[Met_help] [rt.rap.ucar.edu #71822] History for MODE Config - merge flag and merge threshold

[John Halley Gotway via RT]

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

Hi,

I'm trying to understand what the double merging technique means and how
should I choose the threshold. I cannot understand what's being explained
in the manual (v5.0) for the merge_thresh option.

It says: These thresholds should be chosen to define larger objects that
fully contain the originally defined objects. For example, for objects
defined as >=5.0, a merge threshold of >=2.5 will define larger objects
that fully contain the original objects. Any two original objects contained
within the same larger object will be merged.

First of all, what are the numbers used in the example? I assume it's the
object area in grid squares. Even if it is indeed the area, the criteria
for merging doesn't make sense. My interpretation of the same example is,
if object "1" has 5 grid squares and object "2" has 2.5, then object "2" is
merged in a cluster with object "1". Any object can be merged to another as
long as the larger object has an area twice the size of the smaller one.
But how about the distance between the objects, it should be considered
somehow. I'd appreciate if you could try to explain it in a different way...

The I have another question about the merge flag:
For the merge flag there's an option for using BOTH techniques (double
thresh and engine) but my tests show that using BOTH is the same as using
ENGINE. What's the criteria that decides which one is used?

Thanks so much

Maria Eugenia B. Frediani
-------------------------------------------------------------------------------------
PhD Candidate
School of Engineering - University of Connecticut
maria.frediani at uconn.edu

Advanced Study Program Visitor
Research Applications Laboratory
National Center for Atmospheric Research
P.O. Box 3000
Boulder, CO 80307
frediani at ucar.edu <maria.frediani at ucar.edu>


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

Subject: MODE Config - merge flag and merge threshold
From: John Halley Gotway
Time: Thu May 07 09:58:24 2015

Maria,

I see that you have some questions about the "double-thresholding"
technique for MODE.

The values used for the thresholds you referenced are the data values
over
which the MODE objects are being defined.  Let's suppose we have
accumulated precip with units in millimeters.  You're interested in
areas
of accumulated precip exceeding 5 mm.  So you set the object
definition
threshold (conv_thresh) as >=5.0.  But you enable double-thresholding
and
set the merge threshold (merge_thresh) as >=2.5.  MODE creates the 5.0
mm
objects and the 2.5 mm objects.  Any 5.0 mm objects that end up inside
the
same 2.5 mm object get "merged" together.

When double-thresholding is enabled, MODE creates extra pages in the
output
PostScript file showing how it worked on your data.  I've attached
page 6
of MODE PostScript output that's generated by the test scripts.  I
think
the picture explains it much better than my words can!  The areas in
blue
are the objects that were defined.  The blue outlines around the
objects
show the objects defined by the lower merge threshold.  Any two blue
objects falling inside the same outline get merged together.  In this
example, it looks like the objects over Texas and eastern Oklahoma
would be
merged.  And the large object over the Southeast US would be merged
with
the one over Kentucky.

If you have more questions, feel free to stop by.  My office is in
FL3-1018.

Thanks,
John Halley Gotway
met_help at ucar.edu


On Wed, May 6, 2015 at 4:24 PM, Maria Eugenia via RT
<met_help at ucar.edu>
wrote:

>
> Wed May 06 16:24:51 2015: Request 71822 was acted upon.
> Transaction: Ticket created by maria.frediani at gmail.com
>        Queue: met_help
>      Subject: MODE Config - merge flag and merge threshold
>        Owner: Nobody
>   Requestors: maria.frediani at gmail.com
>       Status: new
>  Ticket <URL:
https://rt.rap.ucar.edu/rt/Ticket/Display.html?id=71822 >
>
>
> Hi,
>
> I'm trying to understand what the double merging technique means and
how
> should I choose the threshold. I cannot understand what's being
explained
> in the manual (v5.0) for the merge_thresh option.
>
> It says: These thresholds should be chosen to define larger objects
that
> fully contain the originally defined objects. For example, for
objects
> defined as >=5.0, a merge threshold of >=2.5 will define larger
objects
> that fully contain the original objects. Any two original objects
contained
> within the same larger object will be merged.
>
> First of all, what are the numbers used in the example? I assume
it's the
> object area in grid squares. Even if it is indeed the area, the
criteria
> for merging doesn't make sense. My interpretation of the same
example is,
> if object "1" has 5 grid squares and object "2" has 2.5, then object
"2" is
> merged in a cluster with object "1". Any object can be merged to
another as
> long as the larger object has an area twice the size of the smaller
one.
> But how about the distance between the objects, it should be
considered
> somehow. I'd appreciate if you could try to explain it in a
different
> way...
>
> The I have another question about the merge flag:
> For the merge flag there's an option for using BOTH techniques
(double
> thresh and engine) but my tests show that using BOTH is the same as
using
> ENGINE. What's the criteria that decides which one is used?
>
> Thanks so much
>
> Maria Eugenia B. Frediani
>
>
-------------------------------------------------------------------------------------
> PhD Candidate
> School of Engineering - University of Connecticut
> maria.frediani at uconn.edu
>
> Advanced Study Program Visitor
> Research Applications Laboratory
> National Center for Atmospheric Research
> P.O. Box 3000
> Boulder, CO 80307
> frediani at ucar.edu <maria.frediani at ucar.edu>
>
>

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