Re: SCAT2GRIDGAUSS_YZ

[Billy Kessler <William.S.Kessler@xxxxxxxx>]

Hi Sophie -

First of all, it's probably not a good idea to use the scat2grid algorithm in depth to make a 
section (y-z or x-z). Picture such a section with scattered data like this:

  y->
z    * *    *     *
|    *      * 
v    *            *
     * *    *     *
     *      *     *
     * *    *     *   
     
Consider the second profile, which has a long interval with no values. The scat2grid (either gauss 
or laplace) algorithm is going to use values from the neighboring profiles to fill those points, 
which could easily result in large and unphysical inversions. This would be more obvious with 
temperature than with salinity, but for example consider the situation spanning the equator in the 
Pacific, where there is an intrusion of high-salinity water from the south at thermocline level. If 
you have missing points in a profile at, say, 2N, scat2grid will be influenced by southern values 
and portray the tongue extending much too far north. A much more physically reasonable way to do 
this would be to define a common z-axis (every 5m or it might be stretched). Then interpolate the 
salinity for each profile individually using z=@fln. Then define the regular y-axis that you want, 
and use y=@fln on the filled profiles.

The best use of scat2grid is where interpolation in both directions makes sense, and I don't think 
mixing horizontal and vertical directions in the ocean does. I have used scat2gridgauss many times 
on 3-d ocean data, but only in x-y. I first linearly interpolated each profile to be complete and 
regular in z, then used scat2gridgauss_xy at *each depth separately* to complete the gridding 
process.

Billy K

> Subject: SCAT2GRIDGAUSS_YZ
> From: Sophie Cravatte <Sophie.Cravatte@notos.cst.cnes.fr>
> To: ferret_users@noaa.gov
> Date: Wed, 01 Jun 2005 20:23:06 +0200
> 
> Dear Ferret users,
> 
> I can't figure out how to use the function SCAT2GRIDGAUSS_YZ to grid
> scattered data to a YZ grid.
> Here is my script:
> 
> define axis/y=20S:5S:1/units=degrees yaxs
> define axis/z=0:500:1/units=meters/depth zaxs
> SET DATA/EZ/var="lat,depth,temp,salt,o1,o2,dens" r165e.dat
> 
> let salinite
> =SCAT2GRIDGAUSS_YZ(lat,depth,salt,y[gy=yaxs],z[gz=zaxs],2,2,2,2)
> 
> And instead of having a 2-D YZ field for salinite, I got a 3-D i,Y,Z
> field, and of course with stupid values.
> What's wrong in my script? What didn't I understand?
> 
> Thanks a lot for your help,
> 
> Sophie
> 
> 
> -- 
> Sophie Cravatte
> LEGOS--14 av. Ed. Belin--31400 TOULOUSE
> tel: 05 61 33 30 05
>