Mesoscale Parallel Experiment Change Log

Experiment Name:  Eta-12 with modified cloud microphysics, assimilation of GOES cloud top pressure data, assimilation of NEXRAD radial winds

When Who What
2002/08/19/12 Lin The precipitation assimilation algorthim has been changed:

  • When the "last resort" contingency is invoked to create a layer of precipitating cloud, rather than saturating the air in the layer, the 'saturation deficit' is first calculated, then an amount of water vapor is added at a rate that would have brought the air from its original state to saturation after an hour.
  • 2002/08/26/12 Lin Added assimilation of GOES cloud top pressure data in the forecast component of the EDAS. See main ETAX page for details.
    2002/09/19/12 Rogers Fixed error in GRDETA code which caused fixed surface fields (e.g. vegetation greenness fraction, albedo) to be incorrectly interpolated to the eta-12 grid. Parallel restarted from operational 12 km EDAS.
    2002/10/30/00 Ferrier Fixed error in new ETAPOST code which caused the categorical precipitation type (Baldwin algorthim) to be computed incorrectly.
    2002/11/04/12 Ferrier Fixed bug in sounding post code (POST0) to eliminate supersaturation of posted model profiles.
    2002/11/05/00 Ferrier Changes to the ETAPOST code:
  • Correct PDS values based on Table 2 (and its addendum) of ON388, in which new fields have been added using a value of PDS Octet 4 equal to 129. Look at the web page at
  • Corrections to the precipitable water calculation (includes water vapor only, no condensate), as well as including new fields of total, column-integrated cloud water, cloud ice, rain, snow, and total condensate (the sum of each hydrometeor specie).
  • 2002/11/14/00 Rogers Included two operational subroutines to the new forecast code which were inadvertently excluded:

  • VTADV.F (with a more robust CFL violation check)
  • SFLX.F : Change to the land-surface physics: Under very dry soil moisture conditions (e.g. in deserts, or in regions with significant soil freezing), a negative soil moisture availability ratio may occur. Under these conditions, this ratio should be zero-ed out to yield no bare soil evaporation, rather than being raised to some exponent (coefficient) as was previously done which can give a non-zero or non-physical solution depending on the coefficient and/or compiler options.

    Due to failures caused by CFL violations, cycled EDAS was broken and therefore restarted from the operational 12 km EDAS.

  • 2002/12/16/00 Rogers Due to numerous failures due to IBM-SP hardware errors. the cycled 12-km parallel EDAS was restarted from the operational 12-km EDAS.
    2002/12/28/00 Lin Began use of merged Stage II/IV precipitation analyses in the parallel 12-km EDAS. The Northwest US RFC precipitation data has been masked out in the precipitation analysis code due to a high percentage of bad hourly data. For more details, please read this description of the merged Stage II/IV product
    2003/01/13/12 Rogers, Parrish Turned back on use of NOAA-16 AMSU-A radiances, which are were turned off in production due to bad data.
    2003/02/05/00 Rogers Changed GRDETA code to obtain e-grid veg type, soil type, and slope type values directly from the nearest neighbor interpolator, instead of getting them from the constants file (nhb1260) from the previous cycle
    2003/02/05/12 Mitchell Parameter CZIL reduced from 0.2 to 0.07. This change will reduce aerodyamnic resistance (i.e. surface turbulent exchange coefficients are too low during mid-day)
    2003/02/12/00 DiMego Parameter CZIL changed back to 0.2 pending further developments.
    2003/02/20/12 Ferrier
  • Increased the grid-scale cloud fraction, which should reduce the 2-m cold bias that has developed over northern latitudes during winter. A cloud fraction of 1.0 (total overcast) is calculated when the relative humidity reaches the threshold value for the onset of condensation (currently 97.7% for the 12-km Eta). This modification is incorporated within the Xu and Randall (1996) cloud fraction parameterization.

  • Redefined the calculation of time-averaged total cloud fraction, assuming random overlap of low (>642 mb), middle (350-642 mb), and high (<350 mb) cloud fractions. The operational code calculates total cloud fraction assuming maximum overlap between the cloud layers. This change has also been made in the post (subroutine CLDRAD) in calculating total, instantaneous cloud cover as the random overlap of low, middle, and high cloud fractions.

  • 2003/04/01/00 Rogers Parallel switched from BSP to frost/snow.
    2003/04/02/00 Lin Began using special hourly prepbufr files with GOES cloud top data only in the prepcloud step.
    2003/04/11/00 Parrish

    New Eta 3DVAR code:

  • Assimilates all channels from NOAA-17 and HIRS channels from NOAA-16

  • Reads BUFR radiance files

  • Background error covariances reduced by 25%
  • 2003/04/11/12 Ferrier Fixed error which caused an erroneous value of variable RHgrid in the first hour of the forecast.
    2003/04/16/12 Lin Use Stage3 RFC map to exclude California-Nevada RFC (CNRFC) data in Stage IV domain. CNRFC seldom sends hourly data, so the part of the CNRFC grid that overlaps with the NWRFC grid usually carries the (often) faulty NWRFC value.
    2003/04/26/00 Rogers Turned off NOAA-17 AMSU-A channel 7 due to instrument problems.
    2003/05/14/00 Rogers Turned off NOAA-16 AMSU-A channel 9 due to instrument problems.
    2003/05/15/00 Keyser Fixed error in radial wind dump job (originally found by Bill Facey) which was excluding about 75% of the radial wind data from being dumped and assimilated into the 3DVAR analysis.
    2003/06/02/12 Ferrier Fixed error in RADTN routine in which the convective and stratiform cloud cover arrays were inadvertently swapped.
    2003/06/05/00 Rogers Since NCO has begun a real-time parallel EDAS test of this system, the EMC parallel was changed to use the GOES cloud top pressure data, merged stage II/IV precipitation analysis, and NEXRAD radial wind data which are being generated for NCO's test.
    2003/06/10/00 Lin Since some RFC's (e.g. Southeast RFC) do not mask out their missing data outside of their range (over water mostly sometimes in Canada), the data we receive that says 'zero' should actually be missing. Since the Stage IV product is used as as the primary input and Stage II as supplemental input for most of the region, so problem will lead to an abrupt drop off in rain areas when it's actually just beyoind the range of the radar.

    To address this problem, the code eta_pcp4to12 has been modified to not use zero values from the Stage IV off the CONUS, and use the Stage II instead.

    2003/06/23/00 Rogers Turned off ETAX forecast job; began use of NCO parallel test runs in ETAX slot.

    Page Last Modified: 15 May 2003