[Dart-dev] DART/branches Revision: 11354

dart at ucar.edu dart at ucar.edu
Wed Mar 15 16:57:40 MDT 2017


raeder at ucar.edu
2017-03-15 16:57:40 -0600 (Wed, 15 Mar 2017)
432
Corrected more {in,out}put_restart_file_list to {in,out}put_state_file_list,
but not all? (see below).
Fixed typos.  Wordsmithed a little.

Check;
 > write_all_states_at_end is "more memory intensive but requires less time"
 > single_file_out;  'specify a output_state_files'   or ouput_state_file_list?
 > write_all_stages_at_end; what about ...file_list?
 > "For Lanai style dart";  output_state_files or output_state_file_list?




Modified: DART/branches/rma_trunk/documentation/html/Manhattan_diffs_from_Lanai.html
===================================================================
--- DART/branches/rma_trunk/documentation/html/Manhattan_diffs_from_Lanai.html	2017-03-15 22:57:09 UTC (rev 11353)
+++ DART/branches/rma_trunk/documentation/html/Manhattan_diffs_from_Lanai.html	2017-03-15 22:57:40 UTC (rev 11354)
@@ -124,7 +124,7 @@
 <code>add_domain(info_file, num_vars, var_names)</code> which lets DART read 
 the NetCDF dimensions for a list of variables from a file 
 (<code>info_file</code>). There are several routines that can be used together 
-to create a domain from a decsription:
+to create a domain from a description:
 <code>add_domain, add_dimension_to_variable, finished_adding_domain</code>. 
 This can be used in models such as bgrid_solo where the model is spun up in 
 perfect_model_obs, but the model itself has variable structure 
@@ -191,7 +191,7 @@
    <br>&nbsp;&nbsp;preassim_priorinf_mean.nc
    <br>&nbsp;&nbsp;preassim_priorinf_sd.nc
 <li><b>The number of ensemble members specifed</b> in filter_nml (num_output_state_members):
-   <br>&nbsp;&nbsp;preassim_member.000*.nc
+   <br>&nbsp;&nbsp;preassim_member_####.nc
 </ul>
 
 <p>
@@ -206,7 +206,7 @@
    <br>&nbsp;&nbsp;postassim_priorinf_mean.nc
    <br>&nbsp;&nbsp;postassim_priorinf_sd.nc
 <li><b>The number of ensemble members specifed</b> in filter_nml (num_output_state_members):
-   <br>&nbsp;&nbsp;postassim_member.000*.nc
+   <br>&nbsp;&nbsp;postassim_member_####.nc
 </ul>
 <p>
 
@@ -214,6 +214,9 @@
 Note that restarts will have been clamped if any clamping is applied (given 
 as an arguement to add_domain). This is <em>different</em> to Posterior_Diag.nc
 which contains unclamped values.
+Note also that there are 2 more 
+<a href="#STAGES_TO_WRITE">"stages" </a>
+which might be output, in addition to the preassim and postassim discussed here.
 
 <p> For models with multiple domains the filenames above are appended with the 
 domain number, e.g. preassim_mean.nc becomes preassim_mean_d01.nc, 
@@ -293,8 +296,8 @@
 <TBODY valign=top>
 <TR><TD>single_file_in</TD>
     <TD>logical</TD>
-    <TD>True mean that all of the restart and inflation information is read from
-        a single NetCDF file.  False means that you must specify a input_state_file_list
+    <TD>True means that all of the restart and inflation information is read from
+        a single NetCDF file.  False means that you must specify an input_state_file_list
         and DART will be expecting input_{priorinf,postinf}_{mean,sd}.nc files for 
         inflation.
  </TD></TR>
@@ -301,7 +304,7 @@
 
 <TR><TD>single_file_out</TD>
     <TD>logical</TD>
-    <TD>True mean that all of the restart and inflation information is written to
+    <TD>True means that all of the restart and inflation information is written to
         a single NetCDF file.  False means that you must specify a 
         output_state_files and DART will be output files specified in the
         list.  Inflation files will be written in the form 
@@ -329,7 +332,7 @@
         single_file_out must be set to .false.
  </TD></TR>
 
-<TR><TD>output_state_files</TD>
+<TR><TD>output_state_file_list</TD>
     <TD>character array</TD>
     <TD>A list of files containing output model restarts. For multiple domains you can 
         specify a file for each domain. When specifying a list single_file_in, 
@@ -336,12 +339,13 @@
         single_file_out must be set to .false.
  </TD></TR>
 
+<A NAME="STAGES_TO_WRITE"></A>
 <TBODY valign=top>
 <TR><TD>stages_to_write</TD>
     <TD>character array</TD>
     <TD>Controls which stages to write. Currently there are four options:
          <UL>
-          <LI><code>input</code> -- outputs input mean and sd only</LI>
+          <LI><code>input</code> -- writes input mean and sd only</LI>
           <LI><code>preassim</code> -- before assimilation, before prior inflation 
               is applied</LI>
           <LI><code>postassim</code> -- after assimilation, before posterior inflation 
@@ -354,9 +358,9 @@
 <TR><TD>write_all_stages_at_end</TD>
     <TD>logical</TD>
     <TD>True means output all stages at the end of filter.  This is more memory 
-        intensive but requires less memory.  For larger models IO begins to dominate
-        the overall cost of the assimilation so writting all stages at the end write
-        more files in parallel, reducing the IO time.
+        intensive but requires less time.  For larger models IO begins to dominate
+        the overall cost of the assimilation, so writting all stages at the end writes
+        more files in parallel, reducing the IO time. Filenames are defined in
         <code>output_state_files</code>.
  </TD></TR>
 
@@ -376,7 +380,7 @@
 
 <TR><TD>output_sd</TD>


More information about the Dart-dev mailing list