Changes between Version 1 and Version 2 of Ticket #869


Ignore:
Timestamp:
10/12/20 16:10:11 (5 years ago)
Author:
arango
Comment:

Legend:

Unmodified
Added
Removed
Modified

  • Ticket #869 – Description

    v1 v2  
    8686 * [[span(style=color: #FF0000, '''WARNING:''' )]] The CPP option '''NL_BULK_FLUXES''' is deprecated and replaced with '''FORWARD_FLUXES''' and '''PRIOR_BULK_FLUXES'''.
    8787
    88  * The option '''FORWARD_FLUXES''' indicates that the surface forcing fields for the '''TLM''', '''RPM''', and '''ADM''' kernels is from the background nonlinear model trajectory ('''BLK''' file structure) instead of input atmosphere forcing files ('''FRC''' file structure).  Such fields are now stored in the quicksave ('''QCK''') output NetCDF instead of history ('''FWD''') files.  It allows saving the surface forcing frequently (say every 1, 2, or 3 hours) without increasing the history file size. Currently, we have the statement in mosty all the adjoint-based drivers:
     88 * The option '''FORWARD_FLUXES''' indicates that the surface forcing fields for the '''TLM''', '''RPM''', and '''ADM''' kernels is from the background nonlinear model trajectory ('''BLK''' file structure) instead of input atmosphere forcing files ('''FRC''' file structure).  Such fields are now stored in the quicksave ('''QCK''') output NetCDF instead of history ('''FWD''') files.  It allows saving the surface forcing frequently (say every 1, 2, or 3 hours) without increasing the history file size. Currently, we have the following statement in mosty all the adjoint-based drivers:
    8989 {{{
    9090# ifdef FORWARD_FLUXES
    … …  
    112112# endif
    113113}}}
    114  The User needs to provide a reasonable value for '''NQCK''' in '''roms.in''' to correctly sample the daily cycle. Please avoid activating the '''Qout''' switches for lots of fields to be written in the quicksave file, so it is relatively small. Avoid activating switches for 3D variables. Optimally, we need to activate the following switches:
     114 The User needs to provide a reasonable value for '''NQCK''' in '''roms.in''' to correctly sample the daily cycle. Please avoid activating the '''Qout''' switches for lots of fields written in the quicksave file, so it is relatively small. Avoid activating switches for 3D variables. Optimally, we need to activate the following switches:
    115115 {{{
    116116Qout(idsurT) == T T     ! temp_sur, salt_sur surface temperature and salinity
    … …  
    169169 All the '''_FillValue''' attributes for output variables in the '''MOD''' NetCDF file were removed. Many thanks to John Wilkin for his suggestions to improve the output '''MOD''' NetCDF file.
    170170
    171  * Added option '''REGRID_SHPIRO''' to apply a Shapiro filter in routine '''regrid.F''' called by '''nf_fread2d''' to remove spikes and noise due to the bilinear interpolation of very coarse input 2D data.
     171 * Added option '''REGRID_SHPIRO''' to apply a Shapiro filter in routine '''regrid.F''' called by '''nf_fread2d''' to remove spikes and noise due to the bilinear interpolation of very coarse 2D input data.
    172172
    173173----
    … …  
    181181 * '''split_rbl4dvar_ocean.h:'''  Split RBL4D-Var algorithm. It uses '''submit_split_rbl4dvar.sh''' to run the driver.
    182182
    183 The bash scripts in '''ROMS/Bin''' are complex and well documented. Several test cases were added to the '''test''' repository for the '''WC13''' application.  It took me several months to code and test these new drivers. We will add documentation to '''wikiROMS''' about the new 4D-Var dirvers in the near future.
     183The bash scripts in '''ROMS/Bin''' are complex and well documented. Several test cases were added to the '''test''' repository for the '''WC13''' application.  It took me several months to code and to test these new drivers. We will add documentation to '''wikiROMS''' about the new 4D-Var drivers soon.
    184184
    185 More complexity is being added and tested for these drivers like coarser grid resolution in the '''increment''' phase during the inner-loops. The increment phase may be run at a lower precision.
     185More complexity is being added and tested for these drivers like coarser grid resolution in the '''increment''' phase during the inner-loops. The '''increment''' phase may be run at a lower precision.
     186
    186187I am currently testing the coupled 4D-Var that allows the '''background''' phase to be part of a coupling system. The coupling between ROMS and the atmosphere model is done with the '''ESMF/NUOPC''' library.  Such capabilities will be released in the future.
    187188