| rdist.earth {fields} | R Documentation |
Given two sets of longitude/latitude locations, rdist.earth computes
the Great circle (geographic) distance matrix among all pairings and
rdist.earth.vec computes a vector of pairwise great circle distances
between corresponding elements of the input locations using the Haversine
method and is used in empirical variogram calculations.
rdist.earth(x1, x2, miles = TRUE, R = NULL) RdistEarth(x1, x2=NULL, miles=TRUE, R=NULL) rdist.earth.vec(x1, x2, miles = TRUE, R = NULL)
x1 |
Matrix of first set of lon/lat coordinates first column is the longitudes and second is the latitudes. |
x2 |
Matrix of second set of lon/lat coordinates first column is the longitudes and second is the latitudes. If missing or NULL x1 is used. |
miles |
If true distances are in statute miles if false distances in kilometers. |
R |
Radius to use for sphere to find spherical distances. If NULL the radius is either in miles or kilometers depending on the values of the miles argument. If R=1 then distances are of course in radians. |
Surprisingly the distance matrix is computed efficiently in R by dot products of the
direction cosines. This is the calculation in rdist.earth. Thanks to Qing Yang for pointing this out a long time
ago. A more efficient version has been implemented in C with the
R function RdistEarth by Florian Gerber who has also experimented with parallel versions of fields functions.
As Florian writes:
" The current fields::rdist.earth() is surprisingly fast. In the case where only the argument 'x1' is specified, the new C implementation is faster. In the case where 'x1' and 'x2' are given, fields::rdist.earth() is a bit faster. This might be because fields::rdist.earth() does not check its input arguments and uses a less complicated (probably numerically less stable) formula."
The great circle distance matrix if nrow(x1)=m and nrow( x2)=n then the returned matrix will be mXn.
Doug Nychka, John Paige, Florian Gerber
rdist, stationary.cov, fields.rdist.near
data(ozone2) out<- rdist.earth ( ozone2$lon.lat) #out is a 153X153 distance matrix upper<- col(out)> row( out) # histogram of all pairwise distances. hist( out[upper]) #get pairwise distances between first 10 and second 10 lon/lat points x1 = ozone2$lon.lat[1:10,] x2 = ozone2$lon.lat[11:20,] dists = rdist.earth.vec(x1, x2) print(dists)