Every autumn the same worries come up about Tromsø: is the Kp high enough, is the moon too bright, which week is best. Here is what the numbers say for Tromsø itself — what actually decides a night there, and why it is rarely the Kp.
Each figure names its source and its window: GFZ’s definitive Kp record, ten years of ERA5 cloud, and the sky’s own clock for the 2026–27 season — the same numbers the live Tromsø page reads every night. Between them, frames from the archive’s Troms nights, on Senja, at Elgsnes and over Lyngenfjord, each with the Kp of its hour. The archive holds frames from 11 winters since 2013, every one between February and early April, 9 of them in Northern Norway.
I
It is rarely the Kp.
GFZ · 14 October 2025 – 13 October 2026361/365 days
361/365
days at Kp 1+ — a glow low in the north at Tromsø
225/365
days at Kp 3+ — aurora overhead at Tromsø
GFZ definitive Kp, each day’s highest 3-hour value, 14 October 2025 – 13 October 2026 (365 days). Tromsø’s thresholds — Kp 1 for a glow, Kp 3 overhead, at 67° geomagnetic — are the site’s own, from a dark spot.
Tromsø sits at about 67° geomagnetic latitude, under the auroral oval. Around Kp 1 is enough for a glow low in the north; around Kp 3 puts the aurora overhead. Over the last twelve months the planetary Kp reached 1 or more on 361 of 365 days, and 3 or more on 225. Kp is also a three-hour planetary average — one number for the whole Earth — so don’t stay in on a clear night because an app says “Kp 2”.
01Elgsnes · 24 March 2017 · 23:43 · 10mm · f/2.8 · 30 s · ISO 1600 · Kp 0.33 (GFZ, 21–24 UTC)
02Bergsbotn · 23 February 2018 · 23:21 · 10mm · f/2.8 · 10 s · ISO 1600 · Kp 2.67 (GFZ, 21–24 UTC)
The frame at the top of this page was made at Straumsnes, on Senja, at 21:54 on 26 February 2018; the archive files that hour at Kp 0.33. Elgsnes, north of Harstad, was Kp 0.33 too. The book’s Lyngenfjord frame, with the curtain overhead, was made at Kp 2.67.
Mean cloud cover by calendar month, ERA5 reanalysis (Open-Meteo), 2015–2024, on a ~25 km grid around Tromsø. A regional mean: it can’t see a single valley, and it can’t see tonight.
Tromsø is coastal, and its weather comes off the sea. On ten years of reanalysis data the area averages 80% cloud cover in October, 80% in January and 80% in March, and 86–88% in November and December. Averages hide the gaps, and the gaps are what you are hunting — but it is why one-night trips strike out so often.
03Elgsnes · 24 March 2017 · cloud over the fjord · Kp 0.33 (GFZ, 21–24 UTC)
Inland, behind the mountains, the sky is often different. Not on average: ERA5 puts Skibotn at 87% in October, and Kilpisjärvi, on the Finnish border, at 90%. But an average is not the night you are out. When a front sits on the coast, the mountains often hold it on the sea side, and that is where the chase tours head on bad nights. If the coast is clear, nobody needs to drive anywhere.
04Lyngenfjord, just past Skibotn on the E6 · night of 4 March 2015 · Kp 2.67 (GFZ, 21–24 UTC)
III
The dark, and the moon.
Tromsø · 2026–27 · the sun’s own clock1 September – 10 April
1821000306
15 Sep
21:48–03:35
15 Oct
19:15–05:48
15 Nov
16:25–06:35
15 Dec
15:36–07:42
15 Jan
16:19–07:25
15 Feb
17:59–05:52
15 Mar
19:52–03:48
dark enough for a bright display (sun 12° down)fully dark (18°)
Computed for Tromsø (69.65° N) with the site’s own solar model (NOAA’s algorithm), in local time, noon to noon, on the 15th of each month.
Darkness sets the season. In 2026–27 it is dark enough for a bright display from 1 September to 10 April, and fully dark from 17 September to 25 March. Polar night, about 28 November to 14 January, means no sunrise, not twenty-four hours of dark: there is blue twilight around midday, and in mid-December it is dark enough to watch from 15:36 to 07:42.
05Lyngenfjord · night of 4 March 2015 · moon 99% lit · Kp 2.67
The moon matters less than people think. A full moon washes out faint aurora, not a strong one: the Lyngenfjord frames were made the night before the full moon, with the moon 99% lit, at Kp 2.67. If your dates are flexible, this season’s new moons fall on:
10 Oct 2026
9 Nov 2026
9 Dec 2026
7 Jan 2027
6 Feb 2027
8 Mar 2027
New moons in Tromsø’s local time, from Meeus’ lunar-phase algorithm (it puts the 12 August 2026 and 6 February 2027 solar eclipses on their days).
More nights beat a perfect night. A front can sit on the coast for two or three days, so four or five nights give you several separate shots at a gap. If you can, keep your tour night movable and pick it two or three days out, when the cloud forecast starts to mean something.
Be a good guest
If you haven’t driven on ice in the dark, don’t start here. Take the bus to a darker spot outside the centre, or go with a licensed tour.
Never stop on the road for the lights. Use a proper lay-by, with parking lights on, not hazards.
Find the gap. Wait for the dark. Stay enough nights.
In Tromsø the Kp is usually already there. The live Tromsø page does this sum every night — the dark window, NOAA’s Kp forecast, cloud from Open-Meteo and the moon, in one go or no-go — with the next seven nights (Kp and cloud only about three nights out, because nothing forecasts further), and, when Tromsø is clouded out, whether any nearby point is clearer. It is odds, not a guarantee, and it can’t see through cloud any better than Yr can.
Tonight in Tromsø · live
Maybe. It's borderline.
dark 20:36–04:28 · cloud 61% · forecast peak Kp 4.7
