pdef and interpolation question

[Jennifer Adams]

Hi, Ron --
That is exactly why there are precipitation types in GFS meteograms,
but not the NAM meteograms (on wxmaps.org) -- because of the
interpolation of the index-value fields when using PDEF with the NAM
data. Same thing happens with the land mask; you have land, ocean,
and something in between around the coastlines. You can try using
maskout and const to create a new field with values equal to 0 or 1
to mimic the original, but you have to make a subjective choice about
what threshold to use with maskout. With a land mask, it's reasonably
easy:
   ga-> d const(maskout(land,land-0.5),1)
But with precip types ... it's a little more dicey.

--Jennifer

On Feb 20, 2008, at 1:37 PM, Goodson,Ron [Edm] wrote:

> Hi
>
> Perhaps a silly question.
>
> I have some "indexed" data on a polarstereographic grid.  Here is an
> example for, lets say, precipitation type:
>
> 1 = rain
> 2 = snow
> 3 = freezing rain
> 4 = drizzle
> 5 = mixed rain-snow
>
> etc.
>
> Since this is on a PS grid, when it comes into GrADS all values are
> interpolated.  And, for other reasons, I'm interpolating onto a finer
> grid than what is in pdef
>
> As a result, when the original grid data might be a 1 next to a 5 ..
> GrADS interpolates this to   1 3 5 (if I use 2x finer
> resolution)... so
> I end up with a not-real "freezing rain" between the rain and mixed
> rain-snow.
>
> Is there some sneaky way of telling GrADS to use nearest-neighbour to
> interpolate from the original data -- or some other solution I haven't
> thought of.  I had thought of regrid (or interp) .. but that might not
> work if it happens to grab the new-interpolated value (eg -- the
> "3" in
> the example above.
>
> any thoughts?
>
> tks
>
> ron

--
Jennifer M. Adams
IGES/COLA
4041 Powder Mill Road, Suite 302
Calverton, MD 20705
jma at cola.iges.org



-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://gradsusr.org/pipermail/gradsusr/attachments/20080220/f316528c/attachment.html