Methodology

Live score (0–100)

For every place and every hour of the next three days we multiply four factors: geomagnetic activity × clear sky × darkness × moonlight.

Activity compares NOAA's Kp forecast for that three-hour block with the Kp the place needs, using a smooth curve. For the current hour we also use NOAA's OVATION nowcast for the place and the area towards the pole. Clear sky is one minus MET Norway's forecast cloud cover. Darkness is full when the Sun is more than 12° below the horizon, partial in nautical twilight and zero in daylight. A bright Moon above the horizon lowers the score by up to 30%.

The score ranks conditions; it is not a probability. Fog, local cloud and lights also matter.

Magnetic latitude

We trace the Earth's magnetic field line (IGRF-14 model, epoch 2025) from the ground to its highest point to get each place's magnetic latitude. Values were computed on a 1° global grid and interpolated; they agree within about 1° with the AACGM coordinates used in auroral science, which we checked against magnetometer stations from Tromsø to Macquarie Island.

Where to look and how high

The equatorward edge of the night-side auroral oval sits near 66.5° magnetic latitude at Kp 0 and moves about 2° towards the equator per Kp step. From the distance between a place and that edge we compute the angle of the aurora's lower border (about 105 km up) and ray tops (about 200 km) above the horizon.

A display counts as visible when its lower border is at least 9.7° above the horizon. That matches the long-used rule that aurora can be seen about 4.5° of magnetic latitude beyond the oval edge. The direction to look is the direction in which magnetic latitude increases fastest.

Terrain

For places with at least 1,500 inhabitants and all featured destinations we sample elevation data (AWS Terrain Tiles) along seven lines of sight, 30° either side of the look direction, out to 30 km, correcting for Earth curvature and standard refraction. If hills rise higher than 6.7°, the visibility threshold is raised accordingly — which is why some fjord and valley towns need a higher Kp than their latitude suggests.

Light pollution

We sum night-time light from NASA's VIIRS composite (2016) within 50 km, weighted by distance to the power −2.5, and group the result into six classes calibrated on known places. For brighter places we also search rings 5–40 km out for land with at least ten times less sky glow. These are estimates: they do not know about roads, private land or a single bright lamp next to you.

21 years of nightly odds

For every 1° grid cell we examined each night from 1 January 2005 to 31 December 2025. A night is dark if the Sun is more than 12° below the horizon in at least one three-hour block; active if GFZ Kp in a dark block reached the place's threshold; clear if NASA POWER's night-time cloud amount (CERES SYN1deg) was 70% or less. A good night meets all three.

We also split the odds into years near solar maximum (2005, 2011–2015, 2022–2025) and minimum (2007–2010, 2018–2020). Satellite cloud data covers a whole grid cell and the whole night, so partly cloudy nights can still offer clear spells, and local microclimates are only partly captured.

Storm visibility

For each severe storm since 1989 we list the places whose Kp threshold was reached during a three-hour block in which the Sun was more than 8° below their horizon. Bright storm aurora is visible in nautical twilight, so this limit is looser than the 12° used for the monthly odds. Weather is not included.

Which places get a page

Towns with at least 1,000 inhabitants (GeoNames) where the aurora appears on the horizon at about Kp 6 or less; larger cities in regions where it needs a storm; and capitals or very large cities further from the pole, where people often ask. Smaller places without much demand are kept out of search engines until there is a reason to include them.

Eclipse calculations

Contact times, totality, coverage and Sun altitude come from the astronomy-engine library for each city centre; they usually agree with published predictions within a few seconds. Weather odds use NASA POWER daytime cloud and maximum temperature for 26 July to 9 August, 2005–2025.

Independence

Affiliate partnerships never change scores, rankings or data.