Not a more accurate forecast — a less lossy one.
The NWS publishes on a 2.5 km grid, and forecasters publish their own confidence. Mainstream apps throw both away: they sample one point, stamp it across a whole city, and collapse the day into a single percentage. “90% chance of rain” formally means a hundredth of an inch, somewhere, sometime — it says nothing about timing, duration, or coverage, which are the only things planning depends on.
GulfLine is the translation layer that keeps what’s already in the public data. Zones are sub-city and honestly aggregated, reporting a median and a range rather than inventing precision. Days are rendered as windows — when it starts, when it clears, how confident that timing is. Every payload carries its source, its age, and the reasons behind its confidence grade.
The API is the product; the iOS app and Cartway’s weather layer are both just consumers.
Features
- Sub-city zones — Forecasts on the NWS grid cell, aggregated to zones with the spread shown, not averaged away
- Timeline forecasts — The shape of the day as rain windows and dry stretches, with plain-language coverage character
- Regime awareness — Sea breeze, outflow, capped, tropical surge, or frontal — classified daily with the evidence that fired
- Nowcast & exposure — Odds you stay dry for the next hour, heat guidance, and lightning risk for people who are outside
- Pool lightning clearance — GOES satellite lightning against a pool’s own radius and wait policy; stale data degrades to “assume closed,” never a false all-clear
- Ask — Plain-language questions answered by a local model that phrases the reply while the engine computes every number