[Met_help] [rt.rap.ucar.edu #53003] History for Regarding scale specification in wavelet-stat tool of MET

Fri Jan 27 09:53:14 MST 2012

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

Respected Sir/Madam,

                   I wish to use wavelet stat tool for verification.
I am using tile_dim=64
so NSCALE is 6
now according to my understanding, ISCALE 0 to 5 (i.e. n=0 to n-1) is for
mother wavelet component and ISCALE = 6 is for father wavelet component
(domain mean), then ISCALE=7 stands for what ? There are in total 8
categories of ISCALE (i.e. 0 to 7) are appearing on line type ___isc.txt .
Kindly explain.

Thanks and Regards
Mansi Bhowmick
Project Scientist
India Meteorological Department

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

Subject: Regarding scale specification in wavelet-stat tool of MET
From: John Halley Gotway
Time: Wed Jan 25 08:52:04 2012

Mansi,

I ran a test case of Wavelet-Stat using a tile dimension of 64 (which
is 2 to the 6th power), and I attached some images of the output.

Also below, I've listed an excerpt from section 7.2.1 of the MET
User's Guide:

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

The binary forecast and observation fields obtained from the
thresholding are then decomposed into the sum of components on
different scales, by using a 2D Haar wavelet filter (Fig 7.2). Note
that the
scale components are fields, and their sum adds up to the original
binary field. For a forecast defined over square domain of 2^n grid-
points, the scale components are n+1: n mother wavelet
components + the largest father wavelet (or scale-function) component.
The n mother wavelet components have resolution equal to 1, 2, 4, ...
2^(n-1) grid-points. The largest father wavelet component
is a constant field over the 2^n grid-point domain with value equal to
the field mean.

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

In the case of 64 = 2^6, we have 6 mother wavelet components with
resolutions equal to 2^0 = 1, 2^1 = 2, 2^2 = 4, 2^3 = 8, 2^4 = 16, and
2^5 = 32 grid points.  And we have the largest father wavelet
which is constant over the 2^6 = 64 grid-point domain.  So in the
attached output images, you should see pictures for the raw field, the
thresholded binary field, 6 mother wavelets, and one father
wavelet.

In the process of looking at this, I do see that our documentation
says "2^n x 2^n" in a few places when it should just say "2^n".
Perhaps that's a source of confusion.  I'll make sure we correct
that for the next release.

Hopefully, that answers your question.  If there's still some
confusion, please let us know.

Thanks,
John Halley Gotway
met_help at ucar.edu

On 01/25/2012 03:03 AM, Mansi via RT wrote:
>
> Wed Jan 25 03:03:50 2012: Request 53003 was acted upon. Transaction:
Ticket created by mansibhowmick at gmail.com Queue: met_help Subject:
Regarding scale specification in wavelet-stat tool of MET
> Owner: Nobody Requestors: mansibhowmick at gmail.com Status: new
Ticket<URL: https://rt.rap.ucar.edu/rt/Ticket/Display.html?id=53003>
>
>
> Respected Sir/Madam,
>
> I wish to use wavelet stat tool for verification. I am using
tile_dim=64 so NSCALE is 6 now according to my understanding, ISCALE 0
to 5 (i.e. n=0 to n-1) is for mother wavelet component and ISCALE
> = 6 is for father wavelet component (domain mean), then ISCALE=7
stands for what ? There are in total 8 categories of ISCALE (i.e. 0 to
7) are appearing on line type ___isc.txt . Kindly explain.
>
> Thanks and Regards Mansi Bhowmick Project Scientist India
Meteorological Department

------------------------------------------------
Subject: Re: [rt.rap.ucar.edu #53003] Regarding scale specification in wavelet-stat tool of MET
From: John Halley Gotway
Time: Wed Jan 25 09:02:55 2012

Mansi,

Please disregard my comment about the documentation issue - where it
says 2^n x 2^n in section 7.2.1 of the documentation is correct.  I
was just confusing myself!

Thanks,
John

On 01/25/2012 08:52 AM, John Halley Gotway wrote:
> Mansi,
>
> I ran a test case of Wavelet-Stat using a tile dimension of 64
(which is 2 to the 6th power), and I attached some images of the
output.
>
> Also below, I've listed an excerpt from section 7.2.1 of the MET
User's Guide:
>
>
-----------------------------------------------------------------------------
>
> The binary forecast and observation fields obtained from the
thresholding are then decomposed into the sum of components on
different scales, by using a 2D Haar wavelet filter (Fig 7.2). Note
that the
> scale components are fields, and their sum adds up to the original
binary field. For a forecast defined over square domain of 2^n grid-
points, the scale components are n+1: n mother wavelet
> components + the largest father wavelet (or scale-function)
component. The n mother wavelet components have resolution equal to 1,
2, 4, ... 2^(n-1) grid-points. The largest father wavelet component
> is a constant field over the 2^n grid-point domain with value equal
to the field mean.
>
>
-----------------------------------------------------------------------------
>
> In the case of 64 = 2^6, we have 6 mother wavelet components with
resolutions equal to 2^0 = 1, 2^1 = 2, 2^2 = 4, 2^3 = 8, 2^4 = 16, and
2^5 = 32 grid points. And we have the largest father wavelet
> which is constant over the 2^6 = 64 grid-point domain. So in the
attached output images, you should see pictures for the raw field, the
thresholded binary field, 6 mother wavelets, and one father
> wavelet.
>
> In the process of looking at this, I do see that our documentation
says "2^n x 2^n" in a few places when it should just say "2^n".
Perhaps that's a source of confusion. I'll make sure we correct that
> for the next release.
>
> Hopefully, that answers your question. If there's still some
confusion, please let us know.
>
> Thanks,
> John Halley Gotway
> met_help at ucar.edu
>
>
> On 01/25/2012 03:03 AM, Mansi via RT wrote:
>>
>> Wed Jan 25 03:03:50 2012: Request 53003 was acted upon.
Transaction: Ticket created by mansibhowmick at gmail.com Queue: met_help
Subject: Regarding scale specification in wavelet-stat tool of MET
>> Owner: Nobody Requestors: mansibhowmick at gmail.com Status: new
Ticket<URL: https://rt.rap.ucar.edu/rt/Ticket/Display.html?id=53003>
>>
>>
>> Respected Sir/Madam,
>>
>> I wish to use wavelet stat tool for verification. I am using
tile_dim=64 so NSCALE is 6 now according to my understanding, ISCALE 0
to 5 (i.e. n=0 to n-1) is for mother wavelet component and ISCALE
>> = 6 is for father wavelet component (domain mean), then ISCALE=7
stands for what ? There are in total 8 categories of ISCALE (i.e. 0 to
7) are appearing on line type ___isc.txt . Kindly explain.
>>
>> Thanks and Regards Mansi Bhowmick Project Scientist India
Meteorological Department

------------------------------------------------
Subject: Regarding scale specification in wavelet-stat tool of MET
From: Mansi
Time: Thu Jan 26 23:08:09 2012

Respected Sir,

             Many thanks for the reply and explanation.
Perhaps I have misunderstood ISCALE=0 as 1st mother wavelet component
,which actually stands for thresholded binary field.

Thanks and regards
Mansi

On 25 January 2012 21:32, John Halley Gotway via RT
<met_help at ucar.edu>wrote:

> Mansi,
>
> Please disregard my comment about the documentation issue - where it
says
> 2^n x 2^n in section 7.2.1 of the documentation is correct.  I was
just
> confusing myself!
>
> Thanks,
> John
>
> On 01/25/2012 08:52 AM, John Halley Gotway wrote:
> > Mansi,
> >
> > I ran a test case of Wavelet-Stat using a tile dimension of 64
(which is
> 2 to the 6th power), and I attached some images of the output.
> >
> > Also below, I've listed an excerpt from section 7.2.1 of the MET
User's
> Guide:
> >
> >
>
-----------------------------------------------------------------------------
> >
> > The binary forecast and observation fields obtained from the
> thresholding are then decomposed into the sum of components on
different
> scales, by using a 2D Haar wavelet filter (Fig 7.2). Note that the
> > scale components are fields, and their sum adds up to the original
> binary field. For a forecast defined over square domain of 2^n grid-
points,
> the scale components are n+1: n mother wavelet
> > components + the largest father wavelet (or scale-function)
component.
> The n mother wavelet components have resolution equal to 1, 2, 4,
...
> 2^(n-1) grid-points. The largest father wavelet component
> > is a constant field over the 2^n grid-point domain with value
equal to
> the field mean.
> >
> >
>
-----------------------------------------------------------------------------
> >
> > In the case of 64 = 2^6, we have 6 mother wavelet components with
> resolutions equal to 2^0 = 1, 2^1 = 2, 2^2 = 4, 2^3 = 8, 2^4 = 16,
and 2^5
> = 32 grid points. And we have the largest father wavelet
> > which is constant over the 2^6 = 64 grid-point domain. So in the
> attached output images, you should see pictures for the raw field,
the
> thresholded binary field, 6 mother wavelets, and one father
> > wavelet.
> >
> > In the process of looking at this, I do see that our documentation
says
> "2^n x 2^n" in a few places when it should just say "2^n". Perhaps
that's a
> source of confusion. I'll make sure we correct that
> > for the next release.
> >
> > Hopefully, that answers your question. If there's still some
confusion,
> please let us know.
> >
> > Thanks,
> > John Halley Gotway
> > met_help at ucar.edu
> >
> >
> > On 01/25/2012 03:03 AM, Mansi via RT wrote:
> >>
> >> Wed Jan 25 03:03:50 2012: Request 53003 was acted upon.
Transaction:
> Ticket created by mansibhowmick at gmail.com Queue: met_help Subject:
> Regarding scale specification in wavelet-stat tool of MET
> >> Owner: Nobody Requestors: mansibhowmick at gmail.com Status: new
> Ticket<URL: https://rt.rap.ucar.edu/rt/Ticket/Display.html?id=53003>
> >>
> >>
> >> Respected Sir/Madam,
> >>
> >> I wish to use wavelet stat tool for verification. I am using
> tile_dim=64 so NSCALE is 6 now according to my understanding, ISCALE
0 to 5
> (i.e. n=0 to n-1) is for mother wavelet component and ISCALE
> >> = 6 is for father wavelet component (domain mean), then ISCALE=7
stands
> for what ? There are in total 8 categories of ISCALE (i.e. 0 to 7)
are
> appearing on line type ___isc.txt . Kindly explain.
> >>
> >> Thanks and Regards Mansi Bhowmick Project Scientist India
> Meteorological Department
>
>

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