Loading…
Right now
Next hours
7 days tap a day for details
Hourly charts
Data
Forecasts are kept for 30 minutes and radar for 5 (models only re-run every 1-6 hours, so refreshing more often brings nothing new and wears out the free APIs).
Theme all straight out of the early 2000s
Experts switch sources on or off · off = not even asked
How to read weathermoe
Everything you see is a vote of ~20 experts (weather models). The app shows what most of them say, and how much they agree.
| You see | It means |
|---|---|
| Big temperature, "Partly cloudy" | the consensus: the middle of all forecasts for this hour |
| high medium low | confidence: how much the experts agree. Green = trust it, yellow = roughly right, red = could easily change |
| Speech bubble | the rain answer for the next 24 h, in plain words |
| Right now | what is really happening: radar and the nearest weather station, not a forecast |
| Next hours tiles | icon + consensus temperature per hour, the dot is that hour's confidence |
| 7 days, coloured bar | the day's low → high on one scale for the whole week; ☔ % = chance of a rainy day (≥ 1 mm); tap a day for ranges |
| Small numbers like 17…20° | the likely range: 8 of 10 forecasts fall inside |
| 🌫️ | drizzle possible: too light to measure, but some experts flag it |
| 👑 Expert of the day | which expert got yesterday right, checked against a real station |
| Charts tab | line = consensus, shaded band = likely range, bottom strip = confidence per hour |
Rule of thumb: narrow range + green dot = plan on it. Wide range or red dot = check again later.
What does “chance of rain” mean?
It is the share of forecasts that see rain. Every model and every ensemble gets one vote, so 30 % means about 3 in 10 forecasts make it rain in that hour (≥ 0.1 mm) and 7 in 10 keep it dry. It is not how long it rains, and not how much of the area gets wet.
| Chance | Answer | What to do |
|---|---|---|
| under 15 % | ☀️ Dry | leave the umbrella |
| 15-40 % | 🌤️ Probably dry, small risk | a shower is possible, often short |
| 40-70 % | 🌦️ Maybe, forecasts split | nobody really knows yet, check later |
| over 70 % | 🌧️ Rain expected | take the umbrella |
The 7-day list uses ≥ 1 mm per day, so a few drops do not count.
Drizzle (🌫️) is counted separately: tiny drops from low cloud, often under 0.1 mm per hour. Models flag it in their weather code even when the amount rounds to zero, so the app shows it from 1 in 10 forecasts. Radar cannot see it.
How is confidence calculated?
Every forecast is one opinion. A model counts once. An ensemble (dozens of slightly different runs of one model) also counts once in total, so each of its members gets a small share. That stops 51 ECMWF members from outvoting everyone else.
From all opinions we take the middle (consensus) and the range that holds 8 of 10 of them (likely range). Narrow range = forecasts agree = high confidence. For rain it is about the vote: almost all yes or almost all no = sure, 50/50 = unsure.
🟢 high ≥ 70 · 🟡 medium 40-70 · 🔴 low < 40. Some sources share DNA (yr.no runs on MET Nordic + ECMWF), so agreement is a bit optimistic. Regional models that silently fall back to a global model are detected and dropped as clones.
Who disagrees? (next 72 h)
Mean distance from consensus. Big number = outlier for this place right now. Not a skill score: the crowd can be wrong together.