1

I've got two questions:

I'm using the autoKrige function in automap to generate some raster data and am trying to figure out how to increase the resolution of the output. The output is too granular when I open it in QGIS.

So far, all I've figured out is that it probably involves the "newdata" argument. I've created a 3000x3000 meter grid and used that as "newdata," but the process is taking forever and I'm not sure if it'll finish before crashing.

Q1) How do you set the resolution of autoKrige output?

Q2) Would R's multicore package work with automap? I've realized that it is only using one of my two cores.

Ross Wardrup
  • 311
  • 1
  • 9
  • 26

1 Answers1

3

You already use the correct means of setting the output resolution: increasing the resolution of the grid you pass to newdata. In regard to performance, the bottleneck is often not the output resolution, but the size of the input dataset. Have a look at the maxdist and nmax arguments of krige, which can be passed to autoKrige and will passed on. These limit the number of points used in prediction, i.e. limit the size of the covariance matrix that needs to be inverted. Setting nmax to, say, 100 should help with performance.

In regard to your second question. You can use multiple cores only in the sense to be able to run multiple autoKrige analyses next to each other. automap does not support parallel processing within one call to autoKrige, and it would be quite a bit work to get that working. Have a look at the the parallel package that comes standard with R, and the mclapply function in that package.

Paul Hiemstra
  • 59,984
  • 12
  • 142
  • 149