[Met_help] [rt.rap.ucar.edu #77581] History for MET 5.2 release now has “grid_weight_flag” configuration option
John Halley Gotway via RT
met_help at ucar.edu
Fri Aug 19 13:53:00 MDT 2016
----------------------------------------------------------------
Initial Request
----------------------------------------------------------------
Hello,
I have been a MET 5.1 user and I saw that this new 5.2 release has the
capability to weight by grid cell area or cosine of latitude for
grid-to-grid comparison. I had previously assumed this was taken care in
the grid-stat and point-stat tools statistical output (particularly
ANOM_CORR and RMSE). Was this not done in either grid-stat or point-stat
before? If not, do you foresee enabling this for point-stat (grid-to-point
comparison) in the future?
Thank you,
--
Shannon Rees
Programming Scientist
Engility
Geophysical Fluid Dynamics Lab
201 Forrestal Rd Princeton, NJ
609-452-5384
----------------------------------------------------------------
Complete Ticket History
----------------------------------------------------------------
Subject: MET 5.2 release now has “grid_weight_flag” configuration option
From: John Halley Gotway
Time: Tue Aug 16 10:00:52 2016
Shannon,
Great question. We intentionally did not add support for grid box
area
weighting to the Point-Stat tool. This really only makes sense when
doing
grid-to-grid comparisons in Grid-Stat and Ensemble-Stat. The whole
idea
behind area weighting is that we're combining data from different grid
boxes, each of which may actually represent a different area on the
grid.
In Grid-Stat, the sampling of those grid boxes is determined by the
definition of the grid itself. Applying grid area weighting,
especially
for global lat/lon grids, helps account for the grid distortion.
In Point-Stat, things are different. The sampling of data is not
determined by the definition of the grid. Instead, it's determined by
the
distribution of the observation locations being used for verification.
So
the description above doesn't really apply.
It would be very straightforward to add grid area weighting to Point-
Stat,
but we don't believe there is a good reason to do so.
If you feel strongly otherwise, please do let us know.
Thanks,
John Halley Gotway
On Tue, Aug 16, 2016 at 9:43 AM, Shannon Rees - NOAA Affiliate via RT
<
met_help at ucar.edu> wrote:
>
> Tue Aug 16 09:43:48 2016: Request 77581 was acted upon.
> Transaction: Ticket created by shannon.rees at noaa.gov
> Queue: met_help
> Subject: MET 5.2 release now has “grid_weight_flag”
configuration
> option
> Owner: Nobody
> Requestors: shannon.rees at noaa.gov
> Status: new
> Ticket <URL:
https://rt.rap.ucar.edu/rt/Ticket/Display.html?id=77581 >
>
>
> Hello,
>
> I have been a MET 5.1 user and I saw that this new 5.2 release has
the
> capability to weight by grid cell area or cosine of latitude for
> grid-to-grid comparison. I had previously assumed this was taken
care in
> the grid-stat and point-stat tools statistical output (particularly
> ANOM_CORR and RMSE). Was this not done in either grid-stat or
point-stat
> before? If not, do you foresee enabling this for point-stat (grid-
to-point
> comparison) in the future?
>
> Thank you,
> --
> Shannon Rees
> Programming Scientist
> Engility
> Geophysical Fluid Dynamics Lab
> 201 Forrestal Rd Princeton, NJ
> 609-452-5384
>
>
------------------------------------------------
Subject: MET 5.2 release now has “grid_weight_flag” configuration option
From: Shannon Rees - NOAA Affiliate
Time: Tue Aug 16 10:11:45 2016
Hi John,
Thanks for the quick reply! That does make sense. I have been trying
to
figure out why the grid-stat results do not look identical to results
from
other analysis tools when being compared to the same gridded dataset.
The
other analysis tool does do the grid cell weighting by area. Do you
think
this could explain a significant difference?
On Tue, Aug 16, 2016 at 12:00 PM, John Halley Gotway via RT <
met_help at ucar.edu> wrote:
> Shannon,
>
> Great question. We intentionally did not add support for grid box
area
> weighting to the Point-Stat tool. This really only makes sense when
doing
> grid-to-grid comparisons in Grid-Stat and Ensemble-Stat. The whole
idea
> behind area weighting is that we're combining data from different
grid
> boxes, each of which may actually represent a different area on the
grid.
> In Grid-Stat, the sampling of those grid boxes is determined by the
> definition of the grid itself. Applying grid area weighting,
especially
> for global lat/lon grids, helps account for the grid distortion.
>
> In Point-Stat, things are different. The sampling of data is not
> determined by the definition of the grid. Instead, it's determined
by the
> distribution of the observation locations being used for
verification. So
> the description above doesn't really apply.
>
> It would be very straightforward to add grid area weighting to
Point-Stat,
> but we don't believe there is a good reason to do so.
>
> If you feel strongly otherwise, please do let us know.
>
> Thanks,
> John Halley Gotway
>
> On Tue, Aug 16, 2016 at 9:43 AM, Shannon Rees - NOAA Affiliate via
RT <
> met_help at ucar.edu> wrote:
>
> >
> > Tue Aug 16 09:43:48 2016: Request 77581 was acted upon.
> > Transaction: Ticket created by shannon.rees at noaa.gov
> > Queue: met_help
> > Subject: MET 5.2 release now has “grid_weight_flag”
configuration
> > option
> > Owner: Nobody
> > Requestors: shannon.rees at noaa.gov
> > Status: new
> > Ticket <URL:
https://rt.rap.ucar.edu/rt/Ticket/Display.html?id=77581 >
> >
> >
> > Hello,
> >
> > I have been a MET 5.1 user and I saw that this new 5.2 release has
the
> > capability to weight by grid cell area or cosine of latitude for
> > grid-to-grid comparison. I had previously assumed this was taken
care in
> > the grid-stat and point-stat tools statistical output
(particularly
> > ANOM_CORR and RMSE). Was this not done in either grid-stat or
point-stat
> > before? If not, do you foresee enabling this for point-stat
> (grid-to-point
> > comparison) in the future?
> >
> > Thank you,
> > --
> > Shannon Rees
> > Programming Scientist
> > Engility
> > Geophysical Fluid Dynamics Lab
> > 201 Forrestal Rd Princeton, NJ
> > 609-452-5384
> >
> >
>
>
--
Shannon Rees
Programming Scientist
Engility
Geophysical Fluid Dynamics Lab
201 Forrestal Rd Princeton, NJ
609-452-5384
------------------------------------------------
Subject: MET 5.2 release now has “grid_weight_flag” configuration option
From: John Halley Gotway
Time: Tue Aug 16 12:01:58 2016
Shannon,
Yes, the application of grid weighting would definitely cause a
difference... especially for a global lat/lon grid where the grid box
area
differences are significant. For a regional lambert conformal grid,
the
differences would be much smaller.
Can you tell me more specifics about the analysis tools against which
you're comparing?
Thanks,
John
On Tue, Aug 16, 2016 at 10:11 AM, Shannon Rees - NOAA Affiliate via RT
<
met_help at ucar.edu> wrote:
>
> <URL: https://rt.rap.ucar.edu/rt/Ticket/Display.html?id=77581 >
>
> Hi John,
>
> Thanks for the quick reply! That does make sense. I have been
trying to
> figure out why the grid-stat results do not look identical to
results from
> other analysis tools when being compared to the same gridded
dataset. The
> other analysis tool does do the grid cell weighting by area. Do you
think
> this could explain a significant difference?
>
> On Tue, Aug 16, 2016 at 12:00 PM, John Halley Gotway via RT <
> met_help at ucar.edu> wrote:
>
> > Shannon,
> >
> > Great question. We intentionally did not add support for grid box
area
> > weighting to the Point-Stat tool. This really only makes sense
when
> doing
> > grid-to-grid comparisons in Grid-Stat and Ensemble-Stat. The
whole idea
> > behind area weighting is that we're combining data from different
grid
> > boxes, each of which may actually represent a different area on
the grid.
> > In Grid-Stat, the sampling of those grid boxes is determined by
the
> > definition of the grid itself. Applying grid area weighting,
especially
> > for global lat/lon grids, helps account for the grid distortion.
> >
> > In Point-Stat, things are different. The sampling of data is not
> > determined by the definition of the grid. Instead, it's
determined by
> the
> > distribution of the observation locations being used for
verification.
> So
> > the description above doesn't really apply.
> >
> > It would be very straightforward to add grid area weighting to
> Point-Stat,
> > but we don't believe there is a good reason to do so.
> >
> > If you feel strongly otherwise, please do let us know.
> >
> > Thanks,
> > John Halley Gotway
> >
> > On Tue, Aug 16, 2016 at 9:43 AM, Shannon Rees - NOAA Affiliate via
RT <
> > met_help at ucar.edu> wrote:
> >
> > >
> > > Tue Aug 16 09:43:48 2016: Request 77581 was acted upon.
> > > Transaction: Ticket created by shannon.rees at noaa.gov
> > > Queue: met_help
> > > Subject: MET 5.2 release now has “grid_weight_flag”
configuration
> > > option
> > > Owner: Nobody
> > > Requestors: shannon.rees at noaa.gov
> > > Status: new
> > > Ticket <URL:
https://rt.rap.ucar.edu/rt/Ticket/Display.html?id=77581
> >
> > >
> > >
> > > Hello,
> > >
> > > I have been a MET 5.1 user and I saw that this new 5.2 release
has the
> > > capability to weight by grid cell area or cosine of latitude for
> > > grid-to-grid comparison. I had previously assumed this was
taken care
> in
> > > the grid-stat and point-stat tools statistical output
(particularly
> > > ANOM_CORR and RMSE). Was this not done in either grid-stat or
> point-stat
> > > before? If not, do you foresee enabling this for point-stat
> > (grid-to-point
> > > comparison) in the future?
> > >
> > > Thank you,
> > > --
> > > Shannon Rees
> > > Programming Scientist
> > > Engility
> > > Geophysical Fluid Dynamics Lab
> > > 201 Forrestal Rd Princeton, NJ
> > > 609-452-5384
> > >
> > >
> >
> >
>
>
> --
> Shannon Rees
> Programming Scientist
> Engility
> Geophysical Fluid Dynamics Lab
> 201 Forrestal Rd Princeton, NJ
> 609-452-5384
>
>
------------------------------------------------
Subject: MET 5.2 release now has “grid_weight_flag” configuration option
From: Shannon Rees - NOAA Affiliate
Time: Fri Aug 19 13:50:14 2016
John,
I was confused about this before, but I understand now. We have no
need
for area weighting with pointstat. I do look forward to trying out
the
weighting with gridstat in the newest MET version to see if that give
the
results we are hoping for with global statistics.
The other analysis tool is just a simple in-house one, but it does not
use
point observations, just gridded reanalysis.
Thanks for your help.
Shannon
On Tue, Aug 16, 2016 at 2:01 PM, John Halley Gotway via RT <
met_help at ucar.edu> wrote:
> Shannon,
>
> Yes, the application of grid weighting would definitely cause a
> difference... especially for a global lat/lon grid where the grid
box area
> differences are significant. For a regional lambert conformal grid,
the
> differences would be much smaller.
>
> Can you tell me more specifics about the analysis tools against
which
> you're comparing?
>
> Thanks,
> John
>
>
>
> On Tue, Aug 16, 2016 at 10:11 AM, Shannon Rees - NOAA Affiliate via
RT <
> met_help at ucar.edu> wrote:
>
> >
> > <URL: https://rt.rap.ucar.edu/rt/Ticket/Display.html?id=77581 >
> >
> > Hi John,
> >
> > Thanks for the quick reply! That does make sense. I have been
trying to
> > figure out why the grid-stat results do not look identical to
results
> from
> > other analysis tools when being compared to the same gridded
dataset.
> The
> > other analysis tool does do the grid cell weighting by area. Do
you
> think
> > this could explain a significant difference?
> >
> > On Tue, Aug 16, 2016 at 12:00 PM, John Halley Gotway via RT <
> > met_help at ucar.edu> wrote:
> >
> > > Shannon,
> > >
> > > Great question. We intentionally did not add support for grid
box area
> > > weighting to the Point-Stat tool. This really only makes sense
when
> > doing
> > > grid-to-grid comparisons in Grid-Stat and Ensemble-Stat. The
whole
> idea
> > > behind area weighting is that we're combining data from
different grid
> > > boxes, each of which may actually represent a different area on
the
> grid.
> > > In Grid-Stat, the sampling of those grid boxes is determined by
the
> > > definition of the grid itself. Applying grid area weighting,
> especially
> > > for global lat/lon grids, helps account for the grid distortion.
> > >
> > > In Point-Stat, things are different. The sampling of data is
not
> > > determined by the definition of the grid. Instead, it's
determined by
> > the
> > > distribution of the observation locations being used for
verification.
> > So
> > > the description above doesn't really apply.
> > >
> > > It would be very straightforward to add grid area weighting to
> > Point-Stat,
> > > but we don't believe there is a good reason to do so.
> > >
> > > If you feel strongly otherwise, please do let us know.
> > >
> > > Thanks,
> > > John Halley Gotway
> > >
> > > On Tue, Aug 16, 2016 at 9:43 AM, Shannon Rees - NOAA Affiliate
via RT <
> > > met_help at ucar.edu> wrote:
> > >
> > > >
> > > > Tue Aug 16 09:43:48 2016: Request 77581 was acted upon.
> > > > Transaction: Ticket created by shannon.rees at noaa.gov
> > > > Queue: met_help
> > > > Subject: MET 5.2 release now has “grid_weight_flag”
> configuration
> > > > option
> > > > Owner: Nobody
> > > > Requestors: shannon.rees at noaa.gov
> > > > Status: new
> > > > Ticket <URL: https://rt.rap.ucar.edu/rt/Tic
> ket/Display.html?id=77581
> > >
> > > >
> > > >
> > > > Hello,
> > > >
> > > > I have been a MET 5.1 user and I saw that this new 5.2 release
has
> the
> > > > capability to weight by grid cell area or cosine of latitude
for
> > > > grid-to-grid comparison. I had previously assumed this was
taken
> care
> > in
> > > > the grid-stat and point-stat tools statistical output
(particularly
> > > > ANOM_CORR and RMSE). Was this not done in either grid-stat or
> > point-stat
> > > > before? If not, do you foresee enabling this for point-stat
> > > (grid-to-point
> > > > comparison) in the future?
> > > >
> > > > Thank you,
> > > > --
> > > > Shannon Rees
> > > > Programming Scientist
> > > > Engility
> > > > Geophysical Fluid Dynamics Lab
> > > > 201 Forrestal Rd Princeton, NJ
> > > > 609-452-5384
> > > >
> > > >
> > >
> > >
> >
> >
> > --
> > Shannon Rees
> > Programming Scientist
> > Engility
> > Geophysical Fluid Dynamics Lab
> > 201 Forrestal Rd Princeton, NJ
> > 609-452-5384
> >
> >
>
>
--
Shannon Rees
Programming Scientist
Engility
Geophysical Fluid Dynamics Lab
201 Forrestal Rd Princeton, NJ
609-452-5384
------------------------------------------------
Subject: MET 5.2 release now has “grid_weight_flag” configuration option
From: John Halley Gotway
Time: Fri Aug 19 13:52:47 2016
Shannon,
OK, thanks for getting back to me on that. I'll go ahead and resolve
this
ticket, but feel free to write again if more issues/questions arise.
Thanks,
John
On Fri, Aug 19, 2016 at 1:50 PM, Shannon Rees - NOAA Affiliate via RT
<
met_help at ucar.edu> wrote:
>
> <URL: https://rt.rap.ucar.edu/rt/Ticket/Display.html?id=77581 >
>
> John,
>
> I was confused about this before, but I understand now. We have no
need
> for area weighting with pointstat. I do look forward to trying out
the
> weighting with gridstat in the newest MET version to see if that
give the
> results we are hoping for with global statistics.
>
> The other analysis tool is just a simple in-house one, but it does
not use
> point observations, just gridded reanalysis.
>
> Thanks for your help.
> Shannon
>
>
> On Tue, Aug 16, 2016 at 2:01 PM, John Halley Gotway via RT <
> met_help at ucar.edu> wrote:
>
> > Shannon,
> >
> > Yes, the application of grid weighting would definitely cause a
> > difference... especially for a global lat/lon grid where the grid
box
> area
> > differences are significant. For a regional lambert conformal
grid, the
> > differences would be much smaller.
> >
> > Can you tell me more specifics about the analysis tools against
which
> > you're comparing?
> >
> > Thanks,
> > John
> >
> >
> >
> > On Tue, Aug 16, 2016 at 10:11 AM, Shannon Rees - NOAA Affiliate
via RT <
> > met_help at ucar.edu> wrote:
> >
> > >
> > > <URL: https://rt.rap.ucar.edu/rt/Ticket/Display.html?id=77581 >
> > >
> > > Hi John,
> > >
> > > Thanks for the quick reply! That does make sense. I have been
trying
> to
> > > figure out why the grid-stat results do not look identical to
results
> > from
> > > other analysis tools when being compared to the same gridded
dataset.
> > The
> > > other analysis tool does do the grid cell weighting by area. Do
you
> > think
> > > this could explain a significant difference?
> > >
> > > On Tue, Aug 16, 2016 at 12:00 PM, John Halley Gotway via RT <
> > > met_help at ucar.edu> wrote:
> > >
> > > > Shannon,
> > > >
> > > > Great question. We intentionally did not add support for grid
box
> area
> > > > weighting to the Point-Stat tool. This really only makes
sense when
> > > doing
> > > > grid-to-grid comparisons in Grid-Stat and Ensemble-Stat. The
whole
> > idea
> > > > behind area weighting is that we're combining data from
different
> grid
> > > > boxes, each of which may actually represent a different area
on the
> > grid.
> > > > In Grid-Stat, the sampling of those grid boxes is determined
by the
> > > > definition of the grid itself. Applying grid area weighting,
> > especially
> > > > for global lat/lon grids, helps account for the grid
distortion.
> > > >
> > > > In Point-Stat, things are different. The sampling of data is
not
> > > > determined by the definition of the grid. Instead, it's
determined
> by
> > > the
> > > > distribution of the observation locations being used for
> verification.
> > > So
> > > > the description above doesn't really apply.
> > > >
> > > > It would be very straightforward to add grid area weighting to
> > > Point-Stat,
> > > > but we don't believe there is a good reason to do so.
> > > >
> > > > If you feel strongly otherwise, please do let us know.
> > > >
> > > > Thanks,
> > > > John Halley Gotway
> > > >
> > > > On Tue, Aug 16, 2016 at 9:43 AM, Shannon Rees - NOAA Affiliate
via
> RT <
> > > > met_help at ucar.edu> wrote:
> > > >
> > > > >
> > > > > Tue Aug 16 09:43:48 2016: Request 77581 was acted upon.
> > > > > Transaction: Ticket created by shannon.rees at noaa.gov
> > > > > Queue: met_help
> > > > > Subject: MET 5.2 release now has “grid_weight_flag”
> > configuration
> > > > > option
> > > > > Owner: Nobody
> > > > > Requestors: shannon.rees at noaa.gov
> > > > > Status: new
> > > > > Ticket <URL: https://rt.rap.ucar.edu/rt/Tic
> > ket/Display.html?id=77581
> > > >
> > > > >
> > > > >
> > > > > Hello,
> > > > >
> > > > > I have been a MET 5.1 user and I saw that this new 5.2
release has
> > the
> > > > > capability to weight by grid cell area or cosine of latitude
for
> > > > > grid-to-grid comparison. I had previously assumed this was
taken
> > care
> > > in
> > > > > the grid-stat and point-stat tools statistical output
(particularly
> > > > > ANOM_CORR and RMSE). Was this not done in either grid-stat
or
> > > point-stat
> > > > > before? If not, do you foresee enabling this for point-stat
> > > > (grid-to-point
> > > > > comparison) in the future?
> > > > >
> > > > > Thank you,
> > > > > --
> > > > > Shannon Rees
> > > > > Programming Scientist
> > > > > Engility
> > > > > Geophysical Fluid Dynamics Lab
> > > > > 201 Forrestal Rd Princeton, NJ
> > > > > 609-452-5384
> > > > >
> > > > >
> > > >
> > > >
> > >
> > >
> > > --
> > > Shannon Rees
> > > Programming Scientist
> > > Engility
> > > Geophysical Fluid Dynamics Lab
> > > 201 Forrestal Rd Princeton, NJ
> > > 609-452-5384
> > >
> > >
> >
> >
>
>
> --
> Shannon Rees
> Programming Scientist
> Engility
> Geophysical Fluid Dynamics Lab
> 201 Forrestal Rd Princeton, NJ
> 609-452-5384
>
>
------------------------------------------------
More information about the Met_help
mailing list