... it might be good to compute a trend surface of the square differences between the two sets of coordinates, at the sites of the good coordinates.
More information about formatting options
Before computing transformation matrix
... it might be good to compute a trend surface of the square differences between the two sets of coordinates, at the sites of the good coordinates.
## compute square difference:
c$sqdiff <- with(c, sqrt((nx-x)^2 + (ny-y)^2))
##
library(gstat)
##
coordinates(c) <- c('nx', 'ny')
coordinates(d) <- c('x', 'y')
gridded(d) <- TRUE
x <- gstat(sqdiff ~ 1, data=c, degree=2)
x.trend <- predict(x, newdata=d)
spplot(x.trend, "var1.pred")