical level-dependency
for number of smooth passes.
HYBPOST - hybint.f
Modification to precip type algorithms.
Rain - indicate now for 1.e-3 instead of
1.e-2 cm/h rate. This increases area of rain
area and freezing rain area. Was too limited
before, and definitely inconsistent w/ snow coverage.
Graupel vs. RA/FZ types - Implemented a
choice based on predominant p-type. Now
use graupel instead of rain if graupel rate
> rain rate. Choose rain (or FZ) if
rain rate > 4x graupel rate.
Was previously getting almost 100% same
coverage for FZ and IP, so this change produces
now a reasonable overlap but also a better
distinction.
for number of smooth passes.
HYBCST - exmoisg.F
Further revision of exmoisg.F after
further meeting between Greg Thompson,
Roy Rasmussen and John Brown.
HYBPOST - write_grib.f, tropwnd.f
Fix proper dx/dy setting for 20km and 40km
output from RUC13. Also, now use grid ID 130,
set by NCEP for RUC13 grid.
Also, fix a rare possible
crash condition in tropwnd.f to avoid a divide
by zero.
HYBFRONT - hybanl_qc.f, humisubs.f, gtresids.f, gtresids_pw.f,
gtresids_sfc_tquv.f, hybhqc.f, VARPRMS
Change moisture control variable from ln-qv
(log of water vapor mixing ratio) to pseudo-RH
(RH relative to background temperature/pressure field).
This version now combines assimilation of in situ moisture
observations (surface and raob) and precipitable water
into a single variational calculation of the 3-d moisture
analysis increment (correction from background).
HYBFRONT - rd_cld.f, hybcloud.f -
Fix some problems with the cloud analysis,
including improving coherence of GOES cloud top data
by mapping fields-of-view onto a slightly larger RUC13
volume, and fixing to ensure that METAR-related cloud
clearing is relative to a constant height rather than
height above model terrain.
HYBPOST - writeg_coarse20.f writeg_coarse.f
Remove horizontal smoothing of 2m dewpoint.
HYBFRONT - humisubs.f, hybanl_qc.f, gtresids_pw.f
Improved treatment of precipitable water, including
use of using 3-d weighting of background error and
observation errors using background saturation value
such that precipitable water assimilation has an
implicit vertical weighting. Lots of diagnostics
added also. Also, some revision of limits on
moisture analysis increment fields.
HYBPOST - mapsout_post.f
Use specific humidity, not mixing ratio,
for precipitable water calculation.
HYBFRONT - profin.f, hybanl_qc.f
Fixes to arguments/parameters for profin.f (called
by hybing.f) and call to hybcloud (from hybanl_qc.f).
HYBCST - hybcstdv.F
Comment out call to mapsout after completion
of initialization.
HYBCST - sfctmp.F
Corrected computation of snow albedo in the areas of thin
snow cover (below 10 cm). The albedo in such areas should
be interpolated between snow-free albedo for the land-use type
(not defined before the correction) and snow albedo equal to 0.7.
Now the snow-free albedo IS defined. The correction should
increase the albedo in such areas and make them colder.
Probably thin layers of snow will stay slightly longer now
than before the correction.
HYBFRONT - gtprof.f, profin.f ,hybhqc.f
Fixed gtprof.f to have number of levels for external
wind profiler data at 50 (maximum number in netcdf files),
having larger number was causing array problems. Number
of levels in the netcdf files were not reliable, so just
counting them internally (qc seems to be reliable). Also
added output for otype 204 (external profiler), and added
print of otype in hybhqc output to better track what obs
we are throwing out.
HYBFRONT - getmnet.f
Changed to not use mesonet wind obs until problems are
fixed.
HYBFRONT - humisubs.f
Changed scaling for OEQ and FGQ from qv_sat to
sqrt(qv_sat).
HYBFRONT - hybanl_qc.f
Reset ob location in BUFTIN and BUFQIN from
oi2/oj2/ok2 just before univariate analysis
section. This is a big fix.
HYBCST - inithybv.F, sfctmp.F
The critical snow depth is changed to depend on snow density.
The albedo will now take into account the effects of snow "patchiness"
at larger snow depths, especially in the cold areas where the snow density is lower.
It will help to increase the melting rate.
HYBFRONT - hybhqc.f
Copied background from NCEP to avoid extreme coldness
spot near MT/AB/SA border. This was later this evening determined
to be a problem with mesonet surface pressure, which was
converted into a far-too-cold surface theta-v.
HYBFRONT - hybhqc.f
Flag surface theta and moisture (p-RH) obs if
the surface pres/elev is flagged.
This avoids bad surface theta-v obs getting in
that were due to elevation problem.
HYBPOST - mapsout_post.f
Do not use 2m qv -- just use 5m qv. There was
a lot of spots showing up in the 2m dewpoint from
discontinuities in soil/skin temperature fields
from melting snow spots. The use of the reduction
to 2m was exaggerating the local variations.
HYBPOST - mapsout_post.f
Similar to change from last night, now do not allow
theta at 2m to be any lower than 2 K less than
the 5m theta-v. This avoids
from exaggerating (but not eliminating)
the cold spots over snow cover areas
where 2m temps in RUC13 web products were showing up about
20-30 deg F lower than surrounding grid points.
HYBFRONT - raobin.f
Fix to nasty bug, moved loop filling obs
arrays with flags previous to station loop,
not inside station loop. This fixes
losing the first 2 raobs at NCEP,
and parts of 2-3 raobs subsequent to
beginning at FSL, dependent on how many
raobs were read in from /public.
HYBFRONT- hybanl_qc.f, pirepin.f , ruc13env script
Identify Canadian aircraft separately as otype=428.
Add to ob_grid_stats output.
Remove external profiler data by changing
ruc13env script to dummy directory.
HYBFRONT- NORMA.DAT
Correct NORMA.DAT to intended value,
same as that used in 20km oper RUC.
HYBFRONT- SETOFEQS, hybanl_qc.f
3 changes - 1)return to 20km values for
OEQ for raob wind/height values.
Will force more fit to winds aloft.
2) do not use CNJ (Canadian jets)
anymore for now due to larger O-B values.
3) adjust soil moisture nudging so that
moistening is 1/2 of that previously,
and also now allow soil drying as well
as moistening.
HYBFRONT - hybanl_qc.f
Corrected soil moisture nudging - should be done when there is
no clouds during the day. (The "if" statement worked the opposite way
before).
HYBFRONT - sfcin.f
Added check for at least 100 METARS, otherwise
don't use any mesonet stations.
HYBFRONT - pirepin.f, "REJCTAIR"
Adapted pirepin to read new reject list format, now can
blacklist by variable (T,W,R). New rejctair file
from Bill Moninger created from information gathered the
last week of May 2005. Also fixed bug where U,V were
not initialized with flag- so when no ACARS winds (very
rare), the previous winds were still in the variable.
HYBFRONT - pirepin.f
Fixed logical NO_TEMP to be initialized to FALSE for
each station before checking for missing or bad temperatures.
ACARS with only good observation of T were being tossed
after the first time NO_TEMP was set to true. This only
became an issue after new blacklist would flag winds and
not temperatures. Also added counters for number of observations
of each type being flagged.
HYBPOST - mapsout_post.f
For plotting convinience the boundary values of soil
temperature and moisture are filled in now at all levels
rather than only at the top 2 levels as it was before.
HYBFRONT - pirepin.f, gtpirp.f, ob_grid_stats.f, hybanl_qc.f
Fixed bug in gtpirp.f that only allowed observations
with all three variables through to PIREPIN. Also
added more print statements to pirepin and fixed logical
NO_TEMP to be reset to false for each observation before
checking for a valid temperature, then setting to TRUE if
no valid temperature is found. Put new version of ob_grid_stats
into hybanl_qc, this version also breaks down residuals by
pressure level.
HYBFRONT - pirepin.f
Fixed bug with THETAP variable going out to the
analysis was not always overwritten with flag when there
was no temperature data, so sometimes carried previous
station temperature data in it.
HYBFRONT - hybanl_qc.f, pirepin.f
hybanl_qc.f - Reinstate smoothing of native background
fields for u/v only, similar to previously used in 20km
RUC 3dvar. This was overlooked with the consideration of
the isobaric-based residuals with the development of the
RUC13 3dvar (eventually decided not to use). The smoothing
of temp and qv is now removed, since ob_grid_stats showed
that this was NOT reducing mean-abs residuals for those
variables.
pirepin.f - Add Southwest Airlines to OTYPE
and modify comment.
HYBCST - cup_forcing_ens.F
Instate version from 20km_dev, which does not
pressure average the vertical velocity and does not
zero out the moisture convergence closure contribution
over water.
topoenvl.dat
Introduced more detailed version of 13km topography,
as originally intended. The version in the operational
RUC13 is smoother than originally intended, since
there was some unadvertised smoothing in the WRF_SI used.
HYBFRONT - rejctair.dat
New aircraft reject list.
HYBFRONT - hybanl_qc.f, gtresids_pw.f
Addition of ob-bkg residual threshold checks
before applying buddy check. This change was prompted
by recognition in TAMDAR of lack of buddies for
more robust QC of problematic obs.
HYBCST - cup_forcing_ens.F, cup_output_ens.F,
cupnew_h.F, physics_trial.F, CUGREL
Correction in output (precipitation, temperature
and moisture tendencies) to properly normalize for
number of closures used at each grid point.
HYBFRONT - hybanl_qc.f, ob_grid_stats.f, exsfcobs.f
Improved bkg_thresh and O-B stats, much cleaner now.
Also, finally began to enforce 50mb difference in
exsfcobs, flagging all obs if 50mb diff between
ob and background is exceeded.
HYBFRONT - gtresids_sfc_tquv.f
Finalized code fixes for coastal stations.
Now use land stations over water if station
elevation is less than 35 m (20m setting over
last week). This now catches, for instance,
Key West.
HYBFRONT - rejctair.dat
New aircraft reject list, based on statistics from 5-Oct through
3-Nov-2005. (WRM)
HYBCST - hybcstdv.F
Introduced diabatic DFI to RUC model to improve consistency with
physics in DFI, but also to prevent against intermittent crashes
during forward step with 13km RUC and more detailed topo.
HYBCST - hybcstdv.F
Eliminated write of half-initialized fields (after backward DFI pass).
This may have been overwriting analysis data for ijet prsver raob stats
for the 13km dev RUC.
HYBCST - cupnew_h.F, cup_kbcon.F
Fixed bug -- add i index for cap_max_increment array used
for ensemble cap threshold.