10

This is a follow-up to this question. I have now managed to plot variograms of the data using the answers given to that question. The code I am using (where im is a variable containing the dput data given at https://gist.github.com/2780792 is

library(gstat)
point_data <- as(im, 'SpatialPointsDataFrame')
v <- variogram(band1 ~ 1, data = point_data)
plot(v)

This gives me the following plot: enter image description here

As you can see, the numbers on the distance axis range from 0 to around 150. However, the units of the data are degrees:

> head(coordinates(point_data))
             x        y
[1,] -1.849353 52.06165
[2,] -1.759728 52.06165
[3,] -1.401227 52.06165
[4,] -1.311602 52.06165
[5,] -1.221977 52.06165
[6,] -1.132352 52.06165

Given that, what are the units of the distance measures on the X axis? I would expect them to be in the same units as the co-ordinates. It doesn't make sense for them to be metres, I guess they could be km.

I can't find anything in the gstat manual about this, so I'd have assumed they used the underlying data's units, but this doesn't seem to be the case.

Any ideas?

Community
  • 1
  • 1
robintw
  • 27,571
  • 51
  • 138
  • 205
  • I don't know for certain but they may be great circle distances in km (the distance between the first two observations is ~10km for example). – Gavin Simpson May 26 '12 at 21:47
  • If you want to check the distances manually, this blog may help you set up functions to run those calculations: http://www.r-bloggers.com/great-circle-distance-calculations-in-r/ PS: There is a http://gis.stackexchange.com for GIS related questions. It's a much smaller community than SO, but there are definitely some R users over there! – Bryan Goodrich May 26 '12 at 21:59

1 Answers1

13

I found some people asking similar questions on the mailing lists and rechecked the details of the help files. Your data is unprojected (proj=+longlat). As the documentation and information I found in mailing lists indicate, it should be in km based on great circle distances. Check out the projection (logical) parameter in variogram:

For projected data, Euclidian distances are computed, for unprojected great circle distances (km).

Bryan Goodrich
  • 741
  • 4
  • 7