How this list is built (the safety model).
Every row is generated by a fixed rules engine that compares this week's measured
satellite indices against published agronomic thresholds.
There is no language model
involved in producing any flag, number, or recommendation — so nothing on this page can be
"made up." Click any field to see the exact measured value and the threshold that fired it.
Fields with stale or cloudy imagery are marked
No data and deliberately get no recommendation.
Dollar figures are
wide estimates to prioritise which grower to call, not guaranteed
returns or prescriptions. Thresholds are conservative starting points that get calibrated
against your ground truth per crop.
The agronomist confirms in the field before anyone applies
anything — this tool ranks the calls, it does not make them.
measured = real Sentinel-2 reading for a live field.
sample = placeholder field to illustrate the territory at scale.
Audit it, don't trust it. The badge at the top of this page runs seven of these safety
checks live in your browser every time the page loads. A further 51 run server-side in one
command (
./verify.sh) — including a test that proves this page and
the audited engine are the same engine, on 4,047 cases. Full method:
VERIFICATION.md.