3D krigging

[ELEGER]

Hello,
First of all I would like to thank the GS team for the very rich documentation available. I especially really enjoyed the documentation and examples available on readthedocs and have a big crush for the WellTestPy and AnaFlow packages.

As the title gives a hint, I would like precision on which implementation is faster/more robust for kriging in 3d (lat/long/elevation + Values). I got pretty lost on the GS or Pykrige preferences, especially after finding some opinions. Are pykrige and Gs -tools based on the sames routines ?

On a side note, I would like to perform 3D krigging of Geophysical data (electrical resistivity), and I'm wondering how I could give anysotropical/preferential direction during the krigging. For instance a fault.

With all the best
Emmanuel

[ELEGER]

Hello,
Any idea ?
Cheers ?
e.

[LSchueler]

Hi Emmanuel,

I'm really sorry for the very late reply.
Did you already perform the krigging? Does the computation take too long? - I would suggest first trying out the GSTools routines with parallelisation or even the Rust-core activated, have a look at the installation guide subsections Speeding up GSTools by parallelization and Using experimental GSTools-Core for even more speed.

And regarding your second question, you can simply provide an anisotropic covariance model to the kriging routine. Here is a very simple example, which uses an anisotropic covariance model to generate a spatial random field.