Where to find crop yield history data by parcel

Parcel-level yield data is scattered across a half dozen systems that don't talk to each other, and most of them weren't built with agrivoltaics siting in mind. If you're a BD lead building a candidate list of fields for a dual-use project, or a co-op partnerships officer trying to figure out which members' ground could support an array, the honest answer is that no single clean dataset exists. You have to know where to look and what each source measures.
Where the records live
Combine yield monitors are the closest thing to ground truth. Most row-crop operations running equipment from the last fifteen years have yield monitor data sitting in a cab display, a John Deere Operations Center account, or a Climate FieldView account. This data is tied to GPS points as the combine moves through the field, so it maps to the actual harvested area rather than a program boundary. The catch is that it belongs to the grower, and getting it released for a siting study means asking.
Crop insurance records are the other major source. USDA's Risk Management Agency requires Actual Production History reporting for insured acres, and that data rolls up by insurance unit, which is sometimes a whole farm and sometimes a single field depending on how the grower elected coverage. APH gives useful multi-year averages, reported in bushels per insured unit rather than GPS polygons, so matching it to a specific parcel boundary takes manual work.
USDA NASS publishes county and district yield statistics that are public and free, but they're an average across every field in the county. Good for a regional sense of what a crop typically does. Useless for telling two adjacent fields apart.
FSA farm records (the 578 forms) give you planted acreage and crop history by farm number, which can be cross-referenced against USDA Common Land Unit boundaries. For a lot of developers this is the best free starting point for matching rotation history to an actual field outline, though the data usually lags a season or two depending on the county office.
Why the boundaries never line up
These systems disagree on what counts as one unit of ground, and that mismatch causes more trouble than tracking down the data itself. A crop insurance unit might span three fields a grower farms as one block. An FSA farm number can include acreage across two counties. A combine yield file only covers what the machine drove, which can cut off at a fence line, a center pivot radius, or a terrace the equipment skipped entirely. When a developer hands a co-op a parcel map built off tax lines, and the co-op's agronomist is thinking in insurance units, the two sides end up arguing about different ground before anyone has looked at a single yield number.
Building a siting shortlist from yield history
For agrivoltaics siting, layering multi-year yield trends, where you can get them, against crop type and parcel geometry beats treating any one source as the final word. A field with a consistent low-yield corner, drought stress showing up in the same spot season after season, or a rotation that already includes a lower-value cover year is a different conversation than a field that's pulled top corn yields for a decade straight. Multispectral satellite data fills the gap when grower records aren't available or won't be shared. It won't replace a combine's yield monitor, but tracked seasonally it shows which parts of a parcel have consistently underperformed, which is often enough to start a real siting conversation instead of a dispute over whose map is right.
Agrisolar Siting puts that seasonal yield picture, crop type, and parcel geometry on one map, so a developer and a co-op are looking at the same ground before anyone drafts a lease.
See what your own fields look like on that map.