I’ve been spending some time with the National Agricultural Statistical Service Crop Data Layer (everyone with a passing interest in GIS should know about CropScape).

Since my job is in Scott County, MN, that’s where I’ve played with the data the most – 2006 through 2011 is currently available. First order of business: clean the data. I started by running a majority filter, turning this:

Into something more like this:

Much prettier! But sharp eyes will notice that the majority filter tends to eat away road features, and we want to keep the original (fairly accurate) impervious surface. So after some quick r.mapcalc, I got this:

Fun stuff! Hopefully gets us more accurate fields (assuming farmers aren’t planting tiny plots of soybeans interspersed with their alfalfa), while retaining accurately linear road features.
Just for the heck of it, I thought I would take a look at what areas had been planted in only corn/soybeans during the entire 2006-2011 period.

This is scary stuff for a water quality planner. Most of the northeast of the map is not in straight corn/soy rotation, but a not insignificant minority is; that means the potential for erosion is pretty elevated. Obscuring interpretation here is the fact that much of the northeast is urban, lake, and wetland area that isn’t farmable. The southwest corner of the map is downright terrifying for water quality; nearly every piece of arable land appears to be in strict row-crop production.