[Dart-dev] DART/branches Revision: 11600

[dart at ucar.edu]

nancy at ucar.edu
2017-05-04 09:29:28 -0600 (Thu, 04 May 2017)
174
add references for jon p's particle filter, update
jeff's SEC reference, and add more details about
some of the options including updating the info
about the new SEC table.




Modified: DART/branches/rma_trunk/assimilation_code/modules/assimilation/assim_tools_mod.html
===================================================================
--- DART/branches/rma_trunk/assimilation_code/modules/assimilation/assim_tools_mod.html	2017-05-04 15:10:51 UTC (rev 11599)
+++ DART/branches/rma_trunk/assimilation_code/modules/assimilation/assim_tools_mod.html	2017-05-04 15:29:28 UTC (rev 11600)
@@ -59,12 +59,14 @@
 <a href="#References"> Anderson 2001</a>)
 <li> 2 = ENKF (Ensemble Kalman Filter)
 <li> 3 = Kernel filter
-<li> 4 = Particle filter
-<li> 5 = Random draw from posterior  (talk to Jeff before using)
+<li> 4 = Observation Space Particle filter
+<li> 5 = Random draw from posterior  (contact dart at ucar.edu before using)
 <li> 6 = Deterministic draw from posterior with fixed kurtosis (ditto)
 <li> 7 = Boxcar kernel filter
-<li> 8 = Rank histogram filter (see
+<li> 8 = Rank Histogram filter (see
 <a href="#References"> Anderson 2010</a>)
+<li> 9 = Particle filter (see
+<a href="#References"> Poterjoy 2016</a>)
 </ul>
 <P>
 We recommend using type=1, the EAKF.
@@ -73,6 +75,8 @@
 the EAKF is identical to the
 EnSRF (Ensemble Square Root Filter)
 described by Whitaker and Hamill in 2002.
+Highly non-gaussian distributions may get
+better results from type=8, Rank Histogram filter.
 </P>
 
 
@@ -329,18 +333,29 @@
  <TD> integer </TD>
  <TD> Selects the variant of filter to be used.
 <UL>
- <LI>1 = EAKF (Ensemble Adjustment Kalman Filter)</LI>
- <LI>2 = ENKF (ENsemble Kalman Filter),
- <LI>3 = Kernel filter</LI> 
- <LI>4 = Observation Space Particle filter</LI>  
- <LI>7 = Boxcar Kernel filter</LI>
- <LI>8 = Rank Histogram Filter</LI>
- <LI>9 = New Particle Filter (see details below)</LI>
-</UL>
+<li> 1 = EAKF (Ensemble Adjustment Kalman Filter, see
+<a href="#References"> Anderson 2001</a>)
+<li> 2 = ENKF (Ensemble Kalman Filter)
+<li> 3 = Kernel filter
+<li> 4 = Observation Space Particle filter
+<li> 5 = Random draw from posterior  (contact dart at ucar.edu before using)
+<li> 6 = Deterministic draw from posterior with fixed kurtosis (ditto)
+<li> 7 = Boxcar kernel filter
+<li> 8 = Rank Histogram filter (see
+<a href="#References"> Anderson 2010</a>)
+<li> 9 = Particle filter (see
+<a href="#References"> Poterjoy 2016</a>)
+<br /> <br />
 
 The EAKF is the most commonly used filter. 
+Note that although the algorithm is expressed in a slightly different 
+form, the EAKF is identical to the EnSRF (Ensemble Square Root Filter)
+described by Whitaker and Hamill in 2002.
+<br /> <br />
+
 The Rank Histgram filter can be more successful for highly nongaussian
-distributsion. 
+distributions. 
+<br /> <br />
 
 Jon Poterjoy's Particle filter is included with this code release.
 To use it rename assimilation_code/modules/assimilation/assim_tools_mod.pf.f90 to
@@ -396,14 +411,16 @@
  <TD> sampling_error_correction </TD>
  <TD> logical </TD>
  <TD>
-If true, uses special input files generated by full_error.f90 in the
-system_simulation directory to reduce errors in the regression step.
-The files are generated for a specific ensemble size.
-They have the name "final_full.X" where X is the number of ensemble members,
-and most common ensemble sizes have precomputed files in that same directory.
+If true, apply sampling error corrections to the correlation values
+based on the ensemble size.
+See <a href="#References"> Anderson 2012</a>.
+This option uses special input files generated by the gen_sampling_err_table
+tool in the assimilation_code/programs directory.
+The values are generated for a specific ensemble size
+and most common ensemble sizes have precomputed entries in the table.
 There is no dependence on which model is being used, only on the number of
 ensemble members.  The input file must exist in the directory where the filter
-program is executing.
+program is executing. 
  </TD> </TR>
 
 <TR>
@@ -748,12 +765,24 @@
 doi: 10.1175/2010MWR3253.1</a>
 <br /> <br />
 </li>
-<li>Anderson, J. L., 2011:,
+<li>Anderson, J. L., 2012:,
 Localization and Sampling Error Correction
 in Ensemble Kalman Filter Data Assimilation.