Northern lights cameras from Abisko to Fairbanks, hour by hour

The camera at Cleary Summit sits at 65.05°N on the Steese Highway north of Fairbanks, looks away from town down the far side of the pass, and replaces its frame every thirty seconds. Alaska DOT put it there to show whether the road is drifted over. It happens to point at the same patch of sky that people drive out of Fairbanks at midnight to stand under, and it is free, and it is already live.

Almost every list of northern lights cameras is a list of cameras built for the job: the Lights Over Lapland stills at Abisko, the AuroraMAX all-sky at Yellowknife, the university rig at Fairbanks. Those are the right cameras, and on a good night they are better than anything below. They are also about a dozen cameras for a band that runs most of the way round the planet, and they are all pointed at their own piece of it.

The cameras that are not built for the job are everywhere, and there are a lot of them. Counted from our camera table at 01:05 UTC on 2026-10-10, Sweden has 1,729 live cameras, Canada 5,569 and Iceland 553, with Norway at 1,511 and Alaska a slice of the 60,312 in the United States; the method behind those counts is on the sources page. A fair few of them sit above 62°N with nothing but tundra in the frame.

Which hour, and therefore which camera

The useful thing about the band is that it hands darkness off westward, and the cameras on it are in six time zones.

For 2026-10-12, taking astronomical twilight from the sunrise-sunset API at each camera's own coordinates: Abisko is properly dark from 18:34 UTC until 02:27, Reykjavík from 20:59 to 05:29, Yellowknife from 03:02 to 11:45, and Fairbanks from 05:23 to 13:51. Those windows overlap end to end. Somewhere on the band a camera is in astronomical darkness continuously from 18:34 UTC through 13:51 the following afternoon, which is nineteen hours and seventeen minutes, and whatever hour you are awake in, one of them is yours.

The four and three quarter hours from 13:51 to 18:34 UTC are the hole. Nothing north of 60° is dark in that stretch, and no camera fixes that.

The number that decides whether to bother

Space weather is forecast, and the forecast is short and free. The NOAA Space Weather Prediction Center three day outlook issued at 00:30 UTC on 2026-10-10 puts the greatest expected 3 hr Kp for 10 to 12 October at 5.67, a G2 on the NOAA scale, in the 03:00 to 06:00 UTC slot on 12 October, with Kp 5.33 either side of it. Several arriving CMEs, with a coronal hole high speed stream behind them, is the rationale it gives. Quiet until then, though: the greatest observed Kp in the preceding 24 hours was 4, and the observed series has it reaching 4.0 at 09:00 UTC on 2026-10-09 and back to 2.33 by 21:00.

Kp is a latitude number more than a brightness number. NOAA's own viewing note puts the equatorward edge of the auroral oval at about 66 degrees magnetic latitude when Kp is 0, moving roughly two degrees equatorward per level of Kp, and adds that a person can see aurora even when it is 1000 km further north. So a forecast of 5.67 is not a promise of anything overhead at Fairbanks. It is a reasonable bet that the oval is low enough and bright enough for a thirty second JPEG with a road in the bottom of it to come back green.

Cross that forecast with the windows above and the answer for 12 October is the North American half. At 03:00 UTC Abisko has been light for half an hour, Reykjavík has two and a half hours left, Yellowknife has just gone dark and Fairbanks goes dark at 05:23 inside the same G2 slot.

The cameras, by where they are

Yellowknife is the one to open first, because it sits nearly under the oval rather than south of it. The Bush Pilots Monument camera at 62.4656°N looks east over the old town and Yellowknife Bay with very little in front of it, and refreshes every 150 seconds; the bay camera a few hundred metres away gives you water to catch it. Of the 35 live cameras we carry in the Northwest Territories, the highway ones are the dark sky bet: camera 31307 is on the road at 62.4714°N and two more sit down at 60.77 and 60.56, five minutes between frames, no streetlight anywhere.

Fairbanks is the deepest set, because Alaska DOT runs a thirty second camera at every pass that matters. Chena Hot Springs Road at 64.8844°N points up the road towards the lodge that exists because of this, and Nenana Hills on the Parks Highway at 64.7076°N is 40 km from the nearest town glow. Cleary Summit is the one I would leave open. For a slower, prettier frame, the Chena River camera looks north from the city at 150 seconds.

Abisko earns its reputation on weather rather than latitude: it sits in a rain shadow and is clear when Tromsø is not. The windy camera above Abisko at 68.3504°N refreshes every 150 seconds, and the Björkliden hotel camera eight kilometres west looks out over the lake at thirty seconds. Further east, Trafikverket's road cameras are the dark ones: Junosuando at 67.4238°N and Kiruna itself, both at sixty seconds, and Kebnekaise up at 67.8930°N looking at the mountain.

Iceland is the hinge between the two halves of the night. Mount Esjan from Reykjavík at 64.1258°N looks north across the bay at the mountain, 150 seconds, with the whole city behind the lens rather than in front of it, and Akureyri's fjord camera at 65.6792°N is thirty seconds and a lot further from any light.

For Tromsø and the Norwegian cameras I wrote them up separately a day ago, in the far-north Norway post, and the Fjellheisen camera above the city is the one worth having open from that list.

What a road camera will and will not give you

Exposure is set for the tarmac. A highway camera meters for a lit road sign and a number plate, so a Kp 3 arc that a dark adapted eye would enjoy will not register at all, and what does register is the bright green band low in the north on a G1 or better. The compensation is refresh rate and count: thirty seconds across dozens of cameras finds a clear patch of sky faster than one beautiful all-sky stream can.

We are going to compute each camera's own dark window from its coordinates and show it on the camera page, so a page can say when it is worth opening instead of leaving you to work out what time it is at 65°N. The coordinates and the refresh rate are both already in the row.

The open question is moonlight, which NOAA's note is blunt about and we do nothing with. A full moon at 65°N in October sits high and stays up most of the night, and it will flatten a weak display on a camera long before it bothers an eye. If there is a public source for lunar altitude by coordinate that I can fetch per camera in one call, I would like the link.