Re: objective analysis & topography

[Steve Hankin <hankin@xxxxxxxxxxxxx>]

Hi Keith,

You are exactly correct about the behavior of the current Ferret "GO
objective" script.

Since there are so many ways to approach the "gridding" question
(producing gridded fields from scattered data) it is a prime candidate
for Ferret "external functions". With this appraoch you can take a host
of different gridding techniques (from math libraries, from the
literature, from commercial software) and embed them into functions that
will appear as part of the Ferret "language". You can then compare and
contrast the results, to help in selecting the most appropriate
technique to use.

We are nearing completion of a template external functio for regridding
-- simply plug your FORTRAN or C routine-of-choice into this template
and compile/link. ((It includes a much superior gridding technique
supplied as an example.)

Use of this template, however, requires that latest alpha version of
Ferret (not distributed). Since an earlier version of the External
Functions framework is embeddd in Ferret V4.91, I'll have to check to
see if we have a version of the gridding EF that will work with that and
get back to you.

	Happy Ferreting - steve

===========================================================================

Keith Lindsay wrote:
> 
> Greetings,
> 
> I am contouring scattered data from a vertical cross section of the ocean.
> The basic approach that I can come up with is to first transfer the data
> to a regular grid with the GO tool objective and then to use contour on
> the result.
> 
> However, I do not think that this approach is topographically aware in the
> following sense. From my reading of the Ferret User's Guide, the objective
> analysis is doing some kind of splining/Laplacian interpolation. So a
> point from the scattered data set affects grid points which are "near" it.
> This effect is not aware of bottom topography and for instance could occur
> through an ocean ridge. I would like to avoid this.
> 
> Is my understanding of how objective works correct? If so, is there a way
> to incorporate topography into an objective analysis or is there another
> overall approach to producing a contour plot of scattered data?
> 
> I think the central question is "How can I perform an objective analysis
> on an irregular geometry?"
> 
> ***********************************************************************
> Keith Lindsay                              phone: 303-497-1722
> Climate and Global Dynamics                fax:   303-497-1700
> National Center for Atmospheric Research   email: klindsay@cgd.ucar.edu
> P.O. Box 3000
> Boulder, CO  80307-3000

-- 
Steve Hankin
NOAA/PMEL, 7600 Sand Point Way NE, Seattle, WA 98115-0070
ph. (206) 526-6080 -- FAX (206) 526-6744