Muntin Open Data · Explorer CC0 NOAA GHCN · public domain

The Origin Climate Envelope

The produce a US kitchen depends on grows somewhere US drought and freeze maps can't see. This reads observed weather — never a forecast — at a representative station in each import origin's growing belt, drawn as a 12-month climatological band (1991–2020) with the most-recent readings' departure from that normal laid over the top. A regional proxy, held-until-calibrated: a station is not the farm.

The shape is the exposure

Two belts, two bands. Read which origin dips below the chill line and which never does — at a glance, with zero forecasting.

The thickness of the band is the diurnal range — the belt's climate fingerprint. Michoacan's wide band sags to the chill line at the depth of winter; Costa Rica's thin ribbon floats high and never touches it. Then the recent departure comb fades in: warm rust marks lifting off the winter floor across the grid at once — the winter floor rising inside the record itself.

Bands are 1991–2020 monthly normals; comb marks are each station's own most-recent readings vs that month's normal. Descriptive of the tracked record, not a forecast.

The origin climate grid

One card per growing belt, foods labeled. Filled band = normal TMIN floor to TMAX ceiling, Jan→Dec. The dashed chill line sits at the same temperature on every card, so you can see which winter floors dip below it.

6 °C

Each card is stamped with its own observed-through date. Stations end on different dates — "recent departure" is anchored to each station's own last readings, not a shared week, so the grid never implies simultaneity.

Current thermal tilt, all belts at once

One lane per station. Each mark is a recent night-temperature (TMIN) departure, zero-centered. Rust ▲ = warmer than its own normal, teal ▼ = cooler. Descriptive only — no belt is implied to cause another, and no line is projected past the last real dot.

* recent-departure window spans far more than a ~45-day stretch (sparse daily reporting) — read that mean with a wider error bar than the dense stations. monthly normals built from a thin baseline sample; indicative, not robust. Both are disclosed per station in the source table below and the provenance drawers.

Every station, every signal

The source of record. Normals are robust monthly means over the 1991–2020 baseline; physically-impossible readings are excluded before averaging. This table is present without JavaScript.

Growing-region climate normals and recent departures · NOAA GHCN-Daily, 1991–2020 baseline
Belt / station Foods (import-origin) Diurnal °C Jan TMIN °C Jul TMAX °C Chill ≤4°C Heat ≥38°C Recent Δnight Recent Δday Observed thru
Sinaloa76412 · Culiacan winter-veg belt Tomato, Bell pepper, Jalapeno, Serrano 16.7 10.6 37.0 5 603 +1.5 +1.6 2025-08-24
Sinaloa76458 · Culiacan winter-veg belt Tomato, Bell pepper, Jalapeno, Serrano 10.2 14.2 33.5 5 23 +0.5 +4.2 2026-06-30
Michoacan76665 · Morelia + Zamora Avocado, Lime, Lemon 17.5 5.7 27.1 272 13 +6 +1.6 2026-06-30
Michoacan76662 · Morelia + Zamora Avocado, Lime, Lemon 18.1 6.8 29.3 144 75 +4.8 -0.4 2026-06-30
Baja California76040 · Mexicali Kale, Bok choy, Collards, Brussels 18.9 5.5 43.1 741 3071 +3.4 -0.7 2026-06-29
Bajio76625 · Queretaro + Guanajuato * Raspberry 15.9 6.4 28.0 32 1 +1.3 +0.7 2025-08-24
Bajio76573 · Queretaro + Guanajuato † Raspberry 15.5 4.5 28.9 1 4 +7.6 +0.6 2026-06-30
Peru coast84531 · central coast * Blueberry 7.4 19.2 22.8 0 0 -1.3 -0.6 2025-08-23
Costa Rica81003 · lowlands Pineapple 7.8 20.8 29.7 0 1 +0 +0.7 2025-08-24
Soconusco76903 · Tapachula, GT border Cantaloupe 12.2 21.1 34.4 0 322 -5.1 -4.7 2026-06-30
Ocean-sourcedshrimp · salmonEcuador / Chile / IndiaNo land station — ENSO/ONI regime gauge, absent here

Δnight / Δday = mean of each station's last 45 valid readings minus that month's normal. Chill and heat counts are whole-record event tallies. Ocean-sourced items (shrimp, salmon) have no land station in this file — an ENSO/ONI regime gauge, deliberately absent rather than faked. * the recent-departure window spans far more than a ~45-day stretch (sparse daily reporting) — read that mean with a wider error bar. normals built from a thin baseline sample; indicative, not robust.