[R-sig-Geo] Question about cokriging on sphere

[Edzer Pebesma]

You are still using the deprecated interface that accepts data.frame
objects as input.

Please see demo(cokriging) to find out how the interface with Spatial*
objects works, and read ?proj4string to find out how to specify
projection systems for such objects.

On 10/24/2012 10:39 PM, Zhaofeng Yang wrote:
> Hi all,
> 
> I am new to geostatistics and need some help on projection. I am working on a bivariate model of a sphere and want to do cokriging with gstat package in R. My data look like
> 
>> head(mydata)
>          lon         lat         v1         v2
> 1  2.1238698  0.87854240  1.8051750  1.2222044
> 2 -2.6524104 -0.14883920 -0.4271378 -0.1659977
> 3 -2.5863044 -0.18126753 -0.4142029 -0.2782104
> 4  1.1100337 -0.03636767 -1.1252408  0.4394822
> 5 -0.5421093 -0.60182344  0.8544659  0.6174832
> 6 -3.0686334  0.53252533  0.9143837  0.9566044
> 
> where lon and lat are longitude and latitude in radian. v1and v2 are two collocated variables. The total number of points is 100 (only 6 are shown here). I want to do cokriging at a couple of new locations saved as a data frame called newloc.
> 
>> newloc
>   lon lat
> 1 0.2 0.3
> 2 0.5 1.0
> 
> My thought is to first get the sample variogram/cross variogram and use linear model of coregionalization to fit them and then do cokriging. Here is my code (basically following demo(cokriging)in gstat). I first simply neglected the sphere and just treated lon and lat as Cartesian coordinate.
> 
> g <- gstat(id = "val1", formula = v1~1, locations = ~lon+lat, data = mydata)
> g <- gstat(g, id = "val2", formula = v2~1, locations = ~lon+lat, data = mydata)
> g <- gstat(g, model = vgm(1, "Sph", 1, 1), fill.all = TRUE)
> x <- variogram(g, cutoff = 2)
> myfit <- fit.lmc(x, g)
> plot(x, model = myfit)
> newloc <- data.frame(lon = c(0.2, 0.5), lat = c(0.3, 1.0))
> z <- predict(myfit, newdata = newloc)
> 
> I don't know if the above code is correct or not but it gives me some output. Now comes the question of how to taking into account the curvature of the sphere. I combed through some R-sig-Geo posts and found something using proj4string or proj4 but I can't make it work. I really need some help on how to do projection in computing variogram/cross variogram and doing cokriging.
> 
> Please help me on this. Thank you very much.
> 
> Zhaofeng
> 	[[alternative HTML version deleted]]
> 
> 
> 
> _______________________________________________
> R-sig-Geo mailing list
> R-sig-Geo at r-project.org
> https://stat.ethz.ch/mailman/listinfo/r-sig-geo
> 

-- 
Edzer Pebesma
Institute for Geoinformatics (ifgi), University of Münster
Weseler Straße 253, 48151 Münster, Germany. Phone: +49 251
8333081, Fax: +49 251 8339763  http://ifgi.uni-muenster.de
http://www.52north.org/geostatistics      e.pebesma at wwu.de