Within each county-rectified MLRA, the COMET-Planner team developed a unique random point sample of approximately 100 points per broad land use category, with some variation in sample size depending on the size of the MLRA and the density of agricultural land use within the MLRA (Figure 2 in COMET-Planner Report).
The limitation on number of points is just if there’s enough cropland or grassland. In places with low sample size for cropland practices, it just means there’s very little cropland. As such, it’s also less likely that the tool will be used much in those places. We also don’t want to oversample what little cropland there is, especially if it means we end up with multiple points in the same field. We don’t know actual locations of current practice adoption so that is not used as a criteria, nor can we really determine suitability.