fieldmesh
Plant growing conditions: the list, the dimensions, the sources#
2026-10-08. A plan, saved for later — nothing here is built yet. The idea: for each plantable plant, dig out the conditions it grows best in — soil, water, climate, light — and match them against what a farm actually has, which fieldmesh already measures or looks up. That is a layer-3 (decision support) module: "what grows well here, and what would grow well here in twenty years."
The list is plant-list.csv: 477 plants, 98 families — cereals and pseudocereals, pulses, roots and tubers, vegetables, oilseeds, sugar, fibre, beverage crops, spices and herbs, temperate, berry and tropical fruit, nuts, forage grasses and legumes, cover crops, agroforestry trees, pollinator and beneficial plants, cut and edible flowers, bee forage, timber, Christmas trees and forest farming, plantings against erosion, and a do-not-plant list of invasives. Animals, bees, fertilizers, and what mixes with what, are in animals-bees-and-inputs.md; tree farms and erosion control in trees-and-conservation-planting.md; growing them as nursery stock in nursery.md; growing under cover in hoop-houses.md. Keyed by scientific name, because every data source below is. Each name is to be checked against USDA PLANTS when its conditions are pulled; the common name is a label, not a key, and multilingual names can come from AGROVOC (see Q4) rather than from here.
What to dig out, and where each side comes from#
Every dimension has two sides: what the plant needs, and what the site has. The site side is mostly in reach already.
| dimension | the plant needs | the site has — source |
|---|---|---|
| Soil texture | adapted to coarse / medium / fine soils | sand, silt and clay at each depth — tools/soil_survey.py (USDA SSURGO) in the US; ISRIC SoilGrids worldwide |
| Soil chemistry | pH range; tolerance of salinity, lime (CaCO₃) | the same surveys; a farm's own soil test beats both |
| Drainage | anaerobic (waterlogging) tolerance; rooting depth | survey drainage class and depth to a restrictive layer; fieldmesh soil-moisture probes show how long the soil stays saturated |
| Water | annual rainfall range; drought tolerance; moisture use; seasonal water demand | rainfall from NASA POWER (daily, any point, 1981 on) or CHELSA normals; available water capacity (field capacity − wilting point) from the soil survey; irrigation availability from the farm |
| Cold | minimum temperature survived; frost-free days needed; for fruit trees, chill hours | daily minima from NASA POWER → last and first frost dates, frost-free days, chill hours |
| Heat and season | temperature range for growth (minimum, optimum, maximum); growing degree days to maturity, with a base temperature | daily temperatures → growing degree days and season length |
| Sunlight | shade tolerance; daily light needed; day-length response (short-day, long-day, neutral); C3 or C4 photosynthesis | daily solar radiation from NASA POWER (1984 on); day length computed from latitude and date; shade from trees and buildings only on the farm itself (observation, or a light sensor) |
| Climate class | the Köppen classes it is grown in | Köppen class at the point (already used for the dwell tables, Q24) |
| Climate trend | — | ~/dev/climate-analysis: Berkeley Earth monthly temperature, with an absolute baseline, for 7 continents, 23 countries and 15 US states — "historical versus now" by region. CHELSA's CMIP6 futures for "in 2050" at a point |
Also worth carrying, as found while drafting:
- Invasiveness, by region. Several plants on this list are useful in one place and invasive in another — leucaena, buffelgrass, napier grass, black locust, Kentucky bluegrass outside its range. A suitability list that recommends an invasive is worse than none. USDA PLANTS carries US noxious and invasive status; GRIIS (the Global Register of Introduced and Invasive Species) is the worldwide candidate.
- Day length matters more than it looks. An onion variety bred for long northern days will not bulb in Kenya; many soybean and rice varieties flower by day length. Conditions differ by variety as well as by species — the list stops at species.
- Elevation, slope and aspect (a south slope is warmer and drier in the northern hemisphere) and wind exposure.
- Pollination: whether the crop needs pollinators or a second variety to set fruit.
- Role in a rotation: nitrogen fixation, family (pest and disease carry-over between related crops), rooting depth, allelopathy.
Sources, and whether we may build on them#
Checked 2026-10-08. The licence matters because the project is CC-BY / Apache / AGPL, and a non-commercial source cannot be redistributed in it.
| source | gives | licence | use |
|---|---|---|---|
USDA PLANTS characteristics — plantsservices.sc.egov.usda.gov/api |
the plant side above, almost field for field: soil textures, pH, rainfall range, drought tolerance, moisture use, minimum temperature, frost-free days, salinity, CaCO₃, anaerobic and shade tolerance, root depth, fertility, nitrogen fixation | CC0 (public domain) | the backbone. Probed: maize, sorghum, wheat, alfalfa and tomato all have characteristics. Values are coarse — maize is listed as not adapted to fine-textured soils, which is not true of clay loams — so a starting layer, cross-checked, not an answer |
| NASA POWER — agroclimatology | daily temperature, rainfall, humidity, wind, solar radiation for any point | CC BY 4.0 | the site side, worldwide, per farm |
| CHELSA v2.1 climatologies | 1 km monthly normals, worldwide; CMIP6 futures | CC0 for the climatologies; the futures' licence still to check | site climate where a point series is not enough; projections |
| ISRIC SoilGrids | soil texture, pH, organic carbon worldwide at 250 m | CC BY 4.0 | soil outside the US |
| USDA SSURGO | soil at field scale, US | public domain | already in tools/soil_survey.py |
| DSSAT crop models (CROPGRO genotype files) | per crop: cardinal temperatures of development and day-length response; per cultivar, phase lengths — simulated against a site's weather in modules/advise/dssat_crops.py |
BSD-3-Clause | the food-crop side for 22 CROPGRO crops, CERES maize, sorghum, pearl millet and rice (drill-seeded and transplanted), and Cropsim-CERES wheat, barley and rye (with vernalization and winter kill) |
| FAO ECOCROP | 1,710 crops: temperature, rainfall, pH, photoperiod, soil, killing temperature, season length — the closest thing to the whole list in one table | Corrected 2026-10-11: FAO's data catalog lists ECOCROP under CC BY 4.0 (entry, created 2025-03-18). The CC BY-NC-SA 3.0 IGO assumed before is FAO's licence for its publications. The Recocrop R package (GPL-3) carries the parameters, described as expert opinion |
not yet used: a note asks FAO to confirm the catalog's licence covers the parameters, and for a bulk file (the ECOCROP tool has no documented export) |
| WorldClim | climate normals | reported as non-commercial | avoid; CHELSA covers it |
| Fadón et al. 2020, Agronomy 10:409 | chill and heat requirements of 531 stone-fruit cultivars, each with model, method, place and study | CC BY 4.0 | fruit chill, bloom and frost after bloom, in chill-requirements.csv, modules/advise/fruit_bloom.py and the advisor |
| González Noguer et al. 2023, Front. Hortic. 2:1217689 | 173 published apple requirements — dessert, cider, crab and rootstock — each with model, method, place and study | CC BY 4.0 | apples, in chill-requirements-pome.csv and the advisor |
| Fadón et al. 2023, Agronomy 13:518 | chill and heat of 16 European pears over twenty years at Zaragoza | CC BY 4.0 | pears, likewise; blueberries still lack an open table (routes in site-match-tennessee.md) |
| CITA Aragón, Agroclimatic requirements of temperate fruit trees (2025) | 143 cultivars, three chill models | CC BY-NC-ND 3.0 ES | not used |
| USDA GRIN-Global evaluation data | chill descriptors for a few crops (walnut, almond, some peach accessions) | public domain | sparse, and no bulk download; not used yet |
| FAO-56 (crop coefficients) | water demand by growth stage, rooting depth, allowable depletion | an FAO publication (same licence question); the values are published facts | needs a decision: cite and tabulate facts, or ask |
| USDA Salinity Laboratory (Maas–Hoffman) | salt-tolerance thresholds by crop | US government work | to verify |
| AGROVOC | crop names in 40+ languages | to verify | names, not conditions |
What to build, in order, when this is picked up#
Steps 1–4 are built (2026-10-08) and run on three Tennessee farms: site-match-tennessee.md. The run's main finding is that USDA PLANTS covers conservation, forage and timber plants well and food crops hardly at all, and that its limits are guidance rather than limits.
- Resolve each scientific name to a PLANTS symbol. The old checklist download link now serves the web application; the API's search, or the CC0 Darwin Core archive of PLANTS on Zenodo/GBIF, are the routes.
modules/advise/plant_conditions.py— pull characteristics for the list into a table, with the source and date on every value.- The site side for one point — NASA POWER for a farm's coordinates → frost dates, frost-free days, growing degree days, chill hours, rainfall by month, solar radiation and day length by month; soil from the survey.
- The match — each plant scored against the site, dimension by dimension, saying which condition rules a plant out rather than one opaque score.
- Then the hard parts — variety rather than species, invasiveness by region, and the 2050 version from climate projections.