I’m trying to sample random points to use as background in a species distribution model. I’m using 30 second predictors from worldclim and soilgrids. The majority of the sampled random points are on the ocean, as if there were values in those cells, outside my masked area.