AgricultureOn request

Yield prediction

For each field the record gives a bushels-per-acre estimate with an 80 percent interval, taken from its state and crop’s own back-test. The back-test result is printed beside it, so the reader sees how wide the uncertainty is.

For crop insurers’ claims leads, agronomy teams and agricultural lenders.

States
Iowa, Illinois, Nebraska, Kansas
Interval
80 percent, per field
Delivered as
PDF, GeoTIFF, GeoJSON, KML, shapefile

01 · The order

What you send, and what comes back

You send

  • Fields, drawn or uploaded
  • The season
  • Corn for grain or soybeans

You receive

  • An estimate in bushels per acre for each field, with its 80 percent interval
  • The season, crop and county it rests on
  • The back-test result for that state and crop, printed beside it

02 · How it is read

From the open archive to the record

Public sources

Copernicus Sentinel-2LandsatERA5USDA NASS
  1. 01

    The model is back-tested year by year against USDA county yields, leaving each year out in turn.

  2. 02

    It covers the counties of Iowa, Illinois, Nebraska and Kansas it is trained on. A field outside them gets no estimate, and the reason.

03 · The record behind it

What it has been measured against

Each figure is printed with its denominator and scope in the paper it links to.

3 states
where soybeans clear every pre-registered back-test gate: Illinois, Kansas and Nebraska

A weekly field record from forty-three years of open imagery, Axine Labs, 2026

5.45 bu/ac
error for Illinois soybeans in the year-by-year back-test, against 7.38 for the county-average control

A weekly field record from forty-three years of open imagery, Axine Labs, 2026

Send us twenty fields

Send twenty field boundaries. The record comes back within 48 hours as a PDF with the layers, and the rest of your book can run under your own name.