Sebastian Müller
Sebastian Müller
If we use only K nearest neighbors, we still can't provide a search radius for the moving window. K-nearest neigbors always depend on the densitiy of the input data.
Simple kriging is not implemented in PyKrige ATM. But in GSTools, we provide it: https://geostat-framework.readthedocs.io/projects/gstools/en/stable/examples/05_kriging/00_simple_kriging.html#sphx-glr-examples-05-kriging-00-simple-kriging-py
Another problem with simple kriging and the current implementations in PyKrige is: Simple kriging depends on the covariance function of a variogram model to build up the kriging system. Since...
Hey Malte, Thanks for your thoughts on that. I don't think that it lifts back the distance to 3D, since it directly depends on the great circle distance. I also...
Ahhh! Thanks for the enlightenment! This could change a lot of stuff. I need to think about that.
I would suggest using the Yadrenko models as the standard interpretation for geo-coordinates. Then we can drop the whole haversine procedures and just calculate the fields in 3D (with a...
@mjziebarth : I would suggest doing this within the [GS-Framework 2.0 project](https://github.com/orgs/GeoStat-Framework/projects/1). By increasing the major version of `Pykrige` and `GSTools` we can break backward compatibility. We can still keep...
This is now solved in GSTools and will be used in PyKrige in the future: - https://github.com/GeoStat-Framework/GSTools/pull/113
This issue should also mention, that we need to implement geographic coordinates for universal kriging.
#217 is related.