Skip to content

Latest commit

 

History

3 Commits

Folders and files

NameName
Last commit message
Last commit date
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 

Repository files navigation

weatherodds

A command-line tool that shows a 15-day weather forecast for any US zip code, built on Google DeepMind's WeatherNext 2 model.

uv run weatherodds 02108

15-day weatherodds forecast for Boston, MA, showing daily high/low, rain probability dots, expected amount, wind, sky, and a confidence rating for each day

The problem

Most weather apps stop at 10 days, and — more importantly — they always show a single confident-looking number. "High of 84°" looks the same on the day-2 forecast (very reliable) and the day-10 forecast (largely a guess). The app never tells you which one you're looking at.

How this tool is different

WeatherNext 2 doesn't produce one forecast — it produces 64 slightly different simulations of the next 15 days, each starting from a slightly different estimate of the atmosphere right now. Where the simulations agree, the forecast is trustworthy. Where they scatter, the honest answer is "we don't know yet."

This tool shows you that agreement instead of hiding it:

  • High / Low — the middle value across all 64 simulations.
  • Rain dots (●●●··) — how many of the simulations produce rain that day. Three dots means roughly half of them do.
  • Confidence (●●● high / ●○○ low) — how tightly the simulations agree, cross-checked against a second, independent forecast model run by the European weather agency (ECMWF). When two unrelated models tell the same story, you can plan on it; when they disagree, check back in a day or two.

So a line like Sat · 79° · ●●··· · ●○○ low reads as: "probably around 79 with a chance of rain — but it's far out and the models haven't settled, so don't book the outdoor party on this yet."

Where the data comes from

Forecast data comes from Open-Meteo, a free weather API that serves Google's WeatherNext 2 model and ECMWF's forecasts. No account or API key needed (free for personal use). Zip codes are looked up offline, so your location isn't sent anywhere except as coordinates in the forecast request. The model updates twice a day.

Usage

Requires uv. From this directory:

uv run weatherodds 02108              # 15-day forecast
uv run weatherodds 02108 --days 7     # shorter horizon
uv run weatherodds 02108 --units metric
uv run weatherodds 02108 --json       # machine-readable output
uv run weatherodds 02108 --no-ecmwf   # skip the second-model cross-check

US 5-digit zip codes only.

Development

uv run pytest

Code lives in src/weatherodds/: zip lookup, API fetch, the math that turns 64 simulations into one row per day, and the table rendering.

License

Apache License 2.0 — see LICENSE.

About

15-day ensemble weather forecast TUI for US zip codes, powered by Google DeepMind WeatherNext 2 via Open-Meteo

Resources

Stars

1 star

Watchers

0 watching

Forks

Releases

Packages

Contributors

Languages