[Dart-users] updates to the rma_trunk: endian-ness, external forward operators

[Nancy Collins]

hello all --

i just committed several changes this morning to the rma_trunk version.
if you update, please update from the top level directory to get all the
changes.

a couple changes in particular may be of particular interest to some of you.

first, the obs_sequence read code can now open a binary obs_sequence file
which was written on a machine with a different "endian-ness" and read it.
there is a new namelist item in &obs_sequence_nml:

    read_binary_file_format   = 'native'

valid string values are:   native, little_endian, big_endian


second, the "external forward operator" code was committed.  this is an 
option
to add externally computed forward operator values into the obs_seq.out file
and have filter read and use those values instead of calling the model 
interpolate
code.  this allows the dart filter to use forward operator values 
computed by a
model at the exact time of the observation if it has the ability to 
output those values,
or allow filter to use forward operators computed by an external 
program, either
to take advantage of bias correction or specialized forward operator 
code, or to
do head-to-head comparisons between assimilation systems using the same 
values.
thanks to craig schwartz for the implementation of this new feature.

if you're interested in using this new facility, email us at 
dart at ucar.edu for more
details.

if you're looking at using the RMA version for the first time, be warned 
that it is
not 100% backwards compatible like we've tried to do in the past. check 
out the
code and look at rma_doc/rma.html for a list of major differences in the 
functionality
and then the second half of that document describes namelist differences 
from the
current version.

as always, please email dart at ucar.edu if you've got any other questions 
or problems,
nancy