[Dart-dev] DART/branches Revision: 11490

[dart at ucar.edu]

nancy at ucar.edu
2017-04-14 10:31:53 -0600 (Fri, 14 Apr 2017)
262
add a pointer to where the original lanai values for SEC
are located, and add a little usage help if someone looks
here.  we still have the problem that people need to 
find this place from their model/work directory so we
need a pointer the other way to here.




Modified: DART/branches/rma_trunk/assimilation_code/programs/gen_sampling_err_table/work/README
===================================================================
--- DART/branches/rma_trunk/assimilation_code/programs/gen_sampling_err_table/work/README	2017-04-14 16:14:07 UTC (rev 11489)
+++ DART/branches/rma_trunk/assimilation_code/programs/gen_sampling_err_table/work/README	2017-04-14 16:31:53 UTC (rev 11490)
@@ -4,6 +4,26 @@
 #
 # DART $Id$
 
+Usage Notes:
+
+The netcdf file in this directory should be copied to the
+working directory where filter is running.  If you need an
+ensemble size which isn't currently in the file the
+gen_sampling_error_table program can be used to append
+new ensemble sizes to the existing file.
+
+The values here are generated by running a random number
+generator for a long time.  Statistically they are equivalent
+to the previous data.  If you are trying to bitwise
+reproduce results from the Lanai release and SEC was enabled, 
+one of the (many) things you will need to do is pick up the 
+sampling error correcton file from the 'system_simulation/work' 
+directory.  That one was constructed from the original final_full.XX
+files and not regenerated with different random values.
+
+
+Theory Discussion (from Jeff Anderson):
+
 Looking at sampling error correction factor for regression
 for ensemble filters. Basic idea here is that there is 
 some noise so that the sample statistics provide a