GSI Change log

Changes to the GSI code are posted below.  Click on the link in the Details column for more information regarding a given update.  To return to the main GSI code page, click here.

Page last modified: 9/18/2006 2:54 PM

Posted

Brief description of changes

Details

3/22/2004

·        Initial gsi release

20040319_emc.change

3/24/2004

o       Bug fixes

o       Code cleanup

20040323_dp_jd.change

3/30/2004

·        Code cleanup/speedup

·        Modify limq behavior

·        Bug fixes

20040329_jd.change

4/7/2004

o       Code cleanup

o       Bug fixes

o       Add radar wind superobs

o       Improve skin temperature analysis

20040407_emc.change

5/5/2004

·        Code cleanup

·        Improve efficiency/memory usage

·        Bug fixes

·        Modify negative moisture constraint

·        Alter radiance assimilation

·        Add SGI SCSL library calls

·        Add option to output regional skin temperature analysis

20040505_emc_gmao.change

5/11/2004

o       Bug fix in routine htoper.f90

20040511_jd.change

6/2/2004

·        Code cleanup

·        Improved documentation

·        Ability to process/assimilate GPS local refractivity

·         Simplify and unify multiplication by background error routines

20040602.txt

7/2/2004

o       Remove obsolete routines

o       All routines documented

o       Changes in code structure

o       Bug fixes

20040702.txt

8/4/2004

·        Add WRF MASS core option

·        Redefine recursive filter lookup table

·        Add generalized grid transform (regional only)

·        Alter radiance thinning algorithm to use less memory

·        Add new snow/ice emissivity model for AMSU-A/B

·        Convert documentation to Protex-compliant format (only a few routines)

·        Improve code reproducibility; bug fixes and general code clean up

20040804.txt

9/16/2004

o       Several bug fixes

o       Ability to directly read/write WRF mass core and nmm core input/output files in either binary or netcdf format

o       Reorder code for application of background error

o       Change definition of scaling parameters for background error

o       Improve efficiency of horizontal recursive filter

o       Move radiative transfer code and related routines to library

o       Replace grid « spectral transforms with calls to NCEP splib routines

o       Replace type_kinds module with kinds module (redefine data types)

o       Remove –qrealsize=8 from NCEP CSS gsi build

o       Increase weight given to IR and MSU brightness temperatures

20040916.txt

11/04/2004

·        Correct code bugs and clean up code

·        Add option to perform nonlinear quality control of observations

·        Update CRTM and IRSSE libraries

·        Modify how code adjusts observation error for certain observations (conventional and brightness temperature observations)

20041104.txt

12/08/2004

o       Add OpenMP directives to select routines

o       Include regularity in table definition for horizontal recursive filter

o       Replace weighting of observations around boundary with logical true false flag

o       Modify regional input/output to use MPI Input/Output for WRF binary format files

o       Expand variance array "dssv" to include longitude dimension (variance still only function of latitude and level … addition of longitude dimension for future applications)

o       Declare parameters for the smallest and largest machine-representable floating point numbers

o       Numerous changes to processing and assimilation of GPS local refractivity

o       Replace IBM MPI extensions (so-called MPE routines) with standard MPI calls

20041208.txt

01/19/2005

·        Modify AIRS, ozone, and Navy AVHRR data read routines

·        Correct comments, replace complex intrinsic function calls with generic interface (e.g., use exp instead of zexp)

·        Modify brightness temperature thinning algorithm to only allocate thinning grid no larger than analysis domain.  Also correct small bug in specification of latitudes for thinning grid boxes.  Result of bug fix is to allow slightly more observations into analysis.

·        Replace use of geopotential function from utility library with simple integration of hypsometric equation.

·        Various namelist changes

­       Add logical array to namelist to toggle on/off generation of innovation diagnostic files

­       Add logical flag to toggle on/off SST retrieval

­       Remove obsolete namelist variables à wrf_anl_filename and wrf_ges_filename

­       Increase jpch from 493 to 535 (see NOAA-18 change below)

·        Prepare for future use of NOAA-18 HIRS/3, AMSU-A, MHS, and AVHRR by expanding fix files related to brightness temperature assimilation (e.g., global_satinfo.txt, CRTM & IRSSE coefficients, etc).  This change increase the number of entries in the satinfo file from 493 to 535.

·        Explicitly type all integers to be type to be type i_kind.  As currently set in kinds.f90, type i_kind is a four byte integer

20050119.txt

02/02/2005

o       Modify background related routines to simplify the task of adding an anisotropic recursive filter option.

o       Fix code errors (bugs)

o       Reorganize GSI namelist variables into more function-based namelist groupings

o       Make global background error consistent with statistics code that generates the error file.

o       Add ability to assimilate SSM/I brightness temperatures

o       Upgrade global surface file input/output to soon-to-be implemented sfcio module

o       Modify thinning algorithm for brightness temperature observations.  Correct algorithm error and improve granularity of surface types.

o       Make numerous changes to processing of local GPS refractivity data.

o       Set “dval” parameter for non-assimilated brightness temperature data sets to 0.0 to avoid this data from influencing the use of other brightness temperature data.

20050202.txt

03/14/2005

·        Modify algorithm for determining surface type in setuprad

·        Make some changes to quality control for satellite data

·        Update prepp to make it less likely for observations with large height differences to pass quality control

·        Modify surface emissivity adjoint routines to simplify

·        Clean up qcssmi routine

·        Add option (qoption=2) for normalized RH variable in place of current pseudo RH (qoption=1)

·        Add option for anisotropic background errors.

·        Correct error in intref.f90 which caused NaNQ's to appear when nonlinear qc turned on

·        Modify statspcp.f90 to include printout of number of precip obs rejected by nonlinear qc

·        Change nonlinear qc to account for inflation of observation errors

·        Read GPS radio occultation (ro) observations from BUFR format files

·        Modify the QC checks for the GPS ro observations

·        Reduce the value of the gross error for GPS ro observations

·        Reorganize and clean up portions of code related to sst retrieval

·        Correct error in horizontal scale weights (hzscl)

·        Clean up diagnostic output

·        Add #ifdef to isolate sections of code which require WRF library and / or module files

·        Clean up code, correct variable declaration sections, correct allocate / deallocate errors

·        Honor METAR surface pressure quality flags and observation errors

20050314.txt

04/15/2005

o       Add observation time as quantity saved in obsmod for future use in minimization

o       Replace mlath in regional background stats code with mlat

o       Address quality control issues with GPS refractivity data

o       Replace spectral method to obtain vorticity and divergence   increments following inner loop with compact differencing   method

o       Merge functionality of linear and nonlinear quality control int*90 and stp*90 routines

o       Include qoption=1 and =2 moisture statistics in global background error statistic file.  Correct bug in calculation of qoption=1 error statistics.

20050415.txt

06/09/2005

·        Add capability to handle GMAO background fields on their primitive vertical coordinate system.

·        Add TLM code/capability. Only inner loop has been dealt with so far.  Setup for TLM is still somewhat preliminary and is in works.

·        Move some of the fixed files inside GSI source code directory:  global_ozinfo.txt, global_satinfo.txt, global_pcpinfo.txt

·        Add code to compute horizontal derivatives. 

·        Improve QC checks for gps local refractivity data

·        Add 2DVAR (surface analysis) option to regional GSI.

·        Provide option to fix problem with cascade interpolation near the pole. 

·        Change runtime innovation information (fort.2* files) to include more level by level information. 

·        Modify minimization algorithm.  Remove terms and calculations no longer needed.

·        Change lower limits on error bounds for T, ps, and uv when reading in observation error table

·        Add option to break model layers into smaller sub-layers when calling community radiative transfer model. 

·        Change regional and global background error statistics files from 8-byte to 4-byte reals.  The same change was made for the optional guess solution input/output file.

20050601.txt

08/16/2005

o       Added modules to interface with GMAO horizontal grid. In all   platforms this adds dependency on other GMAO libraries (hermes, mpeu, transf, gfio, hdf, etc), except on AIX.

o       Bug fix that now allows GSI to read two or more prepbufr files correctly.

o       Modify to use NOAA-18 and EARS data.  Remove satellite dependent thinning criterion from read_bufrtovs, read_airs, and read_goesndr.  Changes made to thinning of AMSU-B observations.

o       Modify to allow monitoring of observations (qc and compute innovations but do not assimilate).  Observations with quality marks of 9 and 14 will be monitored.  Changes printout to reflect stats for rejected observations.

o       In order to apply background error covariances calculated from central NMM forecasts to the whole North America domain, sin(lat) is separated from the balance projection

o       Include both qoption=1 and =2 statistics in single regional background error statistics file.

o       Add capability to assimilate OMI total ozone

o       Update PINT (3d non-hydrostatic pressure field) in regional NMM output

o       Bug fix; allow regional background error statistics from a smaller domain to be used in a larger domain.

o       Add preliminary code for dynamic constraint term

o       Allow specification of conventional data usage, error limits, and variation qc parameters through text input file.

o       Added capability to process level2 radar bufr files.

20050801.txt

10/20/2005

·        Fix minor bugs and clean up (simplify) code

·        Add logic to identify and skip over bad NOAA-17 SBUV data

·        Simplify and make more efficient the data handling, setup, int and stp routines

·        Tune regional observation error table based on adaptive tuning experiments

·        Change time window in regional_convinfo.txt file to 1.5 hours (was 3.0 hours.  1.5 hours is consistent with 3 hourly cycle of regional data assimilation system)

·        Regional mode bug fix --> set ozmz values

·        Turn off ozone channels in regional_satinfo.txt since regional model does not carry ozone as a prognostic variable

·        Expand dynamic constraint term, making using of generalized tendency routine.  Code may be applied in both global and wrf_nmm regional mode

·        Fix bug in update_guess routine

·        Fix bug in setupsrw which gave incorrect diagnostic output

·        Fix date handling in setupsrw to be same as other data type read routines

·        Add ability to read and process AMRSE and SSM/IS

·        Fix problems with diagnostic files