[R] Re: Oceanographic lattice plots?

[W. C. Thacker]

Mike,

If depth is positive downward, you can simply plot its negative.  Fir
example, suppose you have a dataframe with ctd data:

> ctd[1,]
            id p      t      s 
104613 5993512 2 25.248 36.238 
> unique(ctd$id)
 [1] 5993512 3319148 3358648 3358538 3317809 3317562 3304786 3300555
3313863
[10] 3317901 3249908 3249987

If you are not fussy about the axis labels indicating negative values
of pressure, you can plot them like this:

xyplot(-p~t|factor(id),
  data=ctd,
  type="l",
  xlab="temperature (C)",
  ylab="pressure (dbar)")

If you want to get rid of the minus signs:

xyplot(-p~t|factor(id),
  data=ctd,type="l",
  xlab="temperature (C)",
  ylab="pressure (dbar)",
scales=list(y=list(at=seq(-3000,0,by=500),
               labels=format(-seq(-3000,0,by=500)))))

Hope this helps.

Carlisle
-- 

William Carlisle Thacker                            
                                                    
Atlantic Oceanographic and Meteorological Laboratory
4301 Rickenbacker Causeway, Miami, Florida 33149 USA
Office: (305) 361-4323           Fax: (305) 361-4392

"Too many have dispensed with generosity 
     in order to practice charity."     Albert Camus

> Subject: [R] Oceanographic lattice plots?
> Date: Sat, 18 Oct 2003 14:14:14 -0400
> From: "Mike Prager" <Mike.Prager at noaa.gov>
> To: R Help List <r-help at stat.math.ethz.ch>
> CC: Michael Prager <Mike.Prager at noaa.gov>
> 
> R 1.8.0 on Windows XP Professional.  A huge THANK YOU to the R Team for
> this marvelous software.
> 
> I am making lattice plots of oceanographic data.  The usual layout does not
> conform to plotting conventions that marine scientists use when depth is
> the independent variable.  Under those conventions, plots are made with the
> origin at the upper left, depth on the vertical axis (increasing as it goes
> down), and the dependent variable on the horizontal axis (increasing to the
> right).
> 
> That convention has implications not just in how axes are labeled and set
> up, but also when using smoothing routines such as panel.lowess(), because
> the smoothed values are on the horizontal axis, not the vertical axis.
> 
> Before I start looking at and modifying the R code that makes up the
> relevant routines, I wonder if any reader has already developed R routines
> for this purpose?
> 
> --
> Michael Prager, Ph.D.
> NOAA Beaufort Laboratory
> Beaufort, North Carolina  28516
> http://shrimp.ccfhrb.noaa.gov/~mprager/
> ***
> 
>   ----------------------------------------------------------------------
> 
> Subject: Re: [R] Oceanographic lattice plots?
> Date: Sat, 18 Oct 2003 13:29:14 -0500
> From: Deepayan Sarkar <deepayan at stat.wisc.edu>
> To: "Mike Prager" <Mike.Prager at noaa.gov>,
>      R Help List <r-help at stat.math.ethz.ch>
> CC: Michael Prager <Mike.Prager at noaa.gov>
> References: <6.0.0.22.0.20031018140413.01b88508 at hermes.nos.noaa.gov>
> 
> On Saturday 18 October 2003 13:14, Mike Prager wrote:
> > R 1.8.0 on Windows XP Professional.  A huge THANK YOU to the R Team for
> > this marvelous software.
> >
> > I am making lattice plots of oceanographic data.  The usual layout does not
> > conform to plotting conventions that marine scientists use when depth is
> > the independent variable.  Under those conventions, plots are made with the
> > origin at the upper left, depth on the vertical axis (increasing as it goes
> > down), and the dependent variable on the horizontal axis (increasing to the
> > right).
> 
> In case you decide to work on this yourself, this might be useful:
> 
> Ideally, specifications like
> 
> xyplot(depth ~ x, ylim = c(10, 0))
> 
> should reverse the y-axis direction. As pointed out some time back, this
> doesn't work in lattice currently, but that should be fixed in the future (it
> already works in my development version).
> 
> > That convention has implications not just in how axes are labeled and set
> > up, but also when using smoothing routines such as panel.lowess(), because
> > the smoothed values are on the horizontal axis, not the vertical axis.
> 
> These should be easy to modify.
> 
> Deepayan
> 
> > Before I start looking at and modifying the R code that makes up the
> > relevant routines, I wonder if any reader has already developed R routines
> > for this purpose?
> 
>   ----------------------------------------------------------------------