AURORA

How Aurora decides

Aurora scores every curated viewing spot near you for every half hour of tonight — location × time — and only then decides where to send you. A spot whose clear spell ends before you could get there doesn't count.

1 · Your chance of seeing the aurora

For every half hour, the chance is two gates multiplied — you need both, so one can't make up for the other:

1
A clear view of the skyLow and mid-level cloud blocks the aurora; broken cloud leaves gaps, which help most when the aurora is overhead. Thin high cloud only dims it, fog and rain obstruct, and a forecast is trusted a little less the further ahead it is.
2
Aurora bright enough to seeActivity comes from NOAA's real-time Kp for the next hour or two, then NOAA's 3-hour forecast (pulled towards a typical night, because Kp forecasts are only weakly skilful) or IMO's midnight forecast. An auroral-oval model places the oval for that activity and time of night, and each spot's geomagnetic latitude — about 63° on the south coast, 65.5° in the north — decides whether it is overhead or low in the north. The display's expected brightness is then compared with the faintest aurora your eyes can pick out against tonight's sky: about the Milky Way's brightness in a dark sky, roughly four times that at the end of twilight, and seven times under a high full moon.
  • The chance for your window allows for the fact that a quiet night tends to stay quiet: extra hours help, but less than a fresh throw of the dice. Uncertainty in the activity forecast is carried through five scenarios rather than ignored.
  • The model reproduces the Finnish Meteorological Institute's statistics for Lapland — how often aurora is seen on clear nights from 61° to 67° geomagnetic latitude — to within about ten points.
  • 70% or more is a very good chance, 45% good, 20% some chance. Chances are shown to the nearest 5% and are good to roughly ±10–15 points; phone cameras pick up fainter aurora than your eyes. Missing data never counts as bad data — without an activity forecast the chance assumes a typical night.
The science behind it
  • Auroral oval: Starkov (1994), coefficients from Sigernes et al. (2011), J. Space Weather Space Clim. 1, A03
  • Geomagnetic latitudes: NASA OMNIWeb corrected-geomagnetic model (IGRF, 2026)
  • Moonlight: Krisciunas & Schaefer (1991), PASP 103, 1033 · Twilight: Patat et al. (2006), A&A 455, 385
  • Visual threshold: Crumey (2014), MNRAS 442, 2600 · Brightness classes: International Brightness Coefficient
  • Calibration: Finnish Meteorological Institute all-sky-camera statistics (share of clear nights with aurora, 1973–1997) with the GFZ Kp record

2 · Recommendation: is it sensible to go?

  • Only the hours you can reach count: arrival = now + drive time + five minutes to park.
  • The best window is the longest steady stretch close to the night's reachable peak, not a single timestamp.
  • Leave-by time = window start − drive − a buffer (10 minutes or 15% of the drive).
  • Trip value starts from the chance and subtracts what the trip costs: about 8 points per hour of driving (more for drives beyond 1½ hours), and a few points for icy or unverified roads, winter-only access or strong wind. None of it changes the chance itself.
  • The setting counts too: a sighting above an iconic backdrop like Kirkjufell, Jökulsárlón or Vestrahorn is worth up to 10% more than above a plain lakeshore. It decides between similar chances, but never beats a clearly better one.
  • Closed or difficult roads, and storm-force gusts (30 m/s, about 108 km/h) at the destination, are a hard stop: the next safe alternative is recommended instead.

Please read

Aurora recommendations are probabilistic and do not guarantee visibility. Cloud forecasts can be wrong, auroral activity changes quickly, and Icelandic roads and weather can turn dangerous fast. Always check umferdin.is, vedur.is and safetravel.is before you go, and never stop on the road to look at the sky.

Data sources & attribution

Find tonight's best spot