-
01AGRONOMY
Crop rules are a first approximation.
Crop traits and scoring weights are hackathon approximations from public extension guidance. They need agronomist review. New combinations from the rotation search are less proven: try them on part of a field first.
WHAT IT MEANSHow much heat, water or frost each crop tolerates, and how much each factor weighs in the score, comes from general public guidance, not from a study of your region.
FOR YOUUse the scores to compare options and start a conversation, not as a final verdict. If something doesn’t match your experience, trust your experience and tell the bot.
CLOSING THE GAPReview by agronomists, validation with farmers in real fields, and the know-how of experienced farmers feeding the knowledge base the AI consults.
-
02RESOLUTION
Satellites see in blocks, not in fields.
POWER is a ~50 km model grid, and a MODIS pixel (250 m) can mix fields. CDL can mislabel some years. GRACE is a ~300 km signal: the regional aquifer trend, not one well.
WHAT IT MEANSNASA POWER climate describes a square about 50 km wide. One MODIS vegetation pixel covers 250 × 250 m, so a small or narrow field can be blended with its neighbors. The USDA crop map (CDL) is itself made from satellites and sometimes gets a year wrong. GRACE groundwater trends describe a region about 300 km across.
FOR YOUClimate reads describe your area rather than your exact plot. Microclimates such as valleys, slopes or frost pockets may not show. If the crop history looks wrong, correct it in the chat.
CLOSING THE GAPFiner-resolution sources such as Landsat (30 m) for vegetation, plus the farmer’s own corrections, which always take priority over the satellite reading.
-
03PRICES & YIELDS
No local prices yet.
No local prices yet. “Income” is a relative index plus the trial-based yield effect, and the bot asks farmers for local prices and buyers.
WHAT IT MEANSFarmPartner doesn’t know your market prices or your own yields. The “income” score ranks rotations against each other, using a relative market index and the yield gain that 3,663 field trials show after each previous crop (Mudare et al. 2025); it is not a forecast in dollars or euros.
FOR YOUBefore switching crops, check local prices, buyers and contracts. The bot asks you about them so the comparison reflects your real market.
CLOSING THE GAPConnecting local price and yield data, and learning from the prices and buyers farmers share in the conversation.
-
04COVERAGE
Deepest in the U.S., useful worldwide.
U.S.-deepest: CDL and SSURGO are U.S.-only. Climate, NDVI and the forecast work worldwide; outside the U.S. the farmer types their crops and FAOSTAT limits the search to crops the country grows.
WHAT IT MEANSCrop history (USDA CDL) and detailed soil data (USDA SSURGO) only exist for the United States. NASA climate, vegetation health (NDVI), soil moisture, crop stress, groundwater trends and the weather forecast cover the whole planet.
FOR YOUOutside the U.S., tell the bot which crops you grow and what your soil is like. The climate analysis works just the same; crop history and soil detail come from you.
CLOSING THE GAPGlobal soil and crop-type datasets, so farmers everywhere get the same depth of analysis without typing it in.
-
05SATELLITE TIMING
Some satellites answer minutes later, and not every day.
SMAP is a 9 km grid. ECOSTRESS passes are irregular, and clouds leave gaps. Both arrive 20–30 minutes after onboarding, and until then the report lists them as “processing”.
WHAT IT MEANSSMAP (soil moisture) and ECOSTRESS (crop water stress) are requested from NASA AppEEARS for each new field. NASA processes them as a queue, so they come in 20–30 minutes after the first answer. SMAP describes a 9 km square; ECOSTRESS sees 70 m but only on the days the space station passes over, and clouds hide some of those passes.
FOR YOUYour first comparison uses climate, vegetation and crop maps; soil moisture and crop stress are added to the report and to the watering advice once they arrive. If your soil is wetter or drier than the bot thinks, say so.
CLOSING THE GAPRequesting the satellite data ahead of time for common farming regions, and adding finer, more frequent soil-moisture products as they become available.
■ WHY WE SAY THIS
A recommendation is only useful if you know how far to trust it. That’s why every answer comes with its evidence, the data gaps and the risks to check locally.
■ FarmPartner supports, and does not replace, farmer judgment and local agronomic expertise.Known limits on GitHub ↗