I'm trying to interpolate some point-data to a raster-map. Until now i used kriging to interpolate the point-data to the surface. now i have to use a generalized additive model for that task.
my question:
i have some soil data (texture: sand, clay, silt, bulk-density ect.) from points. i used some pedotransfer-functions to interpolate field capacity (Beke and McCormick 1985 and Batjes 1996). then i made a GAM to predict the interpolated field capacity from the soil data. i interpolated the soil data to the raster-map using kriging and then interpolated the field capacity with the GAM. but i think it is not the best way to first interpolate the field data to the raster map and then using this interpolated data for the GAM-model.So is there a way to skip the first kriging of the field-data to the raster and use only the GAM?
some code (e.g for the soil-data "clay"):
mod
my question:
i have some soil data (texture: sand, clay, silt, bulk-density ect.) from points. i used some pedotransfer-functions to interpolate field capacity (Beke and McCormick 1985 and Batjes 1996). then i made a GAM to predict the interpolated field capacity from the soil data. i interpolated the soil data to the raster-map using kriging and then interpolated the field capacity with the GAM. but i think it is not the best way to first interpolate the field data to the raster map and then using this interpolated data for the GAM-model.So is there a way to skip the first kriging of the field-data to the raster and use only the GAM?
some code (e.g for the soil-data "clay"):
mod