One question: can I see it from here, at this hour?
Many aurora apps look only at Kp and may alert you under a fully overcast sky. Our score estimates your chance of seeing the aurora by eye or camera at a specific place and hour. It is an estimate, not a guarantee.
The formula
Score = 100 × A × C × D × M × L
Each factor runs from 0 to 1. Because they are multiplied, any factor near zero pulls the score to zero: a great aurora behind solid cloud is still invisible.
A: Aurora activity
- Now (first half hour): from NOAA's OVATION model, which uses solar wind measured at L1 to estimate aurora about 30–90 minutes ahead. We look not only overhead but along your poleward horizon up to about 1,000 km, discounting by distance. OVATION tends to under-state probability, so we map it through a saturating curve.
- Later hours: over the first 3 hours we blend from OVATION into the Kp forecast. Beyond that we use NOAA's 3-day Kp forecast with SWPC's rule that the oval's edge is near 66° geomagnetic latitude at Kp 0 and moves about 2° per Kp step. Because SWPC Kp forecasts tend to run high at Kp 3+, we damp predicted values slightly.
C: Clear sky
Cloud comes in three layers, weighted:
- Low cloud: 0.70 (blocks aurora completely)
- Mid cloud: 0.25
- High cloud: 0.05 (thin cirrus often lets bright aurora through)
C = 1 − (0.70 × low + 0.25 × mid + 0.05 × high) / 100
If cloud data are missing we assume a middling 0.6 and mark confidence as low.
D: Darkness
Based on how far the Sun is below the horizon. Faint aurora needs the Sun about 16° down; very bright aurora starts to show with the Sun only 6° down, in twilight. So the threshold shifts with activity. In northern summer D stays near zero all night, which is why the score is low then.
M: Moonlight
Uses the Moon's altitude and illuminated fraction. A full moon high in the sky cuts the score by at most about 40%, and less as activity rises, because a bright moon cannot hide a strong display. With the Moon below the horizon, M = 1.
L: Light pollution
By place type: dark countryside 1, village 0.85, town 0.6, city centre 0.4. Driving a few tens of kilometres out of town can raise the score a lot.
Score bands
- 80–100: Go! Excellent conditions.
- 60–79: Good chance.
- 40–59: Worth a try.
- 20–39: Low chance.
- 0–19: Unlikely.
The night summary's "best window" covers hours scoring 40 or more.
Confidence
- High (0–1 h): based on measured solar wind and OVATION.
- Medium (1–24 h): Kp and cloud forecasts.
- Low (24–72 h): CME arrival times are uncertain; read it as a scenario.
We lower confidence when OVATION data are stale or cloud data are missing.
The limiting factor
For each hour we show which of A, C, D, M or L is lowest: "clouds", "daylight", "moon", "light pollution" or "aurora activity". It tells you at a glance why the score is low. If clouds are limiting, moving helps; if activity is limiting, you wait.
Why no 27-day night score?
NOAA publishes a weekly 27-day geomagnetic outlook. The Sun rotates about every 27 days, so fast streams from coronal holes can recur. But:
- cloud forecasts are not reliable night-by-night beyond a few days;
- CMEs cannot be predicted in advance.
Saying "63% tonight" for a night four weeks away would mislead. For long-range planning we show only a general activity trend, not a night score.
Why "0%" instead of a blank?
A common complaint about aurora apps is a dash or blank where a number should be: was it not calculated, or is it zero? We show a number for every hour we compute. 0% means "we calculated it and there is no chance" (for example, daylight). If data are actually missing, we say so explicitly and show the source status.
Data sources
- NOAA SWPC: OVATION aurora model, observed and forecast Kp, real-time solar wind. Since 10 June 2026 L1 data also come from NOAA's SOLAR-1 (NOAA).
- Open-Meteo and MET Norway: low, mid and high cloud forecasts (CC BY 4.0).
- Sun and Moon positions: computed on our server with astronomical formulas.
The model is not perfect. Your sighting reports, both "seen" and "not seen", help us calibrate it over time.