[Met_help] [rt.rap.ucar.edu #41699] History for mask_poly?

[RAL HelpDesk {for John Halley Gotway}]

----------------------------------------------------------------
  Initial Request
----------------------------------------------------------------

Good morning John,

Thanks a lot for the kind information about GDAS. These data at ds377 does work for me.

Now I have another question.

(1) How can I define mask_polys for the area I am interested?  And once I define this new mask_poly, do I need to use it for all programs ( pb2nc, point_stat, etc?)  for the same case?

(2) if I use mak_polys in config files, then mask_grids will automatically become invalid?

Thanks a lot!

Ming 

----------------------------------------------------------------
  Complete Ticket History
----------------------------------------------------------------

Subject: Re: [rt.rap.ucar.edu #41699] mask_poly?
From: John Halley Gotway
Time: Fri Oct 22 12:03:02 2010

Ming,

Great, I'm glad you were able to locate the data you needed.

Defining a masking polyline is actually pretty easy.  I'd suggest
looking at one of the sample ones that distributed with the MET
release.  For example, the file "METv3.0/data/poly/CONUS.poly"
contains a rough lat/lon outline for the continental United States.
That file consists of 244 lines of ASCII data.  The first line
contains a name for the polyline ("CONUS") followed by 243 lines
containing lat/lon values (in degrees north and degrees east) that
define the boundary for the verification region.

Here's a few things to note:
(1) You do not need to "close" the polyline.  By default, the last
point you define is connected to the first point to close the polyline
for you.
(2) You have to specify at least 3 lat/lon points to define a
verification region.
(3) All of the grid points falling inside the region are used in the
verification, while those falling outside are not.
(4) In the Point-Stat config file, you can define verification areas
in two places: using "mask_polys" and/or "mask_grids".  You must
define at least one verification area or Point-Stat will error
out.  But if, for example, you define 2 grids and 3 polylines in your
config file, you'll see output for 5 verification areas.  Verification
is performed separately for each one.

Now suppose you've defined a polyline that you want to use.  You asked
the question, "In which tools, should I use it?".  And the answer is,
it's up to you.

I do assume that you'd want to use it in Point-Stat to define the
region over which you want to compute statistics.  You could use it
PB2NC too if you'd like, but you really don't have to.  The PB2NC
tool would use it to filter out which observations are written in the
output NetCDF file.  The advantage would be that you'd retain a
smaller set of observation points - so when you run them through
Point-Stat, it would run faster.  The disadvantage would be that you'd
be "stuck" with just those points - so if you wanted to verify over a
different part of the globe, you wouldn't have observations
for it.

My recommendations is this: try to decide the extent of the geographic
region over which you might want to verify.  And when you run PB2NC on
those global PREPBUFR files, only retain the observations
that fall in that region.

Lastly, if you end up defining a rather complex polyline (on that
contains more than just a few points), I would suggest running it
through the gen_poly_mask tool to define a bitmap.  And then use the
output of the gen_poly_mask tool in the Point-Stat config file.  The
gen_poly_mask tool computes the mask once and then you can use it many
times in the Point-Stat runs.  That'll speed things up a bit
rather than having point_stat compute that mask for your grid every
time you run it.

Hope that helps.

Please send future MET-related questions directly to
"met_help at ucar.edu".  That way we're better able to track our support
work.

John

On 10/22/2010 11:46 AM, RAL HelpDesk {for John Halley Gotway} wrote:
>
> Fri Oct 22 11:46:49 2010: Request 41699 was acted upon.
> Transaction: Ticket created by johnhg
>        Queue: met_help
>      Subject: mask_poly?
>        Owner: johnhg
>   Requestors: chenming at ucar.edu
>       Status: new
>  Ticket <URL:
https://rt.rap.ucar.edu/rt/Ticket/Display.html?id=41699 >
>
>
> Good morning John,
>
> Thanks a lot for the kind information about GDAS. These data at
ds377 does work for me.
>
> Now I have another question.
>
> (1) How can I define mask_polys for the area I am interested?  And
once I define this new mask_poly, do I need to use it for all programs
( pb2nc, point_stat, etc?)  for the same case?
>
> (2) if I use mak_polys in config files, then mask_grids will
automatically become invalid?
>
> Thanks a lot!
>
> Ming

------------------------------------------------