Re: [ferret_users] array subset assignment after smoothing

[Ansley Manke <ansley.b.manke@xxxxxxxx>]

In a slight variation of this, an expression like

   let hsmo = if hsmo_a then hsmo_a else h

is often better replaced by

   let hsmo = MISSING(hsmo_a, h)

Which has the advantage that if there are values of 0 in hsmo_a, the 
first expression will not retain them as valid data (because 0 is false 
in the IF statement!)  but the expression using MISSING will keep zero 
as a valid value. A simple example:

  yes? ! Define a couple of variables
  yes? let aa = IF (i-j) LT 4 then (i-j)
  yes? let bb = i+j

  yes? ! aa is 0 along the diagonal, missing in one corner
  yes? shade/i=1:5/j=1:5 aa

  yes? ! this expression has missing values where aa is 0
  yes? shade/i=1:5/j=1:5 if aa then aa

  yes? ! replaces aa with bb where aa is 0
  yes? shade/i=1:5/j=1:5 if aa then aa else bb

  yes? ! keeps aa intact with its 0's, replacing the missing value
  yes? shade/i=1:5/j=1:5 missing(aa, bb)


William S. Kessler wrote:

> let hsmo_a = h[x=4:221@shn:7,y=4:142@shn:7]
> let hsmo = if hsmo_a then hsmo_a else h
>
> (Actually the dimensions of hsmo1 are always the same as h, but it has 
> blank values at the edges.)
>
> Since the above will likely lead to funny contours at the edges, 
> perhaps a better technique would be to taper the smoothing:
> let hsmo7 = h[x=4:221@shn:7,y=4:142@shn:7]
> let hsmo5 = h[x=4:221@shn:5,y=4:142@shn:5]
> let hsmo3 = h[x=4:221@shn:3,y=4:142@shn:3]
>
> let h1 = if hsmo3 then hsmo3 else h
> let h2 = if hsmo5 then hsmo5 else h1
> let hsmo = if hsm07 then hsmo7 else h2
>
> Billy K
>
> On Oct 2, 2005, at 5:56 PM, John Luick wrote:
>
> > If I smooth a data set with:
> > let hsmo = h[x=4:221@shn:7,y=4:142@shn:7]
> > then hsmo loses the outer three rows and columns of the original h, 
> > ie it is six rows and six columns smaller than h.
> > But I need an array whose dimensions are the same as h - the values 
> > in the outer rows and columns can be identical to h (ie unsmoothed), 
> > but not all '_' symbols.
> > Can this be done? I tried variations on xact, asn, etc. but no joy 
> > (only unjoy).
> > p.s., am I just getting old, or is the ferret syntax really strange? 
> > I've used GMT, GrADS, Matlab, Tecplot, NCAR - you name it - and I'm 
> > finding ferret the least intuitive of all. Is it worth learning to 
> > think like a ferret?
> > Thanks,
> > John (trying ferret out for a specific purpose)
> > (and if you are a long-lost friend: "hi!")
> ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
> William S. Kessler
> NOAA / Pacific Marine Environmental Laboratory
> 7600 Sand Point Way NE
> Seattle WA 98115 USA
>
> william.s.kessler@noaa.gov
> Tel: 206-526-6221
> Fax: 206-526-6744
> Home page: http://www.pmel.noaa.gov/~kessler