Re: cell edges for 3-argument shade

[Steve Hankin <hankin@xxxxxxxxxxxxx>]

Hi Keith,

I'm glad you asked this question, because it points us to an omission of this
topic in the documentation (which we will fix - thx).

In the 3-argument syntax you specify
    yes? SHADE values, xcoords, ycoords

Say
    nVx - is the size of the "x" dimension of the values argument
    nCx - is the size of the "x" dimension of the coordinate arguments

If nCx = nVx then the xcoords argument is presumed to give the locations of
the points in the values argument and (as you say), the boundaries between
points are computed to be the midpoints.

However, if nCx = nVx + 1 then the xcoords argument is presumed to give the
locations of the *boundaries* -- exactly the type of control that I think you
are looking for.

Here's a simple example (plot attached):

     yes? set view upper
     yes? shade i[i=1:5]+j[j=1:5],i[i=1:5]+0*j[j=1:5],0*i[i=1:5]+j[j=1:5]

     yes? set view lower
     yes? shade i[i=1:5]+j[j=1:5],i[i=1:6]+0*j[j=1:5],0*i[i=1:6]+j[j=1:5]

You'll note that in the upper panel that X cell boundaries are at 1.5, 2.5,
... (the midpoints of the xcoord argument), whereas in the lower panel they
are at x=1, 2, 3, ... (the values of the xcoord argument.)

    - steve

P.S. Make sure that the size of the xcoord argument always matches the size of
the ycoord argument.

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

Keith Lindsay wrote:

> Greetings,
>
> I have 2 questions. The short versions are :
>
> 1) How does ferret generate boundaries for the 3-argument shade command?
>
> 2) Can these boundaries instead be user specified?
>
> When a 3-argument shade command is issued, say for data on a curvilinear
> grid, it appears that ferret is internally generating the boundaries or
> edges of the 'cells'. How are these cell edges are generated? I cannot
> find it documented in the Users Guide. I'm guessing it is a generalization
> of what ferret does for rectilinear grids, which is to use midpoints of
> the corresponding dimension variables, which leads to my next question.
>
> For rectilinear grids, the 'midpoint technique' can be overridden with the
> use of dimension edges. Is there a generalization of this that allows a
> user to specify the edges for the 3-argument shade command?
>
> Thanks,
> Keith
>
> ************************************************************************
> Keith Lindsay         http://www.cgd.ucar.edu/oce/klindsay/klindsay.html
> email: klindsay@cgd.ucar.edu    phone: 303-497-1722    fax: 303-497-1700

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