[R-sig-Geo] projecting shape files, in R?

Michael Sumner mdsumner at gmail.com
Fri Oct 3 14:09:49 CEST 2014

It's not really possible to guess from a literally infinite number of
possible projections. Usually there is some information, or you can
infer it from context. Really you need at least the family of the
projection, then you can maybe guess at the details until you find a
good match. Experience can provide the family by looking at the data,
but only careful hunting can find the details. It's important to find
and tell the story of what the data are, clues are hidden in many

For example, this map was fairly simple to work out (for me, given my
background, the people I know, the stuff I've seen 'around here'):


Through experience I know that 71 S is a commonly used latitude for
the "true scale" of a Polar Stereographic projection, and that Polar
Stereographic (the family) is pretty popular for Antarctic data. (It's
because the coast of Antarctica is roughly "averagely" at 70S, and
secant-slicing the ellipsoid at this latitude provides a better fit
for areas around the coast, a good compromise compared to the single
tangent at the south pole). But, perspective on this aspect of PS is
relatively new for me. The rest was easy, I can find two points with
the printed graticule, and also determine the central longitude at
110E and so with tools to transform longlat to PS I can fully
georeference this image - well good enough that I cannot see a
problem. I cannot see any of this metadata on the web page, so I had
to resort to personal experience and socio-historical heuristics. (I'm
still not sure if this uses WGS84, a sphere, or maybe Hughes 1980 - I
have to do some more tests to be sure, but maybe it's not answerable
or maybe doesn't even matter).

So, you need to go hunting in your own social context, show us the
map, give us more clues. It's possible to guess/get lucky, but you
really need to find out all the details. Guessing from nothing is a
no-hope. There are some CRS that won't be candidates given the range
though, for example it's unlikely that for all of Europe that UTM is
used so you need some idea about the sensible scope for different
families - clues in one direction - as well as knowledge about the
data, what it's used for, and so on - from the other direction.

Cheers, Mike.

On Fri, Oct 3, 2014 at 9:32 PM, Alessandra Carioli
<alessandracarioli at gmail.com> wrote:
> Dear Mailing list members,
> I need a projected shape file but I don’t have the .prj information and I was wondering if anything like projecting a shape file to obtain coordinates in longitude and latitude can be done. My issue is that I am producing semivariograms and, although I think distances are in meters, I want to be 100% sure. Also, being the considered area rather big (Europe) considering the projected coordinated would make my computation more precise.
> Any ideas?
> Ale
> _______________________________________________
> R-sig-Geo mailing list
> R-sig-Geo at r-project.org
> https://stat.ethz.ch/mailman/listinfo/r-sig-geo

Michael Sumner
Software and Database Engineer
Australian Antarctic Division
Hobart, Australia
e-mail: mdsumner at gmail.com

More information about the R-sig-Geo mailing list