[Dart-dev] DART/branches Revision: 12148

[dart at ucar.edu]

thoar at ucar.edu
2017-11-28 16:17:35 -0700 (Tue, 28 Nov 2017)
132
Updating instrument list and use 'add_obs_to_seq' to speed up
the conversion of COAMPS observations to DART observation sequences.




Modified: DART/branches/coamps/models/coamps_nest/README
===================================================================
--- DART/branches/coamps/models/coamps_nest/README	2017-11-28 22:05:37 UTC (rev 12147)
+++ DART/branches/coamps/models/coamps_nest/README	2017-11-28 23:17:35 UTC (rev 12148)
@@ -62,6 +62,71 @@
 coamps structure does not actually allocate any space for the state, 
 but does have the variable names we want - except for the nuisance variables.
 
+
+-------------------------------------------------------------------------------
+Mon Nov 13 12:51:13 PST 2017 - conf call with Sasa, Sergey
+
+*) What fields in the HDF5 file are used for restarting - hop test material?
+    which tau do we use comes from ktauf
+
+*) Where do the observations come from? 'Innovations' ... from what?
+
+*) Do you want to use the DART forward operators or the ones defined in obs_def_navdas_mod.f90 - they use non-standard observation types - and there is a clash with the get_expected_altimeter() function
+
+*) time ... what is going on with the 'model_mod: get_dart_time' ... is cdtg always from the convert.nml
+
+*) the default perturb routine needs to be swapped out for the coamps_pert_mod:perturb_state routine
+
+*) go over scripting
+
+*) exner functions ... inflation/perturbation nomenclature ... are these 'mean variables' seems like coamps_intrinsic_mod.f90 defines them based on the standard atmosphere - so they are never changing - is that true?
+
+*) what are assumptions about what needs to be staged to advance coamps till next observation time
+
+*) Meeting with Tim W. indicated the fields in the hdf5 file may be 'unstaggered'.
+   if we update the 'fcstfld' or 'analfld' ... what does coamps use to restart?
+   Do we use the variables in the hdf5 file or do we use the routines based on
+   the input parameters (which are alread in use for the model_interpolate code
+   that takes into account the variable stagger - but get_state_meta_data does not,
+   so the localization distances are off).
+   coamps_interp_mod:extract_neighbors() 
+U(1) is EAST of first mass point, V(1) is NORTH first mass point
+
+*) the coamps_netcdf_mod:nc_write_prognostic_atts() routine writes out stuff I do not believe 
+   to be correct given the staggering uncertainty ... I will remove this as soon as the scripting is done
+
+*) model_mod_check on eddy with PG compiler: -- TEST 2 -- Read and write restart file
+0: DEALLOCATE: memory at (nil) not allocated
+
+
+-------------------------------------------------------------------------------
+Tue Nov 28 09:55:48 PST 2017
+
+Frolov, Dr. Sergey
+Nov 15 (13 days ago)
+
+"Tim,
+In case you want to see an example of the innovation file for this DTG and domain.
+Eddy in /home/nopp/COAMPS_hdf5_files
+ 
+Innovation file innov_a_2013011000
+Some description of this format is in “Innovation Vector Description.pdf”
+However, it is not the exact description. This document is actually for a related system that produces this innovation vector. You can pretty much ignore all the stuff about the radiances.
+ 
+Matlab program to read the file is in rd_ainv.m
+In matlab
+[n ob bk t_bk iv err etc lat lon p vty ity nvp chk dt pf org idp rh_bk] = rd_ainv(fn);"
+
+Tim: updated the instru_list_coamps.inc based on new table (Sergey got me one from Pat Pauley)
+and was able to read the innov_a_2013011000 file. It had instruments not in the .inc file
+and highlighted a problem in the navdas_innov_mod.f90
+I also replaced the calls to 'insert_obs_in_sequence()' with 'add_obs_to_seq()' which
+does not do an exhaustive search through the sequence to insert at proper time. The (optimized) run-time
+went from almost 700 seconds to 14 seconds. Also added the 'ngt' filename to the input.nml.
+'ngt' appears to be something related to vortex locations - should get a much better name ...
+
+
+
 # <next few lines under version control, do not edit>
 # $URL$
 # $Revision$

Modified: DART/branches/coamps/models/coamps_nest/instru_list_coamps.inc
===================================================================
--- DART/branches/coamps/models/coamps_nest/instru_list_coamps.inc	2017-11-28 22:05:37 UTC (rev 12147)
+++ DART/branches/coamps/models/coamps_nest/instru_list_coamps.inc	2017-11-28 23:17:35 UTC (rev 12148)
@@ -1,121 +1,145 @@
 ! This code is not protected by the DART copyright agreement.
 ! DART $Id$
 
-! rcs keywords: &RCSfile& 
-!               &Revision: 2820 & &Date: 2010-07-26 14:35:27 -0500 (Mon, 26 Jul 2010) &
-! SCCS IDENTIFICATION:  %W% %G% 
-!                       %U% %P% 
-!
 !     master list for the instruments identified by NAVDAS
-!
-      integer, parameter :: mx_nm_instru=99
+!     derived from ob_list/instru_list.f
+
+      integer, parameter :: mx_nm_instru=126
       integer :: nc(mx_nm_instru)
       integer :: nsat(mx_nm_instru)   !  set to 1 for satellite platforms
       character(len=16) :: c_label(mx_nm_instru) ,c_rec(mx_nm_instru)
       integer :: ival(mx_nm_instru)
       integer :: m