The science behind Turf Alert

What we measure

Every day, Turf Alert pulls weather data for your rough location: soil temperature, soil moisture, rainfall, and evapotranspiration (how much water is leaving the soil through evaporation and the grass itself). That's a richer picture than the air-temperature-only view most lawn advice is built on — soil temperature is what drives root and shoot growth, and it lags behind air temperature in ways that matter.

Two models

A growth model tracks how much your grass is growing day to day, based on soil temperature and daylight — mow alerts fire on accumulated growth, not a fixed weekly clock, so the interval stretches in a slow spell and tightens in a warm one. A water-balance model tracks soil moisture the way agronomists do — rainfall in, evapotranspiration out — so watering alerts reflect an accumulated deficit, not just "it hasn't rained in a while."

Checked against independent data

Internal consistency isn't enough — a model can be perfectly self-consistent and still wrong. So every threshold in Turf Alert has been checked against independent, external US data:

• The core growth constant, confirmed independently

  PACE Turf — the original published source for this project's growth-potential curve — states the optimum temperature for cool-season turf as 67.5°F (19.7°C), essentially exactly this project's own value. Independent confirmation the US model hasn't drifted from its source.

• Tracks a real, independent US phenology signal

  Checked each covered city's predicted first-mow date against the USA National Phenology Network's Extended Spring Index — a 30-year average tracking when lilac and honeysuckle shrubs leaf out each spring. The gap between the two events holds steady across most cities, and where it doesn't (two Pacific Northwest cities with a notably longer gap), the raw soil-temperature data explains why: a mild winter but a slow, cool spring — a real climate signature, not a model quirk.

• Reconstructed 5 real, named US droughts — across all 12 cities, unprompted

  Ranked each city's driest backtest years purely by the model's own drought-stress numbers, then checked them against real, documented US droughts — the 1988-1990 North American drought, the 2012 US drought (the second most widespread on record), the 2015 Pacific Northwest snow drought, and more. Every one of the 12 cities landed inside a real drought window at least once; several hit three or four of their five driest years.

None of this makes the model perfect — weather forecasting and biology both carry real uncertainty, and we say so plainly in the terms. This validation work is ongoing as coverage expands — but every threshold behind your alerts has already been tested against 86 years of US weather and checked against real, independent data, not picked because it sounded reasonable.