Change Log for 40-km Experimental MAPS

    Main MAPS/RUC homepage

    November 1995

    Initial implementation. 40km grid, 151 x 113 grid points, 40 vertical isentropic/sigma hybrid levels, new surface fixed fields (topography, land use, climo SST). PRMCMP, TRNCMP, HYBNMC, HYBING, HYBHQC, HYBANL, HYBCST.

    SMS version of HYBCST used for model, with appropriate changes for 40km to each of the three SMS hybcst-related programs: hybcst_pre, hybcst, and hybcst_post.

    NetCDF used for all grid i/o (with software designed so that GRIB or another format could be substituted at a later date).

    AVN model data used for boundary conditions until Eta data available at later date.

    Water vapor mixing ratio replaces condensation pressure as stored water vapor variable in all components of MAPS.

    Bugs corrected in HYBHQC.



    Version 0.1 - November 1995 - Initial 40-km software


    December 1995

    Addition of HYBINT for 40-km version to produce isobaric grids, special-level grids (freezing, maximum wind), and mean-layer grids (boundary layer).

    Direct incorporation of surface observations along with upper-level observations in HYBING, HYBHQC, and HYBANL. Colorado mesonet data included.


    January 1996

    HYBING - Modification of buoy data to use bogus moisture so that temperatures and winds will actually get used.


    17 January 1996

    HYBANL, HYBHQC - Extrapolation of surface obs to model terrain modified so that all wind and surface pressure obs are used, regardless of pressure distance between ob and model terrain. Temperature and dew-point observations now used in analysis with pressure separation of no more than 50 mb (increased from 30 mb in 60km version). With this change (exsfcobs.f) and 40km higher resolution terrain, most surface data (almost 95% of temp/td, 100% of wind/pressure) is now used in the 40km version.


    24 January 1996

    HYBINT - Added diagnosis of potential vorticity for isobaric levels. Added diagnosis of surface fields (MAPS SLP, potential temperature, winds, dew-point temp, altimeter setting, 3-h pressure change) using mini-topography (minimum topography from 10km data set within each 40km box).


    25 January 1996

    HYBNMC - Removal of sub-zero interpolation problem for relative humidity. Resulted in part from coarse (2.5 x 5 deg) AVN data.



    Version 1.1 - 30 January 1996 - Addition of HYBINT, direct incorporation of all surface data


    6 February 1996 - 1800 UTC run

    HYBING - RASS virtual temperature data included. About 6 stations, including Vandenberg AFB, Platteville, CO, and stations in KS/OK profiler network.


    9 February 1996 - 2100 UTC run

    HYBANL - Decreased observation errors for surface winds and RASS temperatures (previously set at unrealistically high values). Also decreased observation errors for rawinsonde moisture, surface moisture, and surface pressure, and profiler winds (also too high previously). The net effect of these changes is to cause the analysis to draw more closely for these data. Added univariate wind analysis update for lowest 5 levels.


    13 February 1996 - 2100 UTC run

    HYBCST - Modifications to mixing processes (subroutine MELLYA). Changed to smooth flux Richardson numbers in time and other changes to largely eliminate oscillations in surface temperatures and other surface fields. This had been happening to some extent in stable conditions with moderate wind speed.


    18 February 1996 - 1500 UTC run

    HYBHQC - Relinked with lower observation errors. Improves quality control, including for bad surface pressure obs which had sometimes been not caught.


    20 February 1996 - 0600 UTC run

    HYBHQC - Fixed problem with extrapolated surface obs being written out to MAPS observation files if some value at that station had been flagged. Now, the raw, unextrapolated observation is kept, since it is re-extrapolated in HYBANL. This eliminates other surface pressure bullseye situations.

    HYBANL - Changed condensation pressure analysis to isotropic univariate, instead of elliptical univariate, which had been causing ill-conditioned matrices in some situations. The condensation pressure analysis (in all 60-km as well as 40-km versions) had assumed an elliptical error covariance field previously. Also, improved determination of 700mb temp (using variable instead of fixed lapse rate for surface pressure < 700 mb) for MAPS SLP calculation. This change allows the lapse rate used for the reduction to vary between dry adiabatic and standard, as is done in HYBINT. The lapse rate used is that between the 1st and 5th levels, constrained as described above. This improves the MAPS SLP field in the vicinity of high terrain.


    22 February 1996 - 1200 UTC run

    HYBCST - hybcst_post - Similar to 20 Feb change in HYBANL, improved determination of 700mb temp (using variable instead of fixed lapse rate for surface pressure < 700 mb) for MAPS SLP calculation. This improves the MAPS SLP field in the vicinity of high terrain.


    24 February 1996 - 1800 UTC run

    HYBANL, HYBHQC, HYBING - frstgues.f. Corrected computation of condensation pressure from pressure, virtual temperature, and mixing ratio (another mistaken use of temperature instead of virtual temperature). This corrects a drying bias that was being repeatedly applied in each update time in the assimilation cycle.


    28 February 1996 - 0000 UTC run

    HYBANL - thvpc2tq.f, hybanl.f. Corrected computation of relative humidity from pressure, virtual temperature, and condensation pressure. RH should be e/es * p/pc, instead of previous e/es *(p-e)/(pc-es). This corrects another drying bias that was being repeatedly applied in each update time in the assimilation cycle. In addition, specific humidity was being written out instead of mixing ratio. This, also, was corrected.


    28 February 1996 - 1800 UTC run

    HYBANL - hybanl.f, procob.f, srchsect.f, srchobs.f. Implemented a look-up table specifying level of actual observation for all single-level obs. This change gives identical results, but reduces run time by 15-50%.


    29 February 1996 - 0300 UTC run

    HYBANL - uv.f, hybanl.f. Corrected error in uv.f that caused large errors in boundary layer for moisture and winds (using theta instead of pressure in vertical correlation). Also, set so that lowest level analysis is not affected by any vertical interpolation.


    29 February 1996 - 0600 UTC run

    Use of generic readg.f and writeg.f routines. These routines can be linked to either NetCDF or GRIB libraries. HYBNMC, HYBING, HYBHQC, HYBANL, HYBCST (hybcst_pre, hybcst_post). HYBINT was already using these routines.



    Version 2.1 - 1 March 1996 - readg/writeg, faster analysis w/ closer fit to obs, RASS, etc.


    1 March 1996 - 0000 UTC run

    HYBING - getboy.f. Changed to use actual elevations for buoys in the Great Lakes instead of sea level in the 0000 UTC run.


    1 March 1996 - 2100 UTC run

    HYBING - pirepin.f, readacars.f. Changed to read new netcdf ACARS files and pass pirepin the correct units. Subroutine CONVRT no longer needed in pirepin, location check done directly in pirepin, as are checks for flagged temps, wind speeds and directions.


    11 March 1996 - 1500 UTC run

    HYBCST - cup.f, cuparth.f. Changed to new Grell cumulus parameterization scheme (same as MM5 Version 2). Areas of sub-grid-scale precipitation are more coherent and less noisy than in previous Grell scheme.


    16 March 1996 - 1800 UTC run

    HYBANL - init.f. Decreased scale of horizontal correlation of forecast error for height at lowest levels. This is consistent with recent changes in reducing observation error for surface pressure. Both changes together result in closer fit to surface observations. The first without the second gave the risk of occasional bullseyes in surface pressure, so this problem is now fixed.


    2 April 1996 - 1800 UTC run

    HYBCST - first tests of soil/vegetation model begin.


    11 April 1996 - 1800 UTC run

    HYBCST - Forward/backward initialization started.


    12 April 1996 - 1800 UTC run

    HYBING - Correction to relative humidity calculation from condensation pressure, similar to 28 Feb 96 change.

    HYBING, HYBHQC, HYBANL - Capability for cycling of soil temperature and moisture and canopy water added.


    13 April 1996 - 1800 UTC run

    HYBCST - Cycling of soil temperature/moisture and canopy water begun. Also, convective adjustment reinstated.


    15 April 1996 - 2100 UTC run

    HYBCST - Cycling of soil temperature/moisture and canopy Smoothing of Richardson number removed. Vegetation type look-up table changed to remove tropical forest in all locations except over Cuba and Yucatan. Other isolated points (causing noise) with large roughness length also were changed to different vegetation type.


    16 April 1996 - 0000 UTC run

    HYBNMC, HYBANL, HYBCST, HYBINT - Determination of I4TIME for NetCDF files corrected.


    16 April 1996 - 1800 UTC run

    vegtype_40km.dat - More deforestation required for to avoid 2 dx variations for tropical forest. corrected.


    17 April 1996 - 1500 UTC run

    HYBCST_PRE, soilvegin.f - Roughness length for tree vegetation type class reduced to 70-80 cm.


    22 April 1996 - 1800 UTC run

    HYBCST, nconv.f - Relaxed limit on latent heat release for non-convective precipitation by factor of 10 from 0.1% of current water vapor mixing ratio to 1.0% per IDIAFR time step.



    Version 3.1 - 23 April 1996 - soil/veg model, digital filter initialization


    24 April 1996 - 1200 UTC run

    HYBCST, hybcstdv.f - Changed frequency for physics from every 5 to every 4 time steps.


    25 April 1996 - 1500 UTC run

    HYBCST, simila.f - Set an upper limit on stability functions (PSIM, PSIH) in free convection to avoid negative u* with very large roughness length. hybcst.f - Changed frequency for physics back to 5 time steps.

    HYBCST_PRE, soilvegin.f - Roughness length restored to original values for tree vegetation.

    vegtype_40km.dat - Tropical forest restored to original values now that free convection problems have been solved. Environmentalists around the world cheer wildly.


    26 April 1996 - 1500 UTC run

    HYBCST_PRE, inithybv_pre.f - Started continuous cycling again for soil temperature and moisture, now that PBL is behaving stably.


    30 April 1996 - 2100 UTC run

    HYBCST, nconv.f - Effectively removed any limit in latent heat release (set to 10% of water vapor mixing ratio).


    1 May 1996 - 0000 UTC run

    HYBING, get surface data routines - Changed to pass station pressure for all surface data, including buoy data. Bulleyes from Great Lakes buoys eliminated by this fix. Other smaller bugs also fixed.


    2 May 1996 - 1500 UTC run

    HYBCST, nconv.f - Set latent heat limit back to 1% of water vapor change in any time step (reversal of 30 April change).


    4 May 1996 - 2100 UTC run

    HYBCST_PRE, inithybv_pre.f,getsnow.f - Read in daily U.S. Air Force snow/ice analysis and set land/ocean surface parameters accordingly.


    14 May 1996 - 2100 UTC run

    HYBCST_PRE, inithybv_pre.f,getsnow.f - Corrections to use of snow/ice analysis and setting land/ocean surface parameters.

    HYBCST - panpbl.f, mellya.f, hybcst.f - Smoothing bulk Richardson number to avoid problems with quilt-like surface temperature patters forming under stable conditions (nighttime, esp.) with moderate wind speed.

    HYBCST_POST, HYBINT, HYBANL - Added output of additional fields for surface and hydrology budget studies associated with the GCIP program. There are 42 fields output for the surface, and 15 upper-level fields. These fields are being written in GRIB format.


    16 May 1996 - 1800 UTC run

    HYBCST_PRE, inithybv_pre.f; HYBCST, inithybv.f - Pass monthly albedo, with modifications due to daily snow cover, into model forecast.


    17 May 1996 - 1200 UTC run

    HYBCST_PRE, inithybv_pre.f - Start to use Great Lakes daily lake surface temperature data, as provided by Great Lakes Environmental Research Laboratory.


    18 May 1996 - 2100 UTC run

    HYBCST, panpbl1.f - Changed horizontal averaging of flux Richardson number to lowest 15 levels.

    mellya1.f - Added time averaging of flux Richardson number.

    Relinking also caused mid- and upper-level clouds to work again, after not being present for about 1 week.


    22 May 1996 - 1800 UTC run

    HYBING, readexpf.f readexrs.f - Start to use boundary-layer (915 MHz) wind profiler and RASS temperature data.


    23 May 1996 - 1800 UTC run

    HYBANL, frstgues.f - Added additional read loop for soil temperature/moisture background data. Now, even if MAPS must cold-start using AVN or ETA data, HYBANL will read back through 3-h forecasts (up to 6 days, at 3-h frequency) to find assimilated soil moisture/temp data. This is done since the assimilated soil fields, even if a few days old, are far superior to those from climatology.



    Version 4.1 - 23 May 1996 - improvements to surface physics/fixed fields, soil temp/moisture cycling, boundary-layer profilers, GCIP output


    25 May 1996 - 1800 UTC run

    HYBCST, sfcrad.f - Changed emitting level for back longwave radiation incident surface from level 6 to level 9. Back LW was too high.


    29 May 1996 - 1200 UTC run

    HYBCST, sfcrad.f - Changed upper-level cloud to start at 75% instead of at 60%. Was producing too much upper-level cloud, in part because non-convective precip is with respect to water saturation at all levels (in the current scheme). Also, added thickness diffusion at top 4 levels to enhance upper-level sponge.


    3 June 1996 - 2100 UTC run

    HYBCST, mellya1.f - Removed time averaging of Richardson number. This removes a significant error in upper-level wind and temperature forecasts.


    6 June 1996 - 0000 UTC run

    HYBANL, HYBCST_POST, capcin.f - Changed calculation of convective available potential energy and convective inhibition to use averaged temperature and moisture over lowest 5 levels.


    19 June 1996 - 0000 UTC run

    HYBNMC - 1st pass corrected to run at +4h15min rather than +4h 15 min. This fixes the 3-h cycle dropouts at 0300 and 1500 UTC.


    21 June 1996 - 2100 UTC run

    HYBCST, hybcst.f, lwrad.f, swrad.f, sfcrad.f, RADIAT - Added MM5 longwave/shortwave atmospheric radiation scheme. It is called every 30 minutes. Cloud water mixing ratio is temporarily diagnosed from relative humidity (pending implementation of MM5 microphysics). Radiative temperature tendencies are now calculated at every level; before, radiative effects were only included in the surface energy budget.


    24 June 1996 - 1800 UTC run

    HYBANL, HYBCST_POST, capcin.f - Corrections to routine that calculates convective available potential energy (CAPE) and convective inhibition (CIN). Corrected virtual temperature error. New version gives somewhat lower (and more accurate) amounts.



    Version 5.1 - 24 June 1996 - Addition of MM5 radiation, improve averaging of Richardson number, miscellaneous changes


    25 June 1996 - 1800 UTC run

    HYBCST_POST, makefile - Correction to routines linked to. CAPE/CIN calculation is now corrected for forecasts.



    Version 6.1 - 26 June 1996 - Additional hybcst_post changes.


    27 June 1996 - 1800 UTC run

    HYBING (sfcin.f, spechum.f, use new tv2tq.f), HYBHQC (frstgues.f), HYBANL (frstgues.f), HYBINT (use new tv2tq.f) - fixes to moisture transformations.


    27 June 1996 - 1800 UTC run

    HYBCST (hybcst.f) - Change radiation frequency to every 60 min.


    27 June 1996 - 2100 UTC run

    HYBING (sfcin.f) - Another moisture transformation bug corrected.

    HYBANL, HYBCST_POST (capcin.f) - Fix to virtual temp calc.


    28 June 1996 - 1800 UTC run

    HYBNMC (hybnmc.f, geteta.f) - Started using Eta 48-km data interpolated to AWIPS 80km grid for boundary conditions. Also uses Eta surface wind data at lowest level.


    1 July 1996 - 2100 UTC run

    HYBING (sfcin.f) - Changed over to METARs from SAOs.


    3 July 1996 - 1800 UTC run

    HYBCST (simila1.f,panpbl1.f) - Determines top of boundary layer at all grid points and does horizontal smoothing of Richardson number up to maximum PBL top level only.


    5 July 1996 - 1200 UTC run

    HYBINT (hybint.f) - Corrected diagnosis of surface temperature and dewpoint fields.


    12 July 1996 - 2100 UTC run

    HYBING (profin.f, getvad.f, readvad.f) - Added VAD (velocity-azimuth display) wind profiles from National Weather Service WSR-88D radars. There are currently about 75-80 radars.


    18 July 1996 - 0000 UTC run

    HYBANL (hybanl.f), HYBINT (hybint.f), HYBCST_POST (mapsout_post.f), dbgrib.f, most scripts - Began writing out all grids in GRIB-encoded format. Also made fix to correct writing out of isobaric vertical velocity.


    22 July 1996 - 2100 UTC run

    HYBING (getvad.f, readacars.f, raobin.f) - Corrections to VAD ingest (allow for more stations), ACARS ingest (use full UPS data stream including ascent format with interpolated lat/lon positions), raob ingest (correct calculation of mixing ratio and condensation pressure).


    29 July 1996 - 1800 UTC run

    HYBING (readacars.f, gtpirp.f, pirepin.f) - Resume collecting carrier statistics for ACARS data.


    1 August 1996

    PRSETA - replaces PRSAVN. Start using Eta (48km interpolated to 80km AWIPS grid interpolated back to 40km grid) for isobaric data and for verification.



    Version 7.1 - 8 August 1996 - Version before combining source code with that from 60-km RUC, ingest changes (including VAD), switch to Eta for boundary conditions


    30 August 1996 - 0000 UTC run

    PRSETA (prseta.f) - Write out RH instead of RH*.



    Version 8.1 - 30 August 1996 - Version with combined source code in single directory structure.


    2 September 1996 - 1800 UTC run

    HYBANL (hybanl.f, uv.f, uvth.f, setofeqs.f) - Various analysis fixes, mostly to do with vertical correlation. Vertical correlation for condensation pressure, potential temperature, and winds near ground tightened (smaller correlation for given vertical distance). Interobservation vertical correlation corrected for univariate analyses. Background error increased for temperature and winds at mid levels and for condensation pressure at all levels.

    4 September 1996 - 0000 UTC run

    PRSING (60-km version, which creates obs file used by 40-km PRSVER (fvlev.f)) - No more extrapolating temperatures below surface. This was affecting verification of 850 mb temperatures.


    12 September 1996 - 1800 UTC run

    HYBING (40km_upperair_run) changed to get qc'd acars from the Facility Division's netcdf files.


    13 September 1996 - 1500 UTC run

    HYBING (readvad.f, getvad.f) - Sort VAD winds by time, only take observation files in window from 1 hour before to 1.5 hrs after, and only take actual observations (one per station) starting at 30 minutes after the hour up to 30 minutes before the hour.


    20 September 1996 - 0000 UTC run

    HYBNMC, PRSETA (hybnmc.f, prseta.f) - Took out wind rotation code. Not needed now that both MAPS and NCEP grids are on the CONUS 211 projection and since NCEP grids are already grid-relative.

    Effect on output - This change had the largest change on overall accuracy of 40-km MAPS forecasts of any made so far.


    25 September 1996 - 1500 UTC run

    HYBANL (uvth.f) - Corrected bug in univariate analysis regarding calculation of observation-grid point correlation. Result is closer fitting of moisture observations and winds near surface. needed now that both MAPS and NMC grids are on the CONUS 211 projection.


    1 October 1996 - 1200 UTC run

    HYBCST - Changed to use mpi version of the Scalable Modeling System (SMS) code.


    23 October 1996 - 2100 UTC run

    HYBHQC (procogqc.f,exhzcnup.f) - Changed observation type look-up table so that profilers (including 915 MHz and VADs) can use each other for QC. Previously, this was not so, resulting in very little QC for most profiler data.


    24 October 1996 - 2100 UTC run

    HYBING (readacars.f,gtpirp.f,pirepin.f) - Fixed tail number in ingest. Now can use reject list properly and buddy check (which is set to not use reports from the same tail number for checking a given observation).


    30 October 1996 - 1500 UTC run

    HYBCST (hybcst.f,hybcstdv.f) - Changed definition of hybrid coordinate so that sigma levels are forced to flatten out by the 500 hPa level, rather than 600 hPa as previously. (Parameters now set in oafcprms.f to control this definition, rather than locally defined as previously.) This reduces the variation in layer thickness, especially over high terrain, going from sigma levels to isentropic levels. Also, time step reduced to 50 s in initialization only (still 60 s in main model run). Output frequency changed to specified output times, now with output at 1,2,3,6,9,12 hours. hybcstin.dat file changed for specify these modifications.

    HYBCST_PRE {hybcstdv_pre.f) - Changed to read in new hybcstin.dat file.

    HYBCST_POST {hybcstdv_post.f) - Changed to read in new hybcstin.dat file.

    HYBANL (hybanl.f) - Changed to new hybrid definition (see HYBCST change description for this time.)


    31 October 1996 - 1200 UTC run

    HYBNMC (hybnmc.f) - Changed to new hybrid coordinate definition. Also, corrected CUSHN function so that it agrees with that in HYBANL and HYBCST.


    1 November 1996 - 1500 UTC run

    Soil moisture and temperature. Reconstituted these fields to partially compensate for losing continuous cycle in late September. As of this time, the new fields are the 17 Sept soil moisture, combined with 1 Nov soil temperature except for replacing the lowest level (1.5 m) with a climatological mean field.

    1 November 1996 - 2100 UTC run

    HYBCST (param.f,metcon.f,paramx.f,swrad.f,pemant.f) - Modified solar constant so that it is not constant anymore but varies over year (by about 7%) according to earth-sun distance. Used code from John Augustine, Surface Radiation Research Branch, Air Resources Laboratory.

    HYBCST (oafcprms.f) - Changed RH value for saturation from 95% to 100% to be consistent with surface fluxes over water.


    4 November 1996 - 1800 UTC run

    Introduced HYBPOST (hybcstdv_post.f, mapsout_post.f, hybint.f). This program combines the previous HYBCST_POST and HYBINT, taking advantage of the fact that the model output fields are already available in the post processing to also do the interpolation. This program runs once for each model output time.


    7 November 1996 - 1500 UTC run

    Introduced HYBFRONT, (hybfront.f, ingdrv.f,rdpztd.f,hybhqc.f, hybanl.f,init.f,hybint.f) and changed frstgues.f to read the file SOILSAVE if a first guess for the soil temperature and moisture is not found. HYBFRONT is a combined data ingest, quality control, analysis and grid interpolation routine. This should result in slightly faster runs, since first guess grids and observations are only read in once, and then passed between routines. The analysis now calls the quality control, and passes common parameters so only one routine needs to be changed to update such things as correlation coefficients. Redundant operations done previously in both the analysis and observation quality control were also eliminated from one or the other. This was done with minimal impact on the structure of the system, and is nearly identical to the old version in most of the code. The analysis grid is also passed into the interpolation grid routine (hybint) directly from the analysis subroutine.


    11 November 1996 - 1500 UTC run

    rejctlst.dat - Updated aircraft tail number reject list based on recent statistics from Bill Moninger's automatic mail message.


    13 November 1996 - 2100 UTC run

    HYBFRONT (getvad.f, hybint.f) getvad.f was changed to only pass winds that have the "A" qualifier (confidence of 1.03 m/s). Previously we passed winds of "A" and "B" quality (B confidence of 2.06 m/s), however statistics over 3 months showed degradation of forecasts using B winds (birds?). A minor fix was made to hybint.f to write out a grib file to MAPSG_SPEC.

    Effect on output - could sometimes be significant in lower tropospheric wind fields


    16 November 1996 - 1500 UTC run

    HYBCST (mapsout.f) - Changed to zero out precipitation amounts only on 3-h output times. This may be modified for hourly output in the near future.

    Effect on output - slight, only in precip accumulation


    18 November 1996 - 1800 UTC run

    HYBCST (soil123.f) - Corrected a fairly significant error over snow cover. Previously, the ground temperature was forced to be at least 0 deg C, rather than no more than 0 deg C. This gave a significant warm bias over snow covered regions, and also gave too strong surface winds because of too much vertical mixing. The warm bias will still be apparent for a few days in these areas, due to the too warm soil temperatures.

    Effect on output - significant


    19 November 1996 - 1200 UTC run

    HYBNMC (hnmcdr.f, hybnmc.f, cushn.f) - Set up to process Eta forecasts out to 48 h. Also, changed to use included statement function cushn.f.

    Effect on output - none


    20 November 1996 - 1800 UTC run

    HYBFRONT (hybanl.f) - Changed to use included statement function cushn.f. Changed so that ground temperature field update from surface temperature analysis increment now occurs only at land points, as defined by LANDUSE file.

    Effect on output - none on forecasts; analyzed ground temperature is overwritten in forecast model anyway.


    21 November 1996 - 2100 UTC run

    HYBCST_PRE (gtnmctim.f, getbc.f, inithybv_pre.f, parame.f, oafcprms.f, check_input.f), HYBCST (hybcst.f, inithybv.f, parame.f) - Changed to allow for boundary-condition segments of variable length (e.g., updates from the Eta model every 3 h instead of every 6 h). Improved accounting for missing boundary condition files. Also now allows for any duration of forecast.

    Effect on output - none unless there are missing boundary conditions. In some such cases, the nudging targets are now set more accurately.


    22 November 1996 - 1800 UTC run

    HYBFRONT (gtpirp.f,readacars.f) - Added check to ACARS data ingest to screen out aircraft with a roll angle greater than 5 degrees.

    Effect on output - slight improvement in accuracy of winds.


    23 November 1996 - 0000 UTC run

    HYBFRONT (uvth.f,init.f) - Changes for vertical correlation, decreasing in both cases. Decrease vertical correlation in pressure space for theta-v analysis, and changed vertical correlation table based on potential temperature separation for all other variables.

    Effect on output - closer fit to rawinsonde observations, slight decrease in analysis noise.


    23 November 1996 - 1800 UTC run

    HYBFRONT (pirepin.f) - Corrected bug that had resulted in aircraft tail number reject list not being used. Also set so that any aircraft reports with zero wind speed will not be used. These reports are almost always erroneous.

    rejctlst.dat - Updated again.

    Effect on output - Eliminates chance of using bad aircraft observations.


    25 November 1996 - 0300 UTC run

    crontab - Adjusted time of runs at times other than 0000 or 1200 UTC to be at 52 min after analysis time, rather than 1 h 22 min as previous.

    Effect on output - For now, the last hour of ACARS data is not available to the earlier run, reducing the amount of ACARS reports by about one-third.



    Version 9.1 - 25 November 1996 - Substantial changes to most parts of code, including introduction of HYBFRONT and HYBPOST, changing of hybrid-b coordinate definition, and fixing wind rotation bug in boundary conditions.


    6 December 1996 - 1800 UTC run

    HYBCST (simila2.f) - Eliminated 1.0 m/s limit on u*.

    Effect on output - Significant - corrects overforecast of surface wind speed and eliminates crashes in some strong mountain wave events.


    8 December 1996 - 0000 UTC run

    HYBFRONT, HYBPOST (hybint.f, hybanl.f, mapsout_post.f) - Added MAPS sea level pressure to hybint call. This variable is already calculated in hybanl.f and mapsout_post.f, but should also be written in in the sfc grid file.

    Effect on output - Almost none - MAPS SLP had been missing in the surface files since the introduction of HYBFRONT and HYBPOST. This omission is now corrected.


    11 December 1996 - 1500 UTC run

    HYBCST_PRE (inithybv_pre.f) - Changed threshold for assignment of snow cover from 2.5 cm to 6 cm snow depth. The area of 2.5 cm depth from the USAF snow analysis appears to be exaggerated.

    Effect on output - Corrects a cold bias at the surface in regions showing about a 1 inch snow depth from the snow analysis.

    11 December 1996 - 1800 UTC run

    HYBPOST (hybcstdv_post.f, soilvegin.f) - Added call of soilvegin.f back in to allow correct calculation of vertically integrated soil moisture and maximum potential integrated soil moisture fields.

    Effect on output - Corrects integrated and maximum integrated soil moisture fields.

    11 December 1996 - 2100 UTC run

    40km_upperair_run script - Repointed aircraft data directory (PIREPDAT) to location that outputs partial data files for a given hour with the correct name, allowing its use.

    Effect on output - The ACARS data volume for the 3-h window is now nearly complete again, corrected most of the data loss that occurred on 25 Nov 96 when many of the runs were moved up by 30 min.


    12 December 1996 - 1500 UTC run

    HYBCST (hybcstdv.f, optfil.f) - Implemented optimal filter (Huang and Lynch, 1993, Monthly Weather Review) in digital filter initialization for adiabatic forward/backward technique. The weights applied to each time step are prescribed now with this optimal filter.

    Effect on output - 0.3-0.5 mb/hr reduction in mean absolute sfc pressure change with new optimal filter, meaning quieter forecasts in the first few hours.


    13 December 1996 - 1800 UTC run

    HYBCST (sfcrad.f) - Reduced computed cloud fraction using RH-based function. Previously, cloud fraction increased from 0-100% over the RH range 82.5-95%. Now, the RH range is 87.5-100%.

    Effect on output - Reduces cloud fraction slightly at all levels. Slightly more heating in day, more cooling at night, increases diurnal contrast in some areas slightly.


    14 December 1996 - 1800 UTC run

    HYBCST (swrad.f, lwrad.f) - Corrected errors regarding devirtualization of temperature in both radiation routines. There had been a vertical index error such that the top level temperature was being devirtualized with the bottom level water vapor mixing ratio with the same reversal applicable to all levels.

    Effect on output - Improves radiation tendencies slightly.


    18 December 1996 - 2100 UTC run

    HYBCST (cup.f,hybcstdv.f) - Corrected calculation of cloud top. Previously, looked from top down until a level with positive buoyancy was found. Now, set to look from level of free convection upward until a level with negative buoyance is found. Now, capping effect works properly. Also set minimum cloud depth to 1000 m instead of 100 mb. hybcstdv.f change - time check - should be no effect.

    Effect on output - Eliminates areas of erroneous light precipitation especially over ocean areas.


    19 December 1996 - 1500 UTC run

    HYBCST (hybcstdv.f) - Further tweak to time check change so that precipitation is not zeroed out for last output time.

    Effect on output - Gives proper precipitation at last output time.


    21 December 1996 - 0000 UTC run

    HYBCST_PRE (inithybv_pre.f) - Fixed so that ice temperature is set to air temperature at initial time and held constant through model run. Previously, ice temperature was not allowed to go below 0 deg C.

    Effect on output - Eliminates warm spots over ice covered areas, currently part of Hudson's Bay, Lake Winnepeg, Winnepegosis, and Manitoba, and Reindeer Lake in Saskatchewan.



    Version 10.1 - 31 December 1996

    Final 40 km version prior to integration of new microphysics package.


    2 January 1997 - 1500 UTC run - Introduction of explicit cloud microphysics with cycling of cloud variables

    Microphysics from NCAR/Penn State MM5 model includes explicit forecasts of mixing ratios of cloud water, rain water, snow, ice, and graupel, and also a forecast of the number concentration of ice particles. Reference: Reisner, Rasmussen, and Bruintjes (1997, Quart. J. Royal. Meteor. Soc.)

    The MAPS version of this routine includes use of the hybrid isentropic/sigma coordinate and changes determined by John Brown. Also, rather than force the model to recreate cloud fields at the beginning of each model run, the cloud fields in MAPS are "cycled", meaning that the forecast fields for all 6 cloud variables valid at the analysis time are used as initial conditions for the next model run. These fields are already in microphysical balance with respect to the model, which allows the precipitation and cloud forecasts from MAPS to avoid spin-up problems in the first few hours.

    HYBFRONT (mapscon.f, hybanl.f, frstgues.f, hybfront.f, rdpztd.f) - Read in 6 cloud variables, along with other fields in forecast background. Pass these through HYBFRONT to serve as an initial cloud analysis for the next forecast.

    HYBCST_PRE (getbc.f, hybcstdv_pre.f, inithybv_pre.f, modelin.f, point3d.f) - Read in 6 cloud variables and pass through to HYBCST in NNT/SMS format.

    HYBCST (changtq.f, cuparth.f, exmoisg.f, paramr.f, hybcst.f, inithybv.f, gamma.f, mapsout.f, outqve.f, outqvet.f, outqvom.f, simila2.f, dcmpsu_h.f, metcon.f, parame.f, pmoist.f, point2d.f, point3d.f, pointbc.f, paramx.f, pmicrph.f, hybcstdv.f, linkrm.scr, linkInclude.scr, makefile) - Call mixed phase cloud microphysics every 10 time steps, horizontal and vertical advection of 6 cloud variables, corrections to thermodynamic transformations, add all diabatic physics tendencies at physics time steps now rather than distributing over interim time steps.

    HYBPOST (hybcstdv_post.f, mapsout_post.f, hybint.f) - Write out 6 cloud variables.

    Effect on output - Very significant overall. Precipitation forecasts appear to be improved, bias in relative humidity forecasts is greatly reduced.


    4 January 1997 - 0000 UTC run

    HYBCST_PRE (inithybv.f) - Fixed cycling of surface ground temperatures.

    HYBCST (lwrad.f, swrad.f, panpbl2.f) - Fixed minor virtualization/ devirtualization errors.

    Effect on output - Improved ground and surface air temperature forecasts.


    4 January 1997 - 1500 UTC run

    HYBCST (mapsout.f) - Fixed to write out precipitation accumulations up to 3 h regardless of output frequency.

    HYBFRONT (frstgues.f, hybanl.f)

    HYBCST_PRE (modelin.f)

    HYBPOST (mapsout.f) - Read/write out water vapor mixing ratio in g/g (rather than previous g/kg), same as other moisture/cloud mixing ratios.

    HYBNMC (hybnmc.f) - same as above, but updated at 0600 UTC 6 January 1997

    Effect on output - as described above - minor


    7 January 1997 - 0000 UTC run

    HYBCST (hybcst.f) - Fixed bug in absorbing "sponge" at top of model.

    Effect on output - Significant reduction in noise at stratospheric levels in model. Among other things, tropopause pressure diagnosed from potential vorticity is now somewhat smoother.

    HYBPOST (scripts - post_it, FSL_hybcst.run) - Set up so that crontab job does post-processing on model output at each output time, looking for any new raw output files every 10 min.


    11 January 1997 - 1500 UTC run

    HYBCST (hybcst.f, parame.f) - Enhanced sponge layer at top of model, which now has enhanced diffusion at top 7 levels rather than top 3 levels.

    Effect on output - Substantial reduction in overall noise, particularly at upper levels.


    11 January 1997 - 1800 UTC run

    HYBFRONT (frstgues.f, rdpztd.f, hybanl.f) - Fixed bug in determination of of dew-point temperature, used to "virtualize" temperature for obs without moisture measurements, e.g., aircraft, buoys. More importantly, fixed bug regarding assignment of isobaric level for observations and forecast background errors, used in the optimal interpolation analysis.

    Effect on output - significant. Aircraft and buoy temperature observations are now being used correctly (this error was introduced on 2 Jan). The level assignment error means that observations above 175 mb were being assigned unknown weights in the analysis. This error has been in all previous versions of MAPS and the Rapid Update Cycle.


    16 January 1997 - 2100 UTC run

    HYBFRONT (guestim.f) - Order of first guess priority changed from 12h NCEP, 24h NCEP, 12h MAPS, 0h NCEP to 12h MAPS, 0h NCEP, 12h NCEP, 24h NCEP.


    20 January 1997 - 2100 UTC run

    HYBFRONT (profin.f, hybhqc.f, hybanl.f, exsfcobs.f, mulcor.f, gtpirp.f, gtrass.f, writegg.f) - code was slightly altered to run without the "static" option during the compile, mostly filling variables with zeroes or flags.

    Effect on output- minimal. The executable should run slightly faster and use memory more efficiently. A few minor bugs were also uncovered during this upgrade and corrected.


    22 January 1997 - 1800 UTC run

    HYBCST (lwrad.f, swrad.f, soilsnow.f, tqsnow.f, hybcst.f soil123.f) - snow accumulation (from snow or graupel) and melting, ice/snow/graupel effects on radiation added. Also, energy balance of snow layer added.

    Effect on output - significant. Large improvement in radiation at surface, and minimal internal snow physics is now included for first time.


    26 January 1997 - 1500 UTC run - snow depth/temperature cycling

    Now, snow depth and temperature fields are cycled in 40-km MAPS cycle. With this change, snow fields are kept up-to-date according to MAPS forecast snow accumulation and melting. Previously, the same snow could be remelted in each 3-h cycle since the snow field was respecified at each 3-h time. Now, the once-daily updated USAF snow analysis is used to update the 40-km MAPS snow field only at 2100 UTC each day.

    HYBFRONT (frstgues.f, hybanl.f, mapscon.f) - Change to pass through snow temperature and depth. Change hybanl to add atmospheric temperature increment at lowest level to snow temperature as well as soil temperature at top level.

    HYBCST_PRE (inithybv_pre.f, modelin.f, getbc.f) - Read in snow depth/temp from analysis field. Set to read daily USAF snow analysis only at 2100 UTC each day or if snow fields are missing from MAPS/RUC analysis. This means that snow is cycled via MAPS/RUC accumulation/melting between updates from USAF analysis. Also, set ice cover over all of Hudson Bay and Manitoba/Saskatchewan lakes according to ice cover from USAF analysis at point in Hudson Bay.

    HYBPOST (mapsout.f) - Write out snow temperature (actually, temperature at atmosphere/surface interface, whether it is snow or top soil level), for snow depth/temperature cycling.

    Effect on output - moderate at surface. Keep more up-to-date snow field during 24-h cycle.


    27 January 1997 - 1800 UTC run

    HYBCST_PRE (inithybv_pre.f) - Fix error introduced yesterday so that surface parameters are set properly for snow cover areas regardless of time of day.

    HYBPOST (mapsout_post.f) - Write out snow accumulation field properly to NetCDF (was previously working for GRIB but not NetCDF).

    Effect on output - snow melt was too excessive for runs in last 21 hours.


    28 January 1997 - 2100 UTC run

    HYBCST (exmoisg.f) - Limit fall velocity of rain and graupel to 5 m/s. The intent is to keep the size of precipitation time step from getting too small in intense precip episodes.

    Effect on output - minimal.


    29 January 1997 - 2100 UTC run

    HYBCST_PRE (inithybv_pre.f) - Fixes for snow and ice cover. The code now identifies missing data holes in USAF snow cover analysis and fills them in with surrounding data. There were 55 grid points with this problem at 1800 UTC 1/29/97, mostly around the Great Lakes, James Bay, and the Manitoba lakes. This problem has been present since the 40km MAPS started using the USAF analysis in spring 1996. Also, a change was made to ensure using proper ice cover data, fixing a bug introduced two days ago.

    Effect on output - considerable at the "missing snow" points in the regions described above for low-level temperatures.



    Version 11.1 - 3 February 1997

    Explicit cloud microphysics and snow physics added, upper-level sponge fixed/enhanced


    13 February 1997 - 1800 UTC run

    HYBPOST (post_it script) - Change so that SOILSAVE file will be written properly.

    Effect on output - improvement in surface temperature when cold start occurs.


    14 February 1997 - 1800 UTC run

    HYBCST_PRE (inithybv_pre.f) - Fixes to snow cycling code.

    Effect on output - some improvement in surface temperature due to proper setting of snow temp from background.


    20 February 1997 - 1500 UTC run

    HYBFRONT, HYBPOST (hybint.f) - Correction to 3-h pressure tendency, addition of temperature and specific humidity as output files for surface grids.

    Effect on output - 3-h pressure tendency is now correct (decreased by factor of 3)



    Version 12.1 - 20 February 1997

    Fixes to snow cycling and SOILSAVE file



    22 February 1997 - 1800 UTC run

    HYBFRONT (rejctlst.dat) - Modified aircraft tail number reject list to include same aircraft as those used in 60km list, plus a few other aircraft with certain errors in reports.

    Effect on output - Should reduce possibility of bad aircraft reports getting into 40km MAPS.


    24 February 1997 - 0000 UTC run

    HYBCST (lwrad.f, swrad.f, inithybv.f, exmoisg.f, oafcprms.f) - Corrections to radiation: removed background value of snow, ice, and graupel in longwave radiation. Set proper thicknesses of layers consistent with MM5 vertical staggering. (Minor changes - created CONFLX_P for constant flux layer, and removed STOPs in exmoisg.f)

    Effect on output - Removes longwave warm bias, especially near tropopause.


    24 February 1997 - 1800 UTC run

    HYBCST (lwrad.f, swrad.f) - Fix to layer depth at lowest level.

    Effect on output - Very small, except at sfc layer.


    12 March 1997 - 0000 UTC run

    HYBCST_PRE (getsnow.f, inithybv_pre.f) - Set to only update snow cover with NCEP/USAF snow cover data only if data valid for current date is present. This can only occur at 2100 UTC. The NCEP/USAF snow data has been missing since 3 March 1997, and MAPS had been erroneously updating with the 3 March snow field every day since then. Now, the MAPS snow fields, which had already been cycling internally using explicit snow accumulation and snow melting over 24-h periods, will be allowed to continue to cycle internally until up-to-date snow data from NCEP becomes available again.

    Effect on output - Could be significant at low levels; now snow fields will be more accurate.


    14 March 1997 - 2100 UTC run

    HYBCST_PRE (getsnow.f, inithybv_pre.f) - Now reads in ice data from NCEP/USAF snow file, even if that file is not current. Snow fields continue to cycle internally over multi-day periods in MAPS if no current NCEP/USAF data is available.

    Effect on output - Allow ice-covered areas to have more accurate (colder) temperatures near surface - most apparent over Hudson Bay.


    18 March 1997 - 1800 UTC run

    new_file, post_it scripts for HYBPOST - Modified so that HYBPOST post-processing on model output file is done, even if file is being written out when cron initiation is performed.

    Effect on output - Will give more consistent 3-h cycle. A number of 3-h forecasts have not been post-processed, so the effect of this change is quite significant.


    18 March 1997 - 2100 UTC run

    HYBCST_PRE (inithybv_pre.f, getveg.f) - Two major changes - Now using vegetation (greenness) fraction from NESDIS global 0.144 degree 5-year average. This file was provided via NCEP (Ken Mitchell) and originally developed by Garik Gutman. Also, it was discovered that the hydrometeor mixing ratios had been getting zeroed out, meaning that, in fact, these fields were not being cycled from one run to the next. This was corrected so that the cycling will now occur.

    Effect on output - Vegetation fraction - should improve temperature/moisture forecasts near ground in general, since NESDIS values tend to be somewhat higher, especially in summer. The NESDIS values are certainly more horizontally detailed, also, so more local effects may show up. The previous climatological vegetation fraction field was taken from a 1 degree data set. Cloud cycling - Since the microphysics was implemented in January, the precipitation fields from the 40-km MAPS have been too low and a spin-up effect has been apparent in comparing 3-h vs. 12-h forecasts. This fix will have a substantial impact on precipitation forecasts overall.


    21 March 1997 - 1800 UTC run - Burk-Thompson explicit turbulence scheme

    HYBCST_CODE (esatpb.f, hoskeep.f, navypb.f, navypbl.f, parambt.f, parbt.f, presimila.f, svp.f, uvcomp.f, hybcst.f, hybcstdv.f, inithybv.f, panpbl1.f, paramx.f, simila1.f, simila2.f, vilka.f) - Implemented Burk-Thompson explicit turbulence scheme (Burk and Thompson, 1989, Mon. Wea. Rev.). MAPS/RUC model now has turbulence kinetic energy as an additional prognostic variable. This replaces the previous Mellor-Yamada level-2.0 scheme with the level-3.0 BT scheme. The surface layer is still prescribed by Monin-Obukhov similarity theory, as described in Pan et al. (1994, Mon. Wea. Rev.).

    HYBPOST (hybcstdv_post.f, mapsout_post.f) - Changed to add output of TKE.

    Effect on output - Now have explicit TKE forecast for the first time in MAPS/RUC. Maxima of TKE show up frequently in frontal zones, consistent with previous tests of the Burk-Thompson scheme in isentropic coordinates. Tests have also shown higher TKE values (5-20 m**2/s**2) in boundary layer. Winds at the surface tend to be slightly higher than with the previous Pan/Mellor-Yamada level 2.0 turbulence scheme.


    31 March 1997 - 1800 UTC run

    HYBPOST (mapsout_post.f, capcin.f) - Fixed SOILSAVE so that boundary values are valid. Also, fixed calculation of CAPE to start from level of free convection up rather than from cloud top down to max moist energy level.

    Effect on output - Diagnostics only. CAPE is now smaller in some regions, and SOILSAVE will work when needed.


    1 April 1997 - 0300 UTC run

    HYBFRONT, HYBCST - topoenvl.dat. Switched from full envelope topography to slope envelope topography conceived by John Brown. Also, corrected offset error in topography. The topo is also calculated from 4x4 fine terrain points, rather than 5x5 overlapping fine terrain boxes. The slope envelope (or "Brown envelope") gives lower values in high terrain regions, but also places slopes more accurately and often with larger gradients.

    Effect on output - Some topographically induced circulations will be different. Surface fields (forecasts and analyses) are more accurate, and the amount of extrapolation of surface temperatures and dewpoint obs is decreased since the slope envelope topo is closer to the actual elevation of most surface stations.


    1 April 1997 - 1800 UTC run

    HYBFRONT, HYBCST - topomini.dat. Also now uses values from 4x4 fine terrain boxes rather than overlapping 5x5 boxes, and location offset is also corrected.

    Effect on output - Diagnostics only. Slightly more accurate surface temperatures in surface grids interpolated to topomini elevation.


    2 April 1997 - 1800 UTC run

    HYBCST (cup.f) - Latent heat limit for convective precip changed from 7.2 K / hr to 1.4 K / hr. Also, maximum cloud base raised from level 27 to level 31 (out of 40 levels).

    Effect on output - Latent heat limit decreases convective precip blow-ups. Cloud base change will allow more high base convection over high-elevation regions.


    4 April 1997 - 1500 UTC run

    HYBCST (navypb.f soil123.f, soilsnow.f) - Roughness length over ocean now passed in from SIMILA rather than using NAVYPB value. Approximation for semi-implicit treatment of net radiation flux in sfc budget is changed.

    Effect on output - Wind speeds over ocean slightly reduced.


    5 April 1997 - 0000 UTC run

    HYBFRONT (hybanl.f, hybint.f), HYBCST (exmoisg.f), HYBPOST (hybint.f) - HYBFRONT - added univariate update for surface height (pressure) analysis after multivariate (height/wind) analysis. Latent heating limit increased up to 2.8 K/hr. Diagnosis of sfc dewpoint temp and dewpoint depression in sfc analysis/forecasts (hybint.f in HYBFRONT and HYBPOST).

    Effect on output - Sfc pressure/SLP now draws for obs slightly more closely. Heavy precip slightly enhanced. Dewpoint and dewpoint depression values more accurate.


    7 April 1997 - 1500 UTC run

    HYBCST (soil123.f, soilsnow.f), - Approximation for semi-implicit treatment of net radiation flux in sfc budget is back to the old version because the new one increased the warm bias. In SOILSNOW the transpiration is now possible and is calculated in correspondence with the vegetation fraction in this area, that is changed from day to day.

    Effect on output - the warm bias of surface temperature is slightly reduced.


    7 April 1997 - 1800 UTC run

    HYBPOST (mapsout_post.f) - Added snow depth to SOILSAVE file, since we haven't had any USAF snow analyses since early March, and we need to ensure that the latest snow depth field cycled through MAPS is available even in the event of a cold start.

    Effect on output - None unless a cold start occurs, in which case the snow field now will be fairly realistic.


    10 April 1997 - 1800 UTC run

    HYBCST (inithybv.f, sfctmp.f) - In the grid points with snow ALB is set to 0.7 and SIGF is not zeroed out any longer so that the transpiration can happen in snow areas.

    Effect on output - the snow melting rate is expected to be smaller.


    14 April 1997 - 1800 UTC run

    HYBCST (soilsnow.f) - Correction is done in calculation of canopy water DD1 in the snow areas.

    Effect on output - very small because in the snow areas the vegetation fraction is very close to zero, but the evaporation of moisture from the canopy should work correctly now.


    15 April 1997 - 0600 UTC run

    HYBCST (soiveg.f, soilvegin1.f, soilvegin2.f, soil123.f, soilsnow.f) - The variable name of volumetric heat capacity of water is changed from CW to CVW to avoid variable name collisions with microphysics. In soil123.f and soilsnow.f the heat contribution of falling rain into surface energy budget is divided by 1000, because the specific heat capacity of water should be used rather than volumetric heat capacity that was in CW variable.

    Effect on output - The rate of snow melting is expected to be reduced in the grid points where the liquid precipitation is falling on the snow.


    16 April 1997 - 1200 UTC run

    HYBFRONT (frstgues.f) - Added snow depth to read of SOILSAVE file. Now, the code is fully prepared for a cold start to provide a realistic snow field. Code also set so that if ground temperature or snow depth in SOILSAVE are not reasonable values, then do not run HYBFRONT.

    Effect on output - None unless a cold start occurs, in which case the snow field now will be fairly realistic.


    16 April 1997 - 1800 UTC run

    HYBCST (tqcan.f,tqsnow.f) - Switch to Schaake/Koren expression for calculation of maximum infiltration rate.

    Effect on output - Surface runoff is expected to increase and be more realistic.


    25 April 1997 - 1500 UTC run

    HYBCST (hybcstdv.f,hybcst.f,cuparth.f,exmoisg.f,lwrad.f,nconv.f, point2d.f,swrad.f) - Changes to all these files are connected with addition of new fields and new include file MOLTS to provide all necessary information for GCIP project output. Also, the bug in computation of instantaneous precipitation rate (PRECIN) is fixed. In the microphysics code (exmoisg.f) th correction of precipitation rates computed in loop 80 is done. IF constructs are replaced with MAX and MIN functions where possible. The sign error for the evaporation of rain when temperature is below 0C is corrected.

    Effect on output - The microphysics code is expected to run faster and produce more realistic fields. Precipitation rate field PRECIN is computed correctly now in the case when the frequency of calling microphysics is not the same as the frequency of calling the rest of physics.


    25 April 1997 - 1500 UTC run

    HYBPOST (hybcstdv.f) - MOLTS include file is added, and MOLTS fields are added to the lists of variables to be read in from HYBCST output.

    Effect on output - All variables for MOLTS output are now available.


    30 April 1997 - 2100 UTC run

    HYBFRONT (mapscon.f, frstgues.f, hybanl.f), HYBCST_PRE (soilvegin.f, modelin.f), HYBCST (soilvegin1.f, soil123.f, soilsnow.f), HYBPOST (soilvegin.f, mapsout_post.f) -

    Add extra level to soil model at 3 m depth, giving 6 levels now.

    Effect on output - Allows slightly stronger diurnal cycle in soil temperature and improved long-range forecasts of soil temperature and moisture in ongoing cycle.


    1 May 1997 - 2100 UTC run

    HYBCST (soil123.f,soilsnow.f) - Set rooting depth variable NROOT to 5 for mixed forests and broadleaf deciduous forests, leaving NROOT=4 for the rest of the vegetation types.

    Effect on output - Evapotranspiration in the forested areas is expected to increase.


    5 May 1997 - 2100 UTC run

    HYBFRONT (procob.f, srchsect.f, oafcprms.f) - Changed observation selection: now, once the search algorithm finds an on-level ob of any type (not just raob or profiler as previously), it stops the search in that sector for that variable. Also, the algorithm now searches for on-level obs first before off-level obs. Finally, the volume over which each analysis solution is determined was changed from a 4x4 to a 3x3 box to improve the fit. This costs a little more CPU-wise, but the previous change improved the efficiency, so this set of changes overall is neutral.

    Effect on output - Smoother analysis increments, better conditioned matrices, especially for high-resolution ascent/descent aircraft reports. Also, closer fit to observations in analyses.


    6 May 1997 - 1500 UTC run

    HYBCST (inithybv.f) - Changed threshold of autoconversion from cloud water to rain water (set in INITHYBV, used in EXMOISG) from 0.5E-3 to 0.1E-3. This done on recommendation of R. Bruintjes - the larger threshold is more appropriate for convective clouds.

    Effect on output - Smaller cloud water mixing ratio, slightly greater area coverage of precipitation, less delay between cloud formation and precipitation in regions of strong upward motion. Other predicted microphysics quantities are minimally affected.


    7 May 1997 - 0900 UTC run

    HYBCST (cup.f) - Eliminated criteria that number of levels between cloud parcel source and level of free convection could be no more than 2. This appeared to be artificially reducing the convective precipitation, especially in the 40km version with finer sigma spacing.

    Effect on output - Increase in precipitation in convective areas. (Remember that with the Grell convective scheme, some of the "convective" precipitation is produced as resolvable-scale, so the field labeled convective precipitation is somewhat less than what would typically be called "convective". This field is more accurately called "sub-grid-scale" precipitation.)


    7 May 1997 - 1500 UTC run

    HYBCST (oafcprms.f) - Changed top of constant flux layer near surface from 10 m to 5 m. This means that the lowest level in the model is forecasting for 5 m above the ground now.

    Effect on output - More accurate temperatures and winds near surface. Stronger diurnal cycle, cooler temps at night and lighter winds.

    HYBCST (cup.f) - Added cap strength term. If negative temperature difference between cloud parcel and environment exceeds 3 deg K, no convection can occur. (Based on Carlson et al., 1980, Mon. Wea. Rev., and subsequent work by Marc Schwartz and Tom Graziano.)

    Effect on output - Will cut off convection in areas with significant "lid" or cap.


    12 May 1997 - 2100 UTC run

    HYBCST (cup.f) - Change maximum separation from cloud parcel source layer and level of free convection from 150 to 350 mb to account for high-based convection in a deep boundary layer.

    Effect on output - Allows convection in deep PBL areas with positive CAPE.


    15 May 1997 - 1800 UTC run

    HYBCST_PRE (inithybv_pre.f) - Temporarily eliminated any ice cover. No USAF snow/ice analyses had been received since 3 March, and the 40-km MAPS had been using the 3 March ice analysis up to this change.

    Effect on output - Eliminates cold bias near Hudson's Bay, over Manitoba lakes, and near Labrador.


    16 May 1997 - 0000 UTC run

    HYBCST (simila2.f) - Changed calculation of u* so that denominator cannot go below 1 regardless of roughness length and stability function (PSIM).

    Effect on output - Should eliminate unstable situations in tropics in free convection regime, large roughness length, and light wind speed.


    16 May 1997 - 0000 UTC run

    1-h cycle

    As of this run, the 1-h cycle seems to be running successfully for the first time. Since the SGI computer is at capacity, script changes to run the 1-h cycle and keep different parts of it from adversely affecting others were difficult, but these problems seem to be mostly solved now. Because of limits in computer power, runs may still be occasionally dropped.

    The optimal filter initialization continues to be run for about 35 minutes forward and backward in each 1-h cycle. This initialization is important for producing 1-h forecasts that are quiet enough to reanalyze with new data.

    HYBING (pirepin1h.f) - The one code change necessary for the 1-h cycle was to change the time window for aircraft data from 3-h (-2 to +1) to 1-h (-1 to 0). All other code changes had been previously incorporated in the 40-km MAPS design.

    Effect on output - Should provide slightly more accurate analyses, especially at surface.



    Version 13.1 - 19 May 1997

    Burk-Thompson explicit turbulence forecasts, 1-h cycle, snow cycling, NESDIS vegetation fraction, slope envelope topography, other surface changes, convection capping criteria, etc.



    20 May 1997 - 1500 UTC run

    HYBCST (hybcstdv.f,hybcst.f) - The frequency of output for GCIP project is not fixed now, but defined by the frequency of MAPS output.

    Effect on output - Averaged over time surface variables for GCIP will be correct in 1h and 2h forecast outputs.


    28 May 1997 - 1200 UTC run

    HYBCST (cup.f) - Changes to convective parameterization. Minimum cloud base is now at a lower level. Got rid of the 3 K cap term, switched back to a 75 mb maximum depth between cloud base and the level of free convection, changed variable to force more of cloud condensate into precipitation and less into evaporation.

    Effect on output - Considerably more sub-grid-scale precipitation.

    HYBPOST (mapsout_post.f) - Output for model output soundings.

    Effect on output - None, other than that soundings data in BUFR will soon be available.


    31 May 1997 - 0000 UTC run

    MOLTSBUFR - Started producing BUFR-packed model output soundings from 40km MAPS runs for 469 locations.

    Effect on output - None, new output format.


    2 June 1997 - 0300 UTC run

    HYBCST (cup.f) - Changes EDT variable for percentage of condensate into evaporation from 0.25 to 0.20

    Effect on output - Further slight enhancement to sub-grid-scale precipitation.


    2 June 1997 - 1800 UTC run

    HYBFRONT (hybanl.f hybhqc.f, exsfcobs.f) - Fixed so that QC now runs for moisture variables. Since the introduction of hybfront (7 Nov 96), there had been no QC on moisture variables. (yikes!) Horizontal scale for moisture obs changed to same as that for other observation types (had been smaller). Also, changed maximum extrapolation pressure difference for surface temp/dewpoint observations from 50 to 70 mb. Pressure and wind obs are always used, regardless of extrapolation distance, and now about 90-92% of surface temp/dewpoint obs west of ~108 W are also used (far better than the 60-km RUC or any of the Eta runs).

    Effect on output - Eliminates dewpoint problems at surface, and probably will improve moisture and precipitation skill overall.


    3 June 1997 - 1500 UTC run

    HYBCST (tqsnow.f,soilsnow.f) - Snow model now includes the process of snow melting from the bottom of snow layer.

    Effect on output - Warm temperatures will not appear any longer under the snow cover. Upper soil layer will give its energy to melting of snow until its temperature cools off to OC if there is enough snow to melt, or if it can stay above OC if the amount of snow is not enough.


    3 June 1997 - 1600 UTC run

    HYBFRONT (hybanl.f) - Added desmooth pass for winds and temperature near surface.

    Effect on output - Slightly closer fit to surface wind and temperature obs.


    3 June 1997 - 1600 UTC run

    HYBFRONT (inithybv.f) - Check for ice in Hudson bay will be done from November till June.

    Effect on output - No ice in the Hudson bay in summer time.


    6 June 1997 - 0900 UTC run

    HYBFRONT (hybanl.f) - Fixed bug resulting from 2 June changes, in which hybfront started to perform QC for moisture obs. In this change, changed NVAR_SRCH_ALL variable from 5 to 6, which means that v components will be used in analysis. v components now will be used again.

    Effect on output - Very large. Statistics have been way off since the 2 June change.


    11 June 1997 - 1900 UTC run

    HYBCST (navypb.f) - The specific humidity at the ocean surface is now equal to saturation value.

    Effect on output - Substantial moistening of surface air over water surfaces.


    11 June 1997 - 2100 UTC run

    HYBFRONT (hybanl.f, uvpw.f, hybpw.f, readpw.f, readgoes.f, readgps.f) - Introduced GOES total precipitable water estimates into MAPS. These data are ftp'd from the University of Wisconsin, courtesy of CIMSS (Cooperative Institute for Meteorological Satellite Studies -- Tim Schmit, Bob Aune, Paul Menzel). Values with sample of size of 9 or greater out of 25 possible are used, and only where the MAPS surface pressure is greater than 950 mb. A univariate analysis of precipitable water (PW) is done to correct the MAPS first guess PW field, and the entire column of water vapor mixing ratio values is corrected according to the change suggested by the GOES data. Further adjustments are made to avoid supersaturation at any points.

    Effect on output - The GOES data seems to improve the overall moisture field over low-elevation land areas, especially at non-raob times, and particularly over oceans.


    13 June 1997 - 1900 UTC run

    HYBFRONT (hybanl.f), HYBCST_PRE (inithybv_pre.f, soilvegin.f), HYBPOST (soilvegin.f) - Changed to read ASCII surface fixed fields in 22F6.2 format instead of 151F6.2). Also, changed all ASCII field field file in fslparms.

    Effect on output - None


    17 June 1997 - 1800 UTC run

    HYBFRONT (hybanl.f, frstgues.f), HYBCST_PRE (inithybv_pre.f, modelin.f), HYBPOST (helic.f, writeout.f, hybcstdv_post.f, soilvegin.f, profout.f, class0.f, class1.f, mapsout_post.f), HYBFRONT and HYBPOST (hb230mb.f, hb2p.f, hb2mw.f, hb2frz.f, hybint.f) - Transfer to the new output format, GRIBTABLE is modified, and variable names in CDL files are changed.

    Effect on output - New MOLTS output files, extra output files for implementation of RUC at NCEP.


    24 June 1997 - 1500 UTC run

    HYBCST (exmoisg.f,pmicrph.f paramr.f,point2d.f,sfctmp.f) - Corrected a bug in the MOLTS diagnostic calculation of latent heating by grid-scale condensation. - Revised the calculation procedure for the microphysical process describing the collection of ice crystals by rain to form graupel. New procedure avoids both a potential underflow problem and a sign error in this term for very small rain mixing ratios. - Introduced sedimentation of cloud ice based on Eq 5 in Heymsfield and Donner 1990, J. Atmos. Sci.

    Effect on output - Possible minor effect on surface temperature forecasts. Effect on other microphysical variables and other aspects of model performance: uncertain.


    26 June 1997 - 2100 UTC run

    HYBFRONT (srchsect.f) - Corrected a bug that was preventing the use of single-level data (surface or aircraft) at off levels.

    Effect on output - Significantly improves retention of information from single-level obs and at least of surface forecasts. Surface data was previously having virtually no effect on subsequent forecasts, since it was only having an effect on a 2 mb deep layer, but not above.


    27 June 1997 - 1500 UTC run

    HYBFRONT (hybanl.f, hybpw.f, hybsat.f, readpw.f, superobs.f, readssmi.f, readcdw.f) - Code added for including in MAPS/RUC analysis GOES cloud drift winds and SSM/I precipitable water data. Normally, this data is not available at FSL except in tests.

    Effect on output - None unless these new data types are present. Cloud drift winds are only used over water areas in hybfront, so the main effects are in winds near 900 mb and at 300-200 mb over ocean areas.


    27 June 1997 - 2100 UTC run

    HYBCST (soil123.f, soilsnow.f) - The number of levels in the root zone is increased from 4 to 5.

    Effect on output - Increase of evapotranspiration.


    30 June 1997 - 2000 UTC run

    HYBCST (hybcst.f) - Unstaggered surface winds for GCIP output are directly calculated rather than get them out of UNSTAGUV routine.

    Effect on output - None while running on castle (FSL), but on CRAY (NCEP) this change should fix the problem with running on multiple proccesors.



    Version 14.1 - 1 July 1997

    Assimilation of satellite precipitable water and drift winds (when available), new output (fixed GRIB variables, BUFR soundings), other smaller fixes



    8 July 1997 - 2100 UTC run

    HYBCST (exmoisg.f) - Minor fix to EXMOISG to correctly compute the diameter of ice particles.

    Effect on output - Minor, but allows execution on Cray at NCEP.


    14 July 1997 - 2000 UTC run

    HYBCST (soil123.f,tqcan.f) - Changed calculation of soil moisture in irrigated areas: soil now cannot dry out below the 0.75 of field capacity value.

    Effect on output - Cooler skin temperature during the day, increased evaporation and soil moisture in cultivated areas.


    18 July 1997 - 1300 UTC run

    HYBCST (tqcan.f) - Changed drying limit for irrigated areas from 0.75 to 0.5 of field capacity value.

    Effect on output - Reduced level of soil moisture in irrigated areas.


    21 July 1997 - 1300 UTC run

    HYBCST (tqcan.f) - Back to the old version of TQCAN with no exceptions of soil moisture calculations in the land-use class - cultivations, because the suggestion of using irrigation in these areas was erroneous, at least for most part of this class.

    Effect on output - Soil moisture field is defined only by natural reasons without artificial addition of water into soil anywhere.


    5 August 1997 - 1300 UTC run

    HYBCST (soilsnow.f) - The thickness of snow layer participating in energy budget is limited by DELTSN=7.5cm, what is reasonable because the largest thermal gradient below the snow surface is near the top of the snow layer due to weak thermal diffusion within the snow.

    Effect on output - When the cold season starts, snow temperature should be cooler and more realistic in the areas with deep snow.


    21 August 1997 - 1200 UTC run

    HYBNMC (hybnmc.f) - Memory saving version - Changes made to reduce memory by using arrays differently and using equivalences where possible. This is in response to recent discovery that all RUC-II programs must use less than 32 MW of memory on the NCEP Cray. This change reduces memory from 36 MW to 31 MW, so this program is now clean with regard to memory.

    Effect on output - None.


    1 September 1997 - 1800 UTC run

    HYBCST_PRE (modelin.f) - Set canopy water to at least zero value to allow cold starting from SOILSAVE file.

    Effect on output - Allow restart to occur properly when restarting with Eta atmospheric variables and SOILSAVE. Previously, values of -99999. were being used for canopy water, causing a crash in hybcst in this situation. Now the SOILSAVE safety net is intact again.


    5 September 1997 - 1200 UTC run

    HYBFRONT, HYBPOST (writegg.f) - A fix was made to the GRIB-writing routine to eliminate the writing of a dummy array if either only 2-d or 3-d grids are being written.

    Effect on output - GRIB files with only 2-d or only 3-d grids are now more easily read.


    5 September 1997 - 1800 UTC run

    HYBCST, HYBCST_PRE - memory-saving version -

    HYBCST - (soilsnow.f, hybcstdv.f, navypb.f panpbl1.f, changtq.f, presimila.f, exmoisg.f, hybcst.f, navypbl.f simila1.f, simila2.f, sfctmp.f, inithybv.f, point3d.f, ueqn.f, veqn.f, pointbc.f, oafcprms.f)

    HYBCST_PRE (oafcprms.f)

    A variety of memory saving measures were made to reduce the memory needed by hybcst to closer to the 32 MW allowed on the NCEP Cray for the RUC-II. These measures included passing of arrays, eliminating unused arrays, redimensioning arrays for the current application. These measures reduce memory use in hybcst from 68 MW to 44 MW, so more is yet to come.

    Effect on output - None, as it should be.


    6 September 1997 - 1500 UTC run

    HYBFRONT - memory-saving version -

    (hybanl.f, hybpw.f, exsfcobs.f, hybfront.f, rdpztd.f, hybhqc.f, capcin.f, tropwnd.f, frstgues.f, gtresids.f, hybint.f, hb2p.f, hb230mb.f, hb2mw.f, hb2frz.f)

    A variety of memory saving measures were made to reduce the memory needed by hybfront to closer to the 32 MW allowed on the NCEP Cray for the RUC-II. These measures included passing of arrays, and changing processing to avoid need for extra 3-d arrays. These measures reduce memory use in hybfront from 79 MW to 39 MW, so more is yet to come. Also, a significant bug was found and corrected in the adjustment of all variables at the top level. This bug has existed through the history of the hybrid-b analyses in both RUC-I and RUC-II. Finally, interpolation near the surface was set to start at level 4 rather than level 3, to allow an extra level to match the surface obs more closely.

    Effect on output - None, except for the changes at the top of the analysis (significant) and near the ground (minor).


    7 September 1997 - 2100 UTC run

    HYBPOST - memory-saving version -

    (hybcstdv_post.f mapsout_post.f, point3d.f)

    Memory saving measures were implemented to reduce the memory needed by hybpost to closer to the 32 MW allowed on the NCEP Cray for the RUC-II. These measures included passing of arrays, changing processing to avoid need for extra 3-d arrays, and making common blocks unique to hybpost to avoid unneeded storage. These measures reduce memory use in hybfront from about 90 MW to about 40 MW, so more is yet to come.

    Effect on output - None.


    8 September 1997 - 2100 UTC run

    GRIB output for analysis and forecast (grib.table) - PDS parameters for GRIB corrected for MORDS output from 30mb layer near surface. Also, increased number of significant digits for specific humidity in this layer.

    Effect on output - None.


    10 September 1997 - 1500 UTC run

    HYBCST (paramx.f, param.f, soilvegin2.f, inithybv.f, point2d.f, hybcst.f, changtq.f, sfctmp.f, exmoisg.f) - memory-saving changes. 2-d arrays that are not used in the latest version of HYBCST are removed. Two 3-D arrays that were used in vertical mixing before incorporation of Burk-Tompson PBL scheme were removed (EXCHGH,EXCHGM). 3-D temperature tendency (EXITTEN) from EXMOISG is replaced by 2-D TTEN, which is added to TA in CHANGTQ like all other microphysics tendencies.

    Effect on output - None.


    10 September 1997 - 2100 UTC run

    HYBFRONT (hybanl.f), HYBINT (new program w/ new driver - hybintdv.f) - Memory saving version. Split off interpolation processing in hybint.f into new program (HYBINT). Also, eliminated use of MAPS database software for further reduction of memory. Now uses 30 MW.

    Effect on output - MAPS database observational files no longer being written out (.ira,.ipr,.ipi, .isa, .ist). No effect on analysis grids.


    15 September 1997 - 1500 UTC run

    HYBPOST, HYBFRONT (grib.table) - Changed precision for vertical velocity, temperature, number concentration for ice particles to appropriate number for GRIB output.

    Effect on output - More appropriate precision for these variables in GRIB files.


    15 September 1997 - 2100 UTC run

    HYBCST - (exmoisg.f) - Fixed bug for calculation of DIACE. This bug was introduced on 7 July 97.

    Effect on output - Should eliminate some occasional crashes of HYBCST and perhaps fix the problem with temperatures at ~200 mb since the 7 July 97 fix.


    16 September 1997 - 2100 UTC run

    HYBPOST - (mapsout_post.f) - Started using HYBINT for interpolation as part of memory-saving version.

    Effect on output - None.


    17 September 1997 - 1800 UTC run

    HYBPOST, HYBFRONT (grib.table) Changed precision for instantaneous precipitation rate from to appropriate number for GRIB output.

    Effect on output - Precip rate will now not be zero in GRIB files.


    17 September 1997 - 2100 UTC run

    HYBPOST, HYBFRONT (grib.table) Changed cloud variable precision to 0.001 g/kg from 0.01 g/kg for GRIB output.

    Effect on output - More precision in cloud variables in GRIB files.


    18 September 1997 - 1200 UTC run

    HYBPOST - (makefile) - Took out routines no longer needed now that HYBINT runs separately. Reduced memory from from 40 MW to 25 MW. This change "realizes" the memory saving made possible in the 16 Sept 97 HYBPOST change.

    Effect on output - None.


    18 September 1997 - 1500 UTC run

    HYBFRONT - (hybanl.f) - Fixed call to TROPWND and CAPCIN so that pressure passed in is in mb instead of Pa. Now hybfront produces reasonable tropopause level fields and CAPE/CIN again.

    Effect on output - Trop level field and CAPE/CIN fixed.


    18 September 1997 - 1500 UTC run

    HYBCST - (makefile) - Took flag for roundoff approximation after test showed that crash resulted from use of this flag. Still keeping in IEEE flag. Hybcst now runs a little slower.

    Effect on output - None.


    19 September 1997 - 1500 UTC run

    HYBFRONT - (hybanl.f) - Fix to serious bug in extrapolation at top level. This bug was causing top level to have some major problems. Extrapolation should occur when temp increment at top level is negative, not positive. Sign error in extrapolation also corrected.

    Effect on output - Improved results near top in model.


    19 September 1997 - 1500 UTC run

    HYBCST - (changtq.f, exmoisg.f, hybcst.f, inithybv.f, point3d.f, soil123.f, soilsnow.f) - "B" arrays for microphysics variables are removed to save memory.

    Effect on output - Very little change of maximum value in precipitation fields without noticable change in spatial distribution of precipitation.


    19 September 1997 - 2100 UTC run

    HYBCST - (cuparth.f, exmoisg.f, lwrad.f, swrad.f, hybcstdv.f) - The calculation and output of four 3-D arrays needed for MOLTS grids are removed from CRAY version to save memory.

    Effect on output - None.


    20 September 1997 - 1200 UTC run

    HYBCST - (nudgeth.f) - Set a single vertical loop to save another 3-d array. More memory saving

    Effect on output - None


    20 September 1997 - 1200 UTC run

    HYBCST - (exmoisg.f) - Changes for vectorization of loops 15 and 30 on Cray.

    Effect on output - None


    23 September 1997 - 1800 UTC run

    HYBCST - (hybcstdv.f, ueqn.f veqn.f) - Changes for memory saving. Equivalence of physics tendency arrays and initialization arrays. With this change, HYBCST is now about 29.8 MW on the Cray at NCEP and all RUC-II programs are now under the 32 MW limit. :)

    Effect on output - None


    24 September 1997 - 2100 UTC run

    HYBFRONT - (hybanl.f) - Fix to prevent writing of gex3 at NCEP.

    Effect on output - None.


    Version 15.1 - 1 October 1997

    Version released to CIMSS (u. Wisconsin).


    3 October 1997 - 1400 UTC run

    HYBCST - (soil123.f,soilsnow.f) - Bug is fixed in calculation of the transpiration rate.

    Effect on output - increase of transpiration rate. Will reduce hot, dry bias in MAPS sfc fcsts.


    9 October 1997 - 2300 UTC run

    HYBFRONT - (hybanl.f, lcl.f, srinv.f) - Changed to vectorizable calculation of condensation pressure. In the process, improved accuracy of PC calculation at top levels/dry conditions. Also, improved vectorization of srinv.f (minor difference in Cray speed here). All of these changes are to improve Cray speed at NCEP.

    Effect on output - Not very much. Will remove slight tendency of dry air aloft to gradually become more moist.


    15 October 1997 - 1400 UTC run

    HYBCST - (soil123.f, soilsnow.f) - Evaporation from the canopy is corrected for the case when it is limited by the amount of water intercepted by the canopy.

    Effect on output - slightly reduced canopy evaporation.


    20 October 1997 - 2200 UTC run

    HYBFRONT - (hybanl.f) - Correction to smoothing of temperature increment from multivariate height/u/v analysis. Was being done only for level 1, but now is accomplished for all levels. Also, now limits this temperature increment to +/- 4 K.

    Effect on output - Eliminates a crash that occurred today at 1800 UTC. In general, will give smoother temperature increments.


    21 October 1997 - 1800 UTC run

    HYBCST - (sfctmp.f, soil123.f, soilsnow.f) - Cleaned-up version of these three routines in accordance to NCEP's CRAY specialist advice.

    Effect on output - None


    24 October 1997 - 1300 UTC run

    HYBFRONT - (hybanl.f) - Added quality control for bird contamination for VAD winds:

    >> Flag VAD winds under following conditions:
    >>   15 Feb-15 Jun - pos. v component
    >>   15 Aug-15 Nov - neg. v component
    >>   temp > -2 deg C
    >>   night time (local zenith angle less than 0)
    

    Effect on output - Will avoid bird contamination of low-level wind fields via VAD winds, but will also throw out some good winds.


    27 October 1997 - 0400 UTC run

    HYBFRONT - (init.f, superobs.f) - Fixes to eliminate occasional ill-conditioned matrices near surface. Set SOAR coefficient for Z at lowest levels to 0.60 instead of 0.50. Change superobbing distance from 50 to 55 km.

    Effect on output - Gets rid of most if not all ill-conditioned matrices in analysis solution near surface.


    27 October 1997 - 2100 UTC run

    HYBCST - (navypb.f, simila1.f, simila2.f, paramx.f, paramx_t.f, nudgeth.f, hybcst.f, panpbl1.f, presimila.f, SOILVEG) - Changes done by Cray expert at NCEP are incorporated in the FSL version.

    Effect on output - None.


    29 October 1997 - 1800 UTC run

    HYBPOST - (mapsout_post.f) - Instanteneous fields of latent and sensible heat fluxes for MORDS output are switched around, so that now they are in the right places.

    Effect on output - MORDS output is correct for instantaneous latent and heat fluxes.


    29 October 1997 - 2100 UTC run

    HYBPOST - (mapsout_post.f) - Units of snow accumulation are changed from [m] to [kg/m**2] to fit existing GRIB.TABLE.

    Effect on output - MORDS output has snow accumulation in right units.


    3 November 1997 - 1200 UTC run

    HYBCST_PRE - (inithybv_pre.f) - Correction is done in the calculation of albedo for the last quater of the year.

    Effect on output - This correction fixed model crashes beginning 1 November because of crazy albedo values.


    7 November 1997 - 2100 UTC run

    HYBCST - (hybcst.f) - Correction is done in the calculation of U10 and V10 used in GCIP output of momentum heat flux.

    Effect on output - GCIP files have correct momentum heat fluxes from now on.


    19 November 1997 - 2200 UTC run

    HYBINT - (hybint.f, grib.table) - Corrections to fields for RUC_SFC, including precip type. Corrections also to GRIB precision.

    Effect on output - Only on post-processing. Corrected RH, vv precision, corrected categorical precip types.


    20 November 1997 - 2100 UTC run

    HYBCST - (hybcst.f) - Correction to instantaneous precipitation rate between the physics time steps.

    Effect on output - Only on post-processing. Non-zero values in RUC_SFC output.


    20 January 1998 - 0300 UTC run

    HYBCST - (cup.f) - Moist feedback at cloud top changed so that scheme tries to saturate toward ice-saturation water vapor mixing ratio rather than that for water-saturation. The variable Tcrit was changed from 50K (total water saturation) to 258K (ice saturation below -15 deg C).

    Effect on output - Much less noise near cloud top in convective areas. Corrected occasional crashes in RUC-2 at NCEP.


    21 January 1998 - 0300 UTC run

    HYBCST - (exmoisg.f) - Changes to microphysics (John Brown)

    Effect on output - Less graupel.


    4 February 1998 - 0000 UTC run

    HYBCST - (simila2.f) - The stability function under stable conditions is limited by -10. This change will allow some turbulent mixing even in stable atmosphere.

    Effect on output - Warm bias at night near the surface is reduced.


    18 March 1998 - 1800 UTC run

    HYBCST - (cup.f, navypb.f, hoskeep.f, uvcomp.f) - CUP is returned back to version 15.1 to avoid afternoon crashes in the mountain areas. Old version of CUP allowed to run 1-h cycle beginning at 1800 UTC for the test-case without crash in initialization at 2100 UTC. The routines of Burk-Thompson PBL are switched to the vectorized versions which give identical results with the old versions. The signs of stability functions (PSIM and PSIH) are changed into the opposite in the calculations of temperature and moisture variances in the surface layer.

    Effect on output - Reduced noise in the wind field at the top of PBL in the mountain areas from CUP change. Very little change of wind and temperature near the surface from PSIM and PSIH change.


    23 March 1998 - 0000 UTC run

    HYBFRONT - (hybanl.f) - Commented out adjustment after univariate potential temperature analysis to eliminate any superadiabatic lapse rates from level 4 upward in analysis. The model forecasts using the Burk-Thompson mixing have a very weak superadiabatic lapse rate up to near the top of the mixed layer, and the analysis adjustment was causing fairly large pressure gradient changes and very large vertical movements of upper mixed layer levels that happened to be above the sigma levels in the hybrid MAPS levels at that point.

    Effect on output - Eliminated crashes in RUC2 at NCEP and occasional crashes on afternoon runs. Smoother height fields at analysis time and quieter forecasts.


    1 April 1998 - 1800 UTC run

    HYBFRONT - (getboy.f) - Index change to flag output array was inadvertantly included in 23 March relink of hybfront, causing no METAR data during this period due to array being overwritten with flags in getboy.

    Effect on output - METARS are once again flowing into the system.



    Changes from this point on are not in the original RUC-2 except for the capcin.f change on 15 May in MAPS40 and on 10 June in RUC-2.

    3 April 1998 - 2100 UTC run

    HYBCST - (soil123.f, soilsnow.f, tqcan.f, tqsnow.f, soilveg.f(include file)) - A version with frozen soil physics.

    Effect on output - Surface temperatures are cooler when the melting process of soil moisture takes place, and warmer when freezing begins. Surface runoff is slightly increased in the areas with frozen soil.


    8 April 1998 - 1900 UTC run

    HYBCST_PRE - (soilvegin.f) - Soil parameters are changed according to Pielke and Dickinson.

    Effect on output - Very little.


    13 April 1998 - 2200 UTC run

    HYBPOST,HYBFRONT - (capcin.f) - CAPE calculation changed to look for cloud top from top down rather than from bottom up. This gets rid of a problem in which the diagnosed CAPE value was near zero when there was a very slight negative buoyancy layer just above the level of free convection.

    Effect on output - Increases CAPE diagnostic in from near zero to much higher amount in some areas, but has no effect on actual analyses and forecasts.


    12 May 1998 - 2100 UTC run

    HYBCST - (simila2.f) - The value of ZMUS is changed from 0.85e-3 to 0.2e-3.

    Effect on output - Increase of evaporation.


    15 May 1998 - 1800 UTC run

    HYBPOST,HYBFRONT - (capcin.f,hybanl.f) - Major revision of CAPE/CIN calculations based on discussions with Storm Prediction Center (Phillip Bothwell, John Hart). CIN values are now lower and somewhat improved over previous values. Also, CIN now shows up in MAPS40 analyses on web site due to a change in sign.

    Effect on output - Improved CIN values, slightly lower CAPE values, possibly larger changes in some situations.


    28 May 1998 - 1500 UTC run

    HYBINT - (hybint.f) - Output surface RH to MAPS_SFC file.

    Effect on output - None.


    29 May 1998 - 1500 UTC run

    HYBCST - (soil123.f, soilsnow.f, tqcan.f, tqsnow.f) - More accurate calculation of canopy water and runoff for the regime of dew formation.

    Effect on output - Improved ratio between evaporation and runoff.


    29 May 1998 - 2100 UTC run

    HYBFRONT, HYBPOST - (tropwnd.f) - Correction to tropopause level calculation.

    Effect on output - Smoother tropopause pressure and tropopause fields in some areas.


    6 June 1998 - 1200 UTC run

    HYBCST_PRE - (soilvegin.f) - Change of wilting point value in clay soil regions back to lower value. This will reduce excessive high temperatures in much of Texas and Louisiana and Arkansas.

    Effect on output - Reduced high temps in these regions.


    Version 16.1 - 10 June 1998 - Changes from NCEP merged in, consistent with RUC-2 code except for changes here since 1 April 1998.


    25 June 1998 - 1900 UTC run

    HYBFRONT - (profin.f) - Correction to treatment of RASS temperatures. This fix eliminates a major bug in use of RASS temps whereby all stations were being assigned the data from a single station. Since RASS usually does not compete with other data sources, these bad data were usually accepted.

    Effect on output - Eliminates errors along northern CA and OR coast where analysis had been putting in horrendous low-level lapse rates, huge CAPEs, resulting in lots of erroneous convective precipitation.


    1 July 1998 - 1500 UTC run

    HYBFRONT - gtrass.f, readpw.f, gti4tm.f, mkfnam.f, readcdw.f, getcdw.f, hybsat.f

    Cleaned up rass ingest a bit, changed number of sample size back to 4 for satellite PW to be used (old readpw had somehow gotten into last relink), fixed 2 Y2K bugs in routines that convert between character time (YYDDDHHMM) and i4 (time in seconds since 1960) time, and set up the ingest routines to read the cloud-drift wind data on /public/data when the script is changed to find them.

    Effect on output - GOES PW once again going into analysis in numbers, other changes should have no immediate effect.


    24 July 1998 - 1500 UTC run

    HYBCST - simila2.f, navypb.f

    Cleaned up versions of SIMILA2 and NAVYPB.

    Effect on output - none


    18 August 1998 - 2100 UTC run

    HYBCST - soil123.f, soilsnow.f,tqcan.f, tqsnow.f, sfctmp.f, hybcstdv.f, point2d.f, makefile; HYBPOST - hybcstdv_post.f, mapsout_post.f

    Some clean-up of soil model. Other changes are connected with the producing correct output of net long- and shortwave radiation at the surface. The list of 2D variables that are passed from HYBCST to HYBPOST is updated: RNET is replaced by new array GLWNET (net longwave radiation, which is defined in SFCTMP routine), and GSW (net shortwave radiation) is added to the list.

    Effect on output - Output has now right fields of net long- and shortwave radiation at the surface.


    13 Sept 1998 - 1400 UTC run

    HYBFRONT - superobs.f

    Changes order of ob preference by ob type so that sfc data would be preferred over profiler data. This is because sfc obs are usually more complete in that they contain moisture information.

    Effect on output - Closer fitting to sfc obs in analysis near profiler stations.


    14 Sept 1998 - 0000 UTC run

    HYBFRONT - uv.f, init.f

    Reduce effect of theta separation and increase effect of horizontal separation in surface analysis. The theta difference was overdominant in the fitting of sfc obs in the sfc analysis.

    Effect on output - Closer fitting to sfc obs in analysis


    19 Sept 1998 - 0300 UTC run

    HYBCST - cup.f

    Set minimum level for maximum moist static energy parcel in a column to level 3 (previously level 1). This is done to reduce an unrealistic feedback between the convective precip and the surface fluxes.

    Effect on output - Reduction in convective precip, esp. over warm water.


    22 Sept 1998 - 2000 UTC run

    HYBCST - PARAME

    Set number of processes equal to 9.

    Effect on output - None


    30 Sept 1998 - 0000 UTC run

    HYBCST - lwrad.f, swrad.f

    Eliminate extra layer of high water vapor from top RUC level (~50-70 mb) to 0 mb in water vapor path length for both longwave and shortwave radiation.

    Effect on output - Major improvement in diurnal cycle near the surface. Improvement evident in temperatures at surface and in 850 mb temperature bias.


    2 October 1998 - 1500 UTC run

    HYBFRONT - readcdw.f, hybsat.f

    Fixed inconsistencies in cloud-drift wind ingest that prevented us from using them for about a month.

    Effect on output- should result in better wind analyses and forecasts, especially over the oceans.


    10 November 1998 - 1200 UTC run

    HYBNMC - hybnmc.f

    Changed RH to mixing ratio conversion to be consistent with Eta RH change made on 3 Nov 98. The Eta RH had been with respect to water saturation, but NCEP changed it (unannounced) to be with respect to ice at < -15 deg C and a mixture for -15 - 0 C. Now the RH is limited to no more than 50% at 200 mb and above also.

    Effect on output - Eliminates massive ice cloud over high levels.


    10 November 1998 - 1800 UTC run

    HYBCST - soilvegin1.f, soil123.f, soilsnow.f, tqcan.f, tqsnow.f, SOILVEG

    The variables of soil model are revised: 1. common/task_soil/ is excluded from SOILVEG, becuase its variables are local for soil model and don't need initialization outside the soil model routines SOIL123, SOILSNOW, TQCAN and TQSNOW. 2. The lists of variables passed between SOIL123 and TQCAN, and SOILSNOW and TQSNOW are rivesed and reduced.

    Effect on output - none


    28 December 1998 - 1300 UTC run

    HYBNMC - hybnmc.f

    Follow-up to 11/10/98 change. On 23 Dec 98, the Eta RH was changed back to be with respect to water only. Today's change makes the 40km MAPS at FSL consistent with this. The RUC at NCEP has already been fixed for this change. We've left in the limitation of 50% for Eta RH at 200 mb and above.

    Effect on output - For the past 5 days, the 40km MAPS has been too dry aloft.


    4 January 1999 - 1800 UTC run

    HYBCST_PRE - soilvegin.f

    One more change of soil parameters, closer to the old original version. The soil should keep less moisture because of the reduction of the wilting point.

    Effect on output - The values of soil moisture in the root zone should reduce in the areas with active transpiration.


    15 January 1999 - 2100 UTC run

    HYBFRONT - 40km_upperair_run

    Took the VAD winds out of the data ingest due to unflagged bad data getting into the files. We will monitor the situation and reverse this decision when we have confidence in the data again.

    Effect on output - Our wind verification should improve especially in the lower to mid troposphere.


    29 January 1999 - 2100 UTC run

    HYBCST_PRE (soilveg.f, linkInclude,scr), HYBCST (soilvegin1.f), HYBPOST (linkInclude.scr), HYBCST_POST (soilvegin.f) - 40km_upperair_run

    Soil and vegetation parameters are moved to include files: soilparm.f and vegparm.f in the include directory. This will help to avoid mistakes connected with the change of parameters' values. The parameters are passed from HYBCST_PRE to HYBCST, but in HYBPOST they are set up again to diagnose some variables. Now the changes should be done to include files, and HYBCST_PRE and HYBCST should be recompiled.

    Effect on output - Diagnostics of some variables in MORDS fields is done right according to values of soil parameters used in computations.


    9 February 1999 - 0000 UTC run

    HYBINT (hybint.f), topomini.f -

    Error corrected for lapse rate near surface in reduction to minimum topography. Error was causing isothermal lapse rate to be used in all situations. Also, new topomini file substituted which started w/ previous topomini and then adjusted values based on METAR surface elevations.

    Effect on output - Much improved surface temperature and dewpoint verification.


    12 February 1999 - 1800 UTC run

    vegfrac_nesdis.[month]

    The NESDIS vegetation fraction fields used in the MAPS 40km runs were inadvertently deleted sometime probably late in December 1998. Since that time, the vegetation fraction has been set to a default summer value, dependent on land use. This, in turn, has resulted in too much evaporation and too little sensible heating and too high soil moisture. Today, these files were restored.

    Effect on output - The effect of the deletion over the past 50 days has been to damp the diurnal cycle in MAPS and give too high dewpoints. With the restoration of the NESDIS vegetation fraction files, the soil moisture fields will gradually recover, but this will take a few weeks.


    19 February 1999 - 1800 UTC run

    HYBCST (parame.f)

    4 processor version in preparation for parallel cycle.

    Effect on output - None.


    20 February 1999 - 2000 UTC run

    HYBCST (exmoisg.f, paramr.f, pmicrph.f)

    exmoisg.f. (Changes from NCAR/RAP and John Brown). Bugfix for 0-intercept for snow (loop 20). Code cleanup. New value of zero-intercept for Marshall-Palmer distribution of rain drop size. Cooper curve replaces Fletcher curve for ice nucleation. Numerous other minor changes from NCAR/RAP.

    Effect on output - Less overproduction of graupel at temperatures near 0 deg C. Overall, somewhat less ice and more supercooled liquid water.


    20 February 1999 - 2200 UTC run

    HYBINT (hb2p.f)

    Changed smoothing coefficients to more passes for winds at all isobaric levels and more for temps at 200 and 150 mb. Based on experiments which show that this will improve verification for forecasts.

    Effect on output - Smoother isobaric wind and upper-level temp output.

    24 February 1999 - 1500 UTC run

    HYBCST, (exmoisg.f, paramr.f, pmicrph.f)

    Backed out microphysics changes made 20 February.

    Effect on output - Avoid excessive graupel fallout over oceans. Which was causing occasional NANs, and resultant crashes.

    11 March 1999 - 2100 UTC run

    HYBCST, (cuparth.f), HYBPOST (soilvegin.f, mapsout_post.f), MOLTSSITES (list of MOLTS stations)

    In HYBCST correction is done to the calculation of convective heating rate (DTCP) in CUPARTH. In HYBPOST bugs are fixed in computations of total soil moisture for GCIP output. MOLTSSITES - 7 stations are added to the list of stations for GCIP sounding output.

    Effect on output - GCIP grids have correct soil moisture at separate levels, and total moisture content in soil domain.

    31 March 1999 - 1800 UTC run

    HYBCST, (paramx.f, soil123.f, soilsnow.f, tqsnow.f tqcan.f) HYBCST_PRE (inithybv_pre.f)

    Most of the changes are connected with the introduction of two-layer snow model. Also changes are done to surface water balance equation (gradient of total water content in the atmosphere instead of gradient of liquid water), and to energy budget equation (changed formulation of heating/cooling by liquid rain).

    Effect on output - Improved temperatures over deep snow pack.

    8 June 1999 - 1800 UTC run

    SOILSAVE

    Soil moisture field is replaced by averaged between MAPS and RUC.

    Effect on output - Reduced moist bias in the surface layer.

    14 June 1999 - 1800 UTC run

    topomini.dat (fixed on 9 July in control, 14 June in parallel)

    Error corrected with wrong elevation in central California resulting from wrong elevation from Vacaville.

    Effect on output - Eliminates "hot spot" near Sacramento in sfc temps.

    26 June 1999 - 1800 UTC run

    HYBFRONT (mulcor.f, mcinit.f, hybhqc.f, exhzcnup.f, makefile) (castle parallel version at this time)

    Awkward function and entry code in mulcor.f replaced by subroutines for mulcor.f and mcinit.f. Also, vertical correlation for theta now included in QC rechecking w/ stations removed one at a time (exhzcnup.f). By some omisssion, it was not there previously, even though it was present in the first QC matrix check in hybhqc.f

    Effect on output - Very slight changes in QC. Code now compiles on IBM SP with external checking option.

    31 July 1999 - 1300 UTC run

    HYBCST_PRE (modelin.f) (both control and parallel versions)

    Convert canopy water from mm back to m, since hybcst uses canopy water in m.

    Effect on output - The fix of this error will significantly change the evolution of soil moisture in MAPS runs. The error was causing an overmoistening of the soil (as well as a less harmful gross overprediction of sfc runoff).

    2 August 1999 - 1800 UTC run

    HYBFRONT (exsfcobs.f) - (both control and parallel versions)

    Change most stable lapse rate from exact isothermal to slightly non-zero lapse rate to avoid division by zero.

    Effect on output - This avoids future crashes, but we are not aware that it has caused a problem up to this time. At least sfc data will now be properly included in the event of a very stable lapse rate in the background field nearby; previously it was not.

    2 August 1999 - 1200 UTC run

    HYBFRONT (readgoes.f,capcin.f) - (both control and parallel versions)

    Eliminate out-of-range errors. When there were more than 5000 GOES prec water obs, there was an error. Also, an error in capcin was also eliminated.

    Effect on output - Avoid crazy sfc wind fields that were showing up intermittently in MAPS.

    3 August 1999 - 1200 UTC run

    soilsave.dat (control version, parallel soil fields taken from control soil fields at 1500 UTC, 4 Aug 1999)

    Restart soil moisture and temperature with new file reconstituted from 3 parts RUC, 1 part MAPS. This Betty Crocker-like recipe was estimated from the relative errors judged in both RUC (too dry, due to canopy water precision problems in GRIB, soon to be fixed) and MAPS (quite a bit too moist, due to canopy m/mm problem fixed on 31 July).

    Effect on output - Decrease cool, moist bias during day and overdone precipitation.

    18 August 1999 - 1800 UTC run

    HYBCST (tqcan.f, tqsnow.f)

    'IF' statement is added to the calculation of infiltration rate in Shaake-Koren' runoff scheme to avoid possible division by 0.

    Effect on output - None

    23 September 1999 - 1400 UTC run

    HYBCST_PRE (inithybv_pre.f, soilvegin.f, param.f), HYBCST (param.f, inithybv.f, soilvegin1.f, soilvegin2.f, soilsnow.f, soil123.f, navypb.f) and new include file, vegparm.f.

    Changed for new vegetation types and cleaned up code.

    Effect on output - Improvements in temperatures, mostly at night.

    28 September 1999 - 1500 UTC run

    HYBINT, HYBPOST, HYBFRONT (calvis.f, hybint.f, capcin.f, mapsout_post.f, hybanl.f)

    Added 3 new output variables, pressure of theta-e max parcel, convective cloud top height, and equilibrium level height.

    Effect on output - three new variables to look at.

    27 October 1999 - 1500 UTC run

    HYBCST_PRE (inithybv.f), vegtype_40km.dat, soiltex_40km.dat

    Vegetation types smoothed- single point types were eliminated, soil texture file now includes Lake Nipigon, ONT, and the Great Salt Lake, UT.

    Effect on output - Surface temperatures should be more representative at these points.

    19 November 1999 - 1800 UTC run

    HYBINT (hybint.f)

    Fixed error with 3h pressure tendency

    27 November 1999 - 2300 UTC run

    HYBFRONT - rejctsfc.dat

    Added KLBL (Liberal, KS) due to consistently too high dew point temp.

    17 December 1999 - 2100 UTC run

    HYBCST (swrad.f, lwrad.f)

    removed extra division by dt for GCIP sw and lw radiation heating rate.

    Effect on output - corrected radiation for GCIP output, no changes to regular fields.

    22 December 1999 - 1500 UTC run

    HYBCST_PRE - (getbc.f inithybv_pre.f)

    Fixed bug about MAPS runs nudging to wrong BC after 3h.

    23 February 2000 - 1500 UTC run

    HYBFRONT - (mapscon.f)

    Changed NSTA_P from 4000 to 6000 and MXST_P from 5000 to 7000 to allow for more aircraft data.

    23 February 2000 - 1500 UTC run

    HYBCST - (soilvegin1.f, navypb.f, simila1.f, simila2.f, hybcstdv.f)

    MAVAIL is initialized in soilvegin1 to be higher than 0.00001. Also MAVAIL is excluded from calculations of AKQS in simila2.f, and from virtualization of surface temperature (TGBV) in both simila1 and simila2.

    4 April 2000 - 1500 UTC run

    HYBCST - (hybcst.f, sfctmp.f, soilvegin1.f, soil123.f, soilsnow.f, tqcan.f, tqsnow.f. ../include - soilveg.f)

    Optimized computation of properties for unfrozen and frozen soils. Algorithm for snow accumulating and snow melting is modified: snow can be intercepted by the canopy until the saturation value is reached. Infiltration into soil for vegetated areas consists only of snow melt. Several non-substantial changes in diagnostic computation of evaporation components.

    Effect on output - more realistic simulation of snow accumulation in snow-covered areas with non-zero vegetation fraction.

    17 April 2000 - 2100 UTC run

    HYBFRONT - (getvad.f, 40km_oper_upperair)

    Re-instituted the VAD wind ingest in the analysis with a check for light winds. If the winds are too light, they are deemed untrustworthy and discarded.

    Effect on output - better wind analyses.

    26 May 2000 - 1500 UTC run

    HYBCST - (new ppp version - see parallel cycle log).

    Effect on output - pervasive on all fields. Again, see para log.

    27 May 2000 - 1200 UTC run

    HYBCST - (same change as in para cycle).

    1 Aug 2000 - 1800 UTC run

    HYBCST - (cup_forcing.F, same change as in para cycle).

    5 Aug 2000 - 1200 UTC run

    HYBCST - (SOILVEG, SOILPARM (new), soilvegin1.F, soilsnow.F soil123.F)
    Implement Johansen soil thermal conductivity.
    Effect on output - slightly better diurnal cycle in sfc temps - 0.1-0.2 at most times of day.

    6 Oct 2000 - 1800 UTC - HYBCST (panpbl1.F), HYBCST_PRE

    Reinstate smoothing of bulk Richardson number FG from control copied over. Use new OAFCPRMS with correct nseg_p in both hybcst_pre and hybcst.


    This page prepared by Stan Benjamin

    `########################## END OF HTML INFORMATION ##############`