Re: NCL vs. WRF_NCL

On Wed, 16 May 2007, Don Morton wrote:

> Hi, we're going to have a couple of students spending the
> summer helping to evaluate WRF performance as compared
> with observations, Real-Time Mesoscale Analysis, etc. This
> will mean lots of graphics produced, lots of "diff" graphics
> (depicting difference between variables in different model
> runs), etc. NCL will be our means for producing all of these.
>
> I'm still quite the beginner with NCL, but thanks to the many
> wonderful examples on the NCL website, I've found it to be
> somewhat trivial to produce visuals of my WRF netCDF output,
> so I'm pretty happy with it all and feel like this will serve our needs,
> and will offer the flexibility to additionally view GRIB output files,
> output files from other models, etc.
>
> However, I'm also aware of, and have tried, WRF_NCL. It's my
> somewhat uneducated view that WRF_NCL is simply a set of
> higher level functions and scripts tuned to the purpose of producing
> nice graphics strictly from the netCDF WRF output files. This is all
> great, and I can see many uses for it, but I suspect it may be a little
> too specialized for our work right now. Am I correct in my assessment?

Hi Don,

You have some good questions here. I will try to explain a little of
what's going on, since I'm going to be involved with helping update
the WRF_NCL graphical scripts. However, I would be delighted if any
other WRF folks who are also familiar with NCL would jump in and offer
their opinions.

You are right that the WRF_NCL scripts are a high-level suite of
scripts for generating a particular suite of plots. In fact, I *think*
the way this suite works is that you indicate which WRF variables you
are interested in seeing, and it goes off and generates a series of
plots for each one. With this suite of scripts, you don't have as
much control over the look of the plots, and that's something that
we're hoping to remedy.

This is different than the way the more generic gsn_csm_xxxx scripts
work, in which you would probably have to set a few resource options
in order to get plots similar to what WRF_NCL produces. But, you do
have more control over how the plots look.

Also, I should point out that the NCL Applications page has a link to
a WRF examples page:

    http://www.ncl.ucar.edu/Applications/wrf.shtml

which is a set of graphical examples that Dennis Shea created. These
examples mostly use the gsn_csm_xxxx scripts, but they also set some
plot resources for you based on attribute information in the WRF file
(like map boundary limits).

> In a few weeks I'll need to be getting these students up to speed on NCL,
> and my inclination is to initially avoid WRF_NCL and focus on NCL
> (they're good computer science students, so I think they'll cope fine).
> Can anybody think of a reason why it would be a mistake to rely solely
> on NCL and avoid WRF_NCL for now? I'm just wondering if perhaps
> WRF_NCL addresses any peculiar issues that NCL might have in dealing
> with the WRF netCDF output files.

I would say that depending on what kind of plots you want to generate,
you could go either way.

If you want a more high-level canned set of scripts with less control
over how they look, then you should go with WRF_NCL. As I mention,
we hope to update the WRF_NCL suite of scripts to make it more accessible
to the user in terms of setting more plot options.

If you want more control over the plots, then you should consider
looking at the WRF page I mentioned above, or even just look at the
gsn_csm_xxx suite of scripts. I have to caution, however, that WRF
netCDF files are somewhat different than CSM files, which the
gsn_csm_xxx scripts were created for. For one example (as mentioned on
Dennis' WRF page), the CCSM grids do not change with time, while the
WRF model allows for this possibility by making the latitude and
longitude arrays a function of time.

So, you have to know your WRF files well, and understand how to use
the information in them to get the correct plots.

Please visit the WRF page above, and then, of course, visit the
WRF_NCL page:

    http://www.ncl.ucar.edu/Applications/wrf.shtml
    http://www.mmm.ucar.edu/wrf/users/graphics/WRF_NCL/NCL.htm

I don't know if they have any graphical examples that you can
view on the WRF_NCL page. I didn't see anything obvious.

Feel free to email me directly if you want to talk some more about
this.

--Mary

> Thanks for any guidance,
>
> Don Morton
> Arctic Region Supercomputing Center / U. Montana
> _______________________________________________
> ncl-talk mailing list
> ncl-talk_at_ucar.edu
> http://mailman.ucar.edu/mailman/listinfo/ncl-talk
>
_______________________________________________
ncl-talk mailing list
ncl-talk_at_ucar.edu
http://mailman.ucar.edu/mailman/listinfo/ncl-talk
Received on Thu May 17 2007 - 10:27:17 MDT