Re: [ferret_users] Vector plot: force maximum?

[Steve Cousins <cousins@xxxxxxxxxxxxxx>]

Hi Billy,

Thanks for your points. This exercise actually is an attempt to *not* 
generate false data. My first (lame) thought when the other researcher 
asked me about this was to create separate NetCDF files that scaled back 
the out of range data. Of course this was immediately shot down by all 
(including myself) for a number of reasons. My second thought was doing a 
log plot but she really wanted it to be linear up to the threshold. So, I 
started exploring with Ferret.

I think Mick's solution will address your concerns too. Having the 
scaled-back values be in red will make them stand out and open to 
question if we ever forget what we did.

Cheers,

Steve
______________________________________________________________________
 Steve Cousins, Ocean Modeling Group    Email: cousins@xxxxxxxxxxxxxx
 Marine Sciences, 452 Aubert Hall       http://rocky.umeoce.maine.edu
 Univ. of Maine, Orono, ME 04469        Phone: (207) 581-4302

On Tue, 5 May 2009, William S. Kessler wrote:

> Whether or not that works (I haven't looked at your code in detail),
> generating false data seems a bad idea. What if you forgot that's what
> you were doing? What if the audience at a talk or reader of a paper
> missed that crucial point? What if you computed any further statistics
> using the false data?
>
> I would do one of three things:
> 1. plot the log of the data (or the square root, not uncommon in
> plotting vectors with a large range of values)
> 2. use your mask to blank out the too-high data: the holes are real.
> 3. use PLOT/VLIMITS to limit the height of a (line) plot
>
> One man's opinion.
>
> Billy K
>
> On May 5, 2009, at 11:08 AM, Steve Cousins wrote:
>
> > > Long time!
> > >
> > > I have a dataset with U and V velocity components. There is a river
> > > that comes into the domain that has high values that I want to
> > > limit. What I have done so far is:
> > >
> > > let wind_mask = IF ABS( (u^2 + v^2)^.5) LT .08 THEN 1
> > > let u2 = u * wind_mask
> > > let v2 = v * wind_mask
> > >
> > > vector/xskip=2/yskip=2 u2,v2
> > >
> > > This works but it puts holes in the data. What we'd like is to have
> > > it so if a speed is greater than 8 cm/sec we scale it down to 8 cm/
> > > sec so the grid doesn't have holes in it and we can still see the
> > > direction.
> > >
> > > Thinking about this as I type (don't try this at home!) I think I
> > > need something like:
> > >
> > > let wind_mask = IF ABS( (u^2 + v^2)^.5) LT .08 THEN 1 ELSE
> > >                  (.08 /  ABS( (u^2 + v^2)^.5))
> > >
> > > ... testing ...
> > >
> > > It works! Ferret is Great! Hold on. I think it is working. It fills
> > > in the gaps. It *looks* like the scaling is correct...
> > >
> > > Is there a better way to do this?
> > >
> > > Thanks,
> > >
> > > Steve
> > > ______________________________________________________________________
> > > Steve Cousins, Ocean Modeling Group    Email: cousins@xxxxxxxxxxxxxx
> > > Marine Sciences, 452 Aubert Hall       http://rocky.umeoce.maine.edu
> > > Univ. of Maine, Orono, ME 04469        Phone: (207) 581-4302