How this is calculated. Forecast temperatures come directly from the National Weather Service via api.weather.gov — the same data that feeds NOAA's website, The Weather Channel, WeatherBug, and every other US weather app. Precipitation forecasts and historical weather come from Open-Meteo. The plant-date yield curve is calibrated against published Corn Belt research: Iowa State (Abendroth & Elmore: 95-100% corn window Apr 12 - May 18), Purdue (Nielsen: prime corn window Apr 20 - May 10), University of Illinois (Nafziger 2007-2021), and Iowa State soy (2020): ≥92% through late May. Corn holds 100% of yield potential through May 7, then declines: ≈99% by May 10, ≈93% by May 20, ≈85% by May 30. Soybeans hold 100% through April 30, then decline: ≈97% by May 6, ≈91% by May 21. We deliberately do not weight against local farm-level yield data because early planters self-select for better-managed operations — that confound is not present in randomized published trials. GDU uses base 50 °F / cap 86 °F. The 10-year normal GDU line is an average of the previous ten years' GDU accumulation at this exact lat/lon (Open-Meteo archive). The decision logic blends three things: where you are on the yield-potential calendar, whether the next 48 hours protect the seed, and how fast the heat will pull the crop out of the ground. The 48-hour rain check looks at both the expected rainfall total and the peak chance-of-rain across the window — a forecast that averages dry but carries a ≥50% chance of showers gets demoted from "go" to "caution," because the expected-value total is probability-weighted and hides the scenario where rain actually hits freshly-planted seed. Cold-rain risk uses the minimum air temperature during the actual rain window (every rainy hour expanded by three hours before and four after), not the overnight low — a 4 AM low has nothing to do with a 7 PM shower. Severity is tiered by the rain amount in that cold window: ≥ 0.20″ with temps below 50 °F is a hard stop (Hold), ≥ 0.10″ is a judgment-call caution (Mixed), and less than 0.10″ doesn't trip the rule.
Field conditions are not factored into the recommendation. Weather-based models can't reliably predict whether a particular field is fit to plant — soil type, drainage, tile, residue cover, slope, and recent foot/tire traffic all matter, and none of that is visible in a forecast. Walk your fields and trust your boots; this tool tells you whether the agronomic and calendar windows are open, not whether the dirt is ready.