Hybrid data assimilation

Post Reply
metroid2012
Posts: 8
Joined: Thu Oct 04, 2018 8:31 pm

Hybrid data assimilation

Post by metroid2012 » Thu Dec 27, 2018 7:24 pm

Hi wrfda community,
I tried to run the tutorials in WRFDA support page : 3DEnVar method.
http://www2.mmm.ucar.edu/wrf/users/wrfd ... ybrid.html
They mentioned that before the assimilation stage, there is a need to provide a set of 10 ensemble forecasts. In the tutorial they are named : "wrfout_d01_2016-01-23_00:00:00.e001" up to "wrfout_d01_2016-01-23_00:00:00.e010". My question is if i want to run a custom experiment, how do i produce such files?

JJGuerrette
Posts: 19
Joined: Mon Feb 25, 2019 8:14 pm

Re: Hybrid data assimilation

Post by JJGuerrette » Fri Mar 01, 2019 8:51 pm

You may want to refer to this help topic: viewtopic.php?f=17&t=638

One way to do this is to create a set of perturbed first guess files (as discussed in that help topic) and then run forecasts from all of those fg's. Then rename the wrfout files to have the appropriate suffixes.

JJGuerrette
Posts: 19
Joined: Mon Feb 25, 2019 8:14 pm

Re: Hybrid data assimilation

Post by JJGuerrette » Mon Jun 03, 2019 5:00 pm

As an addendum to my previous reply, using WRFDA in cycling requires the fg to be a wrfinput file, not a wrfout file. So when performing the ensemble of forecasts, you should set the following namelist variables (durations only given as an example):

Code: Select all

 write_input                         = .true.,
 inputout_interval                   = 60,
 input_outname                       = "wrfinput_out_d<domain>_<date>",

Post Reply

Return to “Miscellaneous DA Topics”