[Met_help] [rt.rap.ucar.edu #39486] History for MET tool performance

[RAL HelpDesk {for Tressa Fowler}]

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

To whom it may concern:

I run WRF 3.2 with analysis nudging with the data from NCEP, ds353.4 and
ds464.0. I am trying to do model performance (compare model with
observational data) with METv2.0. In that case, do you think I still can use
these data to do model performance (compare WRF output with ds353.4 and
ds464.0)?

Thanks

Yang


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

Subject: MET tool performance
From: Tressa Fowler
Time: Fri Jul 23 11:45:31 2010

Hi Yang,

If you are using certain observations as input into your model, you
should not use the same observations to verify the model. The results
of your verification will not tell you about the quality of the model,
since the forecast is not independent of the observations.

However, if you are initializing your model with upper air data from a
certain time (say 0Z), then running your model forward (say to 18Z),
you can use upper air data from 18Z to verify your 18 hour forecast.
So, the data is the same type, but not the same observations you used.
Does this make sense?

Please let us know if you have further questions.

Thanks,

Tressa

On Wed Jul 21 15:15:11 2010, gaoyang83 at gmail.com wrote:
> To whom it may concern:
>
> I run WRF 3.2 with analysis nudging with the data from NCEP, ds353.4
and
> ds464.0. I am trying to do model performance (compare model with
> observational data) with METv2.0. In that case, do you think I still
can use
> these data to do model performance (compare WRF output with ds353.4
and
> ds464.0)?
>
> Thanks
>
> Yang



------------------------------------------------
Subject: MET tool performance
From: Yang Gao
Time: Fri Jul 23 13:51:09 2010

Hi Tressa

 Thanks for your reply. I see now.

  Yang

On Fri, Jul 23, 2010 at 1:45 PM, RAL HelpDesk {for Tressa Fowler} <
met_help at ucar.edu> wrote:

> Hi Yang,
>
> If you are using certain observations as input into your model, you
should
> not use the same observations to verify the model. The results of
your
> verification will not tell you about the quality of the model, since
the
> forecast is not independent of the observations.
>
> However, if you are initializing your model with upper air data from
a
> certain time (say 0Z), then running your model forward (say to 18Z),
you can
> use upper air data from 18Z to verify your 18 hour forecast. So, the
data is
> the same type, but not the same observations you used. Does this
make sense?
>
> Please let us know if you have further questions.
>
> Thanks,
>
> Tressa
>
> On Wed Jul 21 15:15:11 2010, gaoyang83 at gmail.com wrote:
> > To whom it may concern:
> >
> > I run WRF 3.2 with analysis nudging with the data from NCEP,
ds353.4 and
> > ds464.0. I am trying to do model performance (compare model with
> > observational data) with METv2.0. In that case, do you think I
still can
> use
> > these data to do model performance (compare WRF output with
ds353.4 and
> > ds464.0)?
> >
> > Thanks
> >
> > Yang
>
>
>
>


--
************************************************
Yang Gao
Dept. Civil & Environmental Engr.
University of Tennessee, Knoxville
Tel: (865) 924-0621
************************************************

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