[R-sig-Geo] RE : Mass conversions of Lat/Long to UTM (many zone)

Philippe Roy Roy.Philippe at ouranos.ca
Fri Aug 25 17:22:02 CEST 2006

-----Message d'origine-----
De : Roger Bivand [mailto:Roger.Bivand at nhh.no] 
Envoyé : 24 août 2006 15:28
À : Philippe Roy
Cc : r-sig-geo at stat.math.ethz.ch
Objet : Re: [R-sig-Geo] Mass conversions of Lat/Long to UTM (many zone)

On Thu, 24 Aug 2006, Philippe Roy wrote:

> Hi everyone!  I'm new to R.  I'll try using it for some Kriging.  My aim
> in the short-term is to convert my 2300 stations database from lat-long
> to UTM.  This network of stations goes from Virginia to the south, to
> Montréal to the north, east coast to Michigan in the west.  So, I have a
> lot of UTM zones to consider.
> Reading the archives, I've seen this formula to calculate the zones,
> based on lat-long coordinates :
> UTM = floor((lon + 180) / 6) + 1
> So, if I understand correctly, I need to run this formula on each
> station to get their zone, abnd then I use something like convUL ?

If you want to use all the stations together, then they must be in the 
same zone (otherwise the western edge of the zone will jump for stations 
in different zones). I think that for continental North America a central 
UTM zone will not be a very good idea, and that probably a Lambert 
Azimuthal Equal Area is a better choice.

You can then use project() or spTransform() in the rgdal package, for 
example, with EPSG code 2163:

US National Atlas Equal Area

+proj=laea +lat_0=45 +lon_0=-100 +x_0=0 +y_0=0 +a=6370997 +b=6370997 +units=m

with some modification of +lat_0 and +lon_0 to suit your region.

> 	[[alternative HTML version deleted]]

Roger Bivand
Economic Geography Section, Department of Economics, Norwegian School of
Economics and Business Administration, Helleveien 30, N-5045 Bergen,
Norway. voice: +47 55 95 93 55; fax +47 55 95 95 43
e-mail: Roger.Bivand at nhh.no

Ok, so, in my kriging code, I need to specify a maximum distance in km (or meter can't remember).  Is this Lambert Azimuthal Equal Area good for Euclidian distance?  I mean, I need to specify the maximum distance to consider a station or not for each grid point of my regular grid.  Would that be good then?

Sorry if it sounds non-understandable! :/

More information about the R-sig-Geo mailing list