Skip to content

calculating area

5 messages · Pedro S. A. Wolf, Roger Bivand, White.Denis@epamail.epa.gov +2 more

Roger Bivand
#
On Fri, 26 Jan 2007, Pedro S. A. Wolf wrote:

            

      
      
      
    

        
If you think of including humans as animals, you'll find that field
scientists, for example ecologists, do this a good deal - see the package
adehabitat, perhaps function NNCH.area for Home Range Area. The package
trip also provides tools for handling spurious GPS reports. You may find 
that you need to transform the GPS coordinates from geographical to 
projected to make area calculation easier, and that maps and mapproj will 
limit what you can do, perhaps consider moving to sp, maptools and rgdal.

(There will be an R spatial workshop at the Association of American 
Geographers conference in San Francisco on Tuesday 17 April; the "tape" of 
Edzer Pebesma's eSeminar on R spatial classes and methods is not yet 
online, but we'll post when it is ready:

http://www.wun.ac.uk/ggisa/seminars/archive/2006_program/index.html

is where it should turn up).

Roger

            

      
      
      
    

        

      
      
    

        

  
    

      
      
    

  


  


      

    

    

      
      

        


  

        

      White.Denis@epamail.epa.gov
    

    

      
      #
    

  


  

      
r-sig-geo-bounces at stat.math.ethz.ch wrote on 2007-01-26 00:34:22:

            

      
      
      
    

        
experimental data

            

      
      
      
    

        
some

            

      
      
      
    

        
proficient so

            

      
      
      
    

        
with

            

      
      
      
    

        
give me

            

      
      
      
    

        
I have

            

      
      
      
    

        
which these

            

      
      
      
    

        
to know

            

      
      
      
    

        
area

            

      
      
      
    

        
lat and

            

      
      
      
    

        
like

            

      
      
      
    

        
creating

            

      
      
      
    

        
term so

            

      
      
      
    

        
occupy

            

      
      
      
    

        
Is

            

      
      
      
    

        
particular

            

      
      
      
    

        
map

            

      
      
      
    

        
advisors

            

      
      
      
    

        
package

            

      
      
      
    

        
package

            

      
      
      
    

        
find

            

      
      
      
    

        
will

            

      
      
      
    

        
rgdal.

            

      
      
      
    

        
"tape" of

            

      
      
      
    

        
Depending on where on earth your data are from, the area calculations
may
be confounded by the spherical (or ellipsoidal) shape of the earth.  So
think about using the difference in longitude as you approach the poles.
If your data are from a small area and you are only interested in
relative
differences in area then the geographic coordinates may be sufficient,
otherwise you may want to check out a web site on map projections and
choose one that will give you good area fidelity.  One class of map
projections has the property called equal-area, and preserves, in the
plane of the projection, the area relationships on the spherical (or
ellipsoidal) earth.

Denis

            

      
      
      
    

  
  


  


      

    

    
2 days later

    

      
      

        


  

        

      Michael Sumner
    

    

      
      #
    

  


  

      
Pedro S. A. Wolf wrote:

            

      
      
      
    

        
This sounds to me like what marine biologists (seals, penguins, etc.) 
refer to as "time spent", and what seems to be
more generally referred to in terrestrial tracking as a "utilization 
distribution".

We routinely use simple interpolation and grid cell counts to provide a 
simple map of time spent (or diving effort or other foraging-proxy),
and I have functions to perform this for sets of individual tracks in 
the 'trip' package.  Another common method is to use
kernel density to try to smooth over the gaps between location estimates 
in tracks and provided a density
map of time spent - but both linear interpolation and kernel density 
provide only a very simplistic model of the
inferred motion.  

It's fine for a rough first pass, and we often use the count of grids 
cells for comparing different groups - the 'sp' package makes this very 
simple to do.

In terms of calculating area I see it as a matter of choosing the 
appropriate projection for your coordinates, and then defining the
grid of cells to make comparisons between times or groups.

I can provide an example of doing this in the trip package if that would 
help.

Cheers, Mike.

  
  


  


      

    

    
2 days later

    

      
      

        


  

        

      Don MacQueen
    

    

      
      #
    

  


  

      
At 12:14 AM -0700 1/26/07, Pedro S. A. Wolf wrote:

            

      
      
      
    

        
In addition to the good advice already sent by others:

A quick and simple improvement would be to get what's called the 
convex hull. See
   help.search('convex hull')
   ?chull
and some others.

The complex hull is a polygon, instead of the rectangle you're now 
using, that will more closely surround your locations. Then all you 
need is a function to calculate the area of a polygon. I know one or 
more exists in R or a package, but I don't recall the name (or names).

            

      
      
      
    

        

      
      
    

        
-Don