Re: limit on coordinate?

[hankin@xxxxxxxxxxxxxxxxxxx (Steve Hankin)]

Hi Bertrand,

With version 4.4x of Ferret (and earlier) Ferret's ability to grid irrigular
data is, indeed, limited. You could play heuristic games like gridding one half
of the data into one grid and the other half into another grid and then
averaging the two -- the result could give you something close to the correct
field though not valid from a statistical point of view.

Version 5 of Ferret will have a full-fledged facility to define objective
analysis and gridding procedures. That is a few months away (too soon to give a
firm date). Meantime, I recommend that if your gridding needs are exacting you
use code external to Ferret to create gridded fields for Ferret to work with.

One code that comes to mind is a public domain utility from Unidata called
"ncbarne" (a NetCDF utility). Have a look at http://www.unidata.ucar.edu/ to
find that. Other users may be able to recommend further alternatives.

	- steve

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

On Aug 21,  5:08pm, Bertrand REPTIN wrote:
> Subject: Re: limit on coordinate?
> Steve Hankin wrote:
> >
> > Hi Dongxiao,
> >
> > Yes, Ferret does have a limit of 50,000 coordinate points but it need not
get
> > in your way. It is only really a limit if you have 50K **unevenly spaced**
> > coordinate points -- this is rarely the case.
> >
>
>
> Hello Steve,
>
> Unfortunately, I read an ascii file containing 73403 values of water
> depth with irregular and **unevenly spaced** values of latitude and
> longitude, in the form :
>
> lat	lon	depth
> 46.6396  -4.9982 1348.0   ! value 1
> ...
> 45.5245  -3.1742  149.0	! value 73403
>
>
> I read them as follows :
>
> yes? define axis/t=1:73403:1 taxis
> yes? define grid/t=taxis tgrid
> yes? file/var="lat,lon,prof"/grid=tgrid bathyzoe1
> yes? plot/vs lon,lat
> yes? sh data
>      currently SET data sets:
>     1> ./bathyzoe1  (default)
>  name     title                             I         J
> K         L
>  LAT      LAT                              1:1       1:1       1:1
> 1:73403
>  LON      LON                              1:1       1:1       1:1
> 1:73403
>  PROF     PROF                             1:1       1:1       1:1
> 1:73403
>
> This plotting was OK, although the number of points was too big !
>
> but not the following one, to transform data in a regular grid :
>
> yes? go objective lon lat prof "-5:-3:0.01" "45:47:0.01" 0 1
> Your result will be variable "gridded" in data set fer_objective.unf.
>  **ERROR: a program limit has been reached: too many input points -
> limit 50000
> USER/COMMAND=OBJECTIVE/opt1=-5:-3:0.01,45:47:0.01,0,1/file=fer_objective.unf
> lon/10000,lat/10000,prof
> Command file, command group, or REPEAT execution aborted
>
> This last command is OK if I use file/skip=30000 to have less than 50000
> values
>
> How can I easily transform the 73403 irregular (x,y,z) values to a 2D
> regular grid ?
>
> Thanks
>
> Bertrand
>
>
> *******************************************************************************
> Bertrand REPTIN
> STSN/GESMA - BP42 - BREST naval - FRANCE
> Tel. 98-22-53-67
> Fax. 98-22-72-13
> e-mail :  reptin@gesma.fr
> *******************************************************************************
>-- End of excerpt from Bertrand REPTIN




-- 

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