[Met_help] [rt.rap.ucar.edu #40149] History for MET and site based verifications
RAL HelpDesk {for John Halley Gotway}
met_help at ucar.edu
Thu Aug 19 09:10:41 MDT 2010
----------------------------------------------------------------
Initial Request
----------------------------------------------------------------
Dear MET support,
We would like to use the MET in the intended way with WRF, but would also like
to use it to verify site forecasts. My understanding of the documentation and
code is that this is not possible. Whilst observations can be on grids or
site based, model data has to be on an grid.
My questions are:
- are you planing to add support for the verification of site based forecasts?
- if not, do you see any reason why it would be a bad idea if we did take on
this task? We're quite happy to feed any additions that you consider useful
back into the main code.
- if we implement the desired support, would you be prepared to answer
questions about the code now and again?
- what are your plans for MET anyway? Is it still under active development?
Kind regards,
Frank
--
Frank Sonntag
Meteorologist, MetOcean Solutions Ltd
PO Box 441, New Plymouth, NZ
T: 64-7-825 0540, M: 64-21-868400
This e-mail, its contents and any attachments are confidential. If you
are not the intended recipient of this e-mail then you must not 1) copy,
distribute or disclose it or 2) take any action relying upon it.
If you are not the intended recipient of this e-mail then, please delete
it and any attachments and advise the sender by return. MetOcean
Solutions Ltd does not accept legal responsibility for the contents of
e-mails sent over a public network.
----------------------------------------------------------------
Complete Ticket History
----------------------------------------------------------------
Subject: Re: [rt.rap.ucar.edu #40149] MET and site based verifications
From: John Halley Gotway
Time: Wed Aug 18 11:24:33 2010
Frank,
You're correct that MET is set up to verify gridded NWP output, and it
is not set up to verify individual site forecasts. That being said,
there is a potential way to use MET to perform verification
for individual site forecasts.
The Point-Stat tool in MET is used to compare gridded forecasts to
point observations. One of the outputs from Point-Stat is a line type
called MPR, for matched pairs. Those lines contains 21
columns of header data followed by a forecast/observation pair. For
example, suppose Point-Stat has used 50 observations in computing some
statistics in verifying 2-meter temperature. It has the
ability to write out 50 lines of output, where each line contains a
forecast/observation pair.
Next, the Stat-Analysis tool has the ability to read in those MPR
lines from Point-Stat, filter them using whatever criteria the user
selects, and compute statistics for them.
Now, suppose you're in the situation where you have a bunch of site
forecasts and corresponding observation values. You could reformat
those matched pairs to look like that ASCII MPR lines output
from Point-Stat. And you could use them as input to the STAT-Analysis
tool to compute a bunch of statistics on them.
Alternatively, you may choose to read them into the R statistical
programming language and use the verify package within R to perform
analysis on them.
We are continuing development on MET and are planning to release
METv3.0 at the end of this month. It will include new tools for
verifying ensemble forecasts as well a handful of other utilities.
And we are open to contributions of methods and code from the user
community. One big issue I see with handling point forecasts is how
those point forecasts are formatted. I'm not aware if there's
any generally accepted standard format for site forecasts.
Hope that helps. Since you're a registered user of MET, you should
receive an email when the next release is available.
Thanks,
John Halley Gotway
met_help at ucar.edu
RAL HelpDesk {for Frank Sonntag} wrote:
> Tue Aug 17 23:20:26 2010: Request 40149 was acted upon.
> Transaction: Ticket created by f.sonntag at metocean.co.nz
> Queue: met_help
> Subject: MET and site based verifications
> Owner: Nobody
> Requestors: f.sonntag at metocean.co.nz
> Status: new
> Ticket <URL:
https://rt.rap.ucar.edu/rt/Ticket/Display.html?id=40149 >
>
>
> Dear MET support,
>
> We would like to use the MET in the intended way with WRF, but would
also like
> to use it to verify site forecasts. My understanding of the
documentation and
> code is that this is not possible. Whilst observations can be on
grids or
> site based, model data has to be on an grid.
> My questions are:
> - are you planing to add support for the verification of site based
forecasts?
> - if not, do you see any reason why it would be a bad idea if we did
take on
> this task? We're quite happy to feed any additions that you consider
useful
> back into the main code.
> - if we implement the desired support, would you be prepared to
answer
> questions about the code now and again?
> - what are your plans for MET anyway? Is it still under active
development?
>
> Kind regards,
> Frank
>
------------------------------------------------
Subject: Re: [rt.rap.ucar.edu #40149] MET and site based verifications
From: Frank Sonntag
Time: Thu Aug 19 04:36:44 2010
Hello John,
Many thanks for your quick and elaborate reply.
I will look into the work around that you suggested and see if it
suits us.
Good point re. the formatting of site data. If we were to add this as
a new
featue we would probably look at doing this in a way that makes it
easy t plug
in new formats. In any case, we would consult you to make sure we do
this in a
useful way.
Good to hear that MET is under active development. I am looking
forward to the
new release.
Cheers,
Frank
On Thu, 19 Aug 2010 05:24:33 RAL HelpDesk {for John Halley Gotway}
wrote:
> Frank,
>
> You're correct that MET is set up to verify gridded NWP output, and
it is
> not set up to verify individual site forecasts. That being said,
there is
> a potential way to use MET to perform verification for individual
site
> forecasts.
>
> The Point-Stat tool in MET is used to compare gridded forecasts to
point
> observations. One of the outputs from Point-Stat is a line type
called
> MPR, for matched pairs. Those lines contains 21 columns of header
data
> followed by a forecast/observation pair. For example, suppose
Point-Stat
> has used 50 observations in computing some statistics in verifying
2-meter
> temperature. It has the ability to write out 50 lines of output,
where
> each line contains a forecast/observation pair.
>
> Next, the Stat-Analysis tool has the ability to read in those MPR
lines
> from Point-Stat, filter them using whatever criteria the user
selects, and
> compute statistics for them.
>
> Now, suppose you're in the situation where you have a bunch of site
> forecasts and corresponding observation values. You could reformat
those
> matched pairs to look like that ASCII MPR lines output from Point-
Stat.
> And you could use them as input to the STAT-Analysis tool to
compute a
> bunch of statistics on them.
>
> Alternatively, you may choose to read them into the R statistical
> programming language and use the verify package within R to perform
> analysis on them.
>
> We are continuing development on MET and are planning to release
METv3.0 at
> the end of this month. It will include new tools for verifying
ensemble
> forecasts as well a handful of other utilities.
>
> And we are open to contributions of methods and code from the user
> community. One big issue I see with handling point forecasts is
how those
> point forecasts are formatted. I'm not aware if there's any
generally
> accepted standard format for site forecasts.
>
> Hope that helps. Since you're a registered user of MET, you should
receive
> an email when the next release is available.
>
> Thanks,
> John Halley Gotway
> met_help at ucar.edu
>
> RAL HelpDesk {for Frank Sonntag} wrote:
> > Tue Aug 17 23:20:26 2010: Request 40149 was acted upon.
> > Transaction: Ticket created by f.sonntag at metocean.co.nz
> > Queue: met_help
> > Subject: MET and site based verifications
> > Owner: Nobody
> > Requestors: f.sonntag at metocean.co.nz
> > Status: new
> > Ticket <URL:
https://rt.rap.ucar.edu/rt/Ticket/Display.html?id=40149 >
> >
> >
> > Dear MET support,
> >
> > We would like to use the MET in the intended way with WRF, but
would also
> > like to use it to verify site forecasts. My understanding of the
> > documentation and code is that this is not possible. Whilst
observations
> > can be on grids or site based, model data has to be on an grid.
> > My questions are:
> > - are you planing to add support for the verification of site
based
> > forecasts? - if not, do you see any reason why it would be a bad
idea if
> > we did take on this task? We're quite happy to feed any additions
that
> > you consider useful back into the main code.
> > - if we implement the desired support, would you be prepared to
answer
> > questions about the code now and again?
> > - what are your plans for MET anyway? Is it still under active
> > development?
> >
> > Kind regards,
> > Frank
>
------------------------------------------------
Subject: MET and site based verifications
From: John Halley Gotway
Time: Thu Aug 19 09:10:40 2010
Frank,
Great. I'll go ahead and close this MET-Help ticket.
Thanks,
John
------------------------------------------------
More information about the Met_help
mailing list