How to test for spatial non-stationarity in R to determine if local regression model is needed?

Question

I have a dataset for which I implement a regression model and from which I assume that the coefficients vary locally. If a spatial non-stationarity is given, it makes sense to run a local regression model, in my case a Geographically Weighted Regression (GWR).

To find out, if there is a spatial non-stationarity I am aware of the Koenker test which can be calculated with an Ordinary Least Square (OLS) regression model in any GIS software. But for this project I am working with R and I need to find out, if spatial non-stationarity is given or not. Is there any statistical test implemented in R which can give me that result?

I know of the spgwr package in R, but it is only to implement the GWR, not to test if GWR is necessary.

Answer 1

1. There is the GWmodel package on CRAN. See: https://cran.r-project.org/web/packages/GWmodel/index.html You should check out the function gwr.montecarlo()
2. There is a whole book centered around spatial statistics using R with a chapter on stationarity here: https://becarioprecario.bitbucket.io/spde-gitbook/ch-nonstationarity.html