[Met_help] [rt.rap.ucar.edu #47895] History for "Better" interpolation...

[Tressa Fowler via RT]

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

Hi,

I would like to ask for advice, if possible.

I run wrf model on italy domain.
Italy, as we know, has very complicated orography and there are many points
that are both near the sea or mountains (or both).
So, in your opinion, what could be the *"best**" interpolation* to use?
For example LS_FIT or DW_MEAN? How square? 2x2, 3x3 and so on ...?
Or it is better not to interpolate (W=1), and take the grid point closest to
the point of observation?

Thank you very much,
Fabio.


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

Subject: "Better" interpolation...
From: Tressa Fowler
Time: Wed Jun 29 10:41:42 2011

Hi Fabio,

In my opinion, with highly varying topography, it may be best not to
interpolate and just use nearest neighbor. However, it may depend
somewhat on your observation source and variable of interest. I would
probably try interpolating using distance weighting for a couple of
neighborhood sizes (2,3, 5?) and see how that affects your results for
a few cases. This may give you a better idea of the best way to
proceed for your whole set of data.

We would be interested to hear what you discover, if you don't mind
sharing with us. We like to be able to provide guidance to other users
based on experience, and we have no experience with verification over
Italy.

Thanks,

Tressa

On Tue Jun 28 04:35:44 2011, fabio.gervasi70 at gmail.com wrote:
> Hi,
>
> I would like to ask for advice, if possible.
>
> I run wrf model on italy domain.
> Italy, as we know, has very complicated orography and there are many
points
> that are both near the sea or mountains (or both).
> So, in your opinion, what could be the *"best**" interpolation* to
use?
> For example LS_FIT or DW_MEAN? How square? 2x2, 3x3 and so on ...?
> Or it is better not to interpolate (W=1), and take the grid point
closest to
> the point of observation?
>
> Thank you very much,
> Fabio.



------------------------------------------------
Subject: "Better" interpolation...
From: Fabio F.Gervasi
Time: Mon Jul 04 10:30:33 2011

Hi,
thank you for your suggests!

We will be happy to let you know about our simulations and our tests
with
MET.

Fabio.


2011/6/29 RAL HelpDesk {for Tressa Fowler} <met_help at ucar.edu>

> Hi Fabio,
>
> In my opinion, with highly varying topography, it may be best not to
> interpolate and just use nearest neighbor. However, it may depend
somewhat
> on your observation source and variable of interest. I would
probably try
> interpolating using distance weighting for a couple of neighborhood
sizes
> (2,3, 5?) and see how that affects your results for a few cases.
This may
> give you a better idea of the best way to proceed for your whole set
of
> data.
>
> We would be interested to hear what you discover, if you don't mind
sharing
> with us. We like to be able to provide guidance to other users based
on
> experience, and we have no experience with verification over Italy.
>
> Thanks,
>
> Tressa
>
> On Tue Jun 28 04:35:44 2011, fabio.gervasi70 at gmail.com wrote:
> > Hi,
> >
> > I would like to ask for advice, if possible.
> >
> > I run wrf model on italy domain.
> > Italy, as we know, has very complicated orography and there are
many
> points
> > that are both near the sea or mountains (or both).
> > So, in your opinion, what could be the *"best**" interpolation* to
use?
> > For example LS_FIT or DW_MEAN? How square? 2x2, 3x3 and so on ...?
> > Or it is better not to interpolate (W=1), and take the grid point
closest
> to
> > the point of observation?
> >
> > Thank you very much,
> > Fabio.
>
>
>
>

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