Updated on October 5, 2026
What Kp is
Kp is a number from 0 to 9 that describes how disturbed the Earth's magnetic field is. NOAA's Space Weather Prediction Center (SWPC) says it "quantifies disturbances in the horizontal component of earth's magnetic field". It is one planetary value, published in 3-hour blocks. It is not a local reading.
The scale
- Kp 0 to 2: calm. SWPC describes the aurora as far north and quite dim.
- Kp 3 to 5: the aurora moves further from the poles and gets brighter.
- Kp 5 and above is a geomagnetic storm. Kp 5 is rated G1, Kp 6 is G2, and so on up to Kp 9, which is G5.
- Kp 6 to 7: aurora may be visible from the northern edge of the United States.
- Kp 8 to 9: overhead in northern US states.
Thresholds such as Kp 5 or 6 belong to those lower latitudes. They do not apply to the places on this site.
What Kp you need at our twelve places
The aurora forms an oval around the magnetic pole. The University of Alaska Fairbanks puts its average position at around 67 to 68 degrees of geomagnetic latitude. Our twelve places sit at that band or a few degrees south of it, so low values are normally enough.
| Place | Geomagnetic latitude (approx.) | Aurora overhead typically from |
|---|---|---|
| Yellowknife | 68.8 | Kp 1 |
| Tromsø | 66.6 | Kp 1 |
| Abisko | 65.3 | Kp 2 |
| Akureyri | 65.3 | Kp 2 |
| Saariselkä | 65.0 | Kp 2 |
| Fairbanks | 64.9 | Kp 2 |
| Kiruna | 64.7 | Kp 2 |
| Levi | 64.5 | Kp 2 |
| Reykjavík | 64.0 | Kp 2 |
| Whitehorse | 63.5 | Kp 3 |
| Rovaniemi | 63.0 | Kp 3 |
| Luleå | 62.2 | Kp 3 |
Treat these as approximate. We found no official per-city table. Most of the latitudes come from published tables for magnetometer stations in or near each town, and where no station is close we estimated from the nearest ones. The figures for Fairbanks and Yellowknife come from a third-party site that does not state which magnetic field model or year it used. The Kp values are our own derivation from the latitudes, and several places sit near a boundary: Saariselkä and Akureyri could be read as Kp 1 to 2, Whitehorse and Luleå as Kp 2 to 3. The detail is on the methodology page.
The practical reading is that Rovaniemi, Whitehorse and Luleå need a somewhat more active night than the others, and Yellowknife and Tromsø need the least.
Why cloud beats Kp
Kp 7 under a solid overcast is worth nothing. You cannot see through cloud, and the storm does not wait for it to clear.
The reverse case is the common one. At high latitude, a clear dark sky at Kp 1 or 2 often still shows aurora, because you are standing under the oval and do not need it to expand. The Finnish Meteorological Institute says that in Lapland auroras are "quite common always, even during solar minimum".
So we rank places by sky first. In February, 72% of nights in Yellowknife had a usable sky (at least two dark hours in a row with half the sky or more free of cloud) against 19% in Tromsø, though both typically need only Kp 1. That gap matters more than any difference in Kp. The figures are on the twelve places compared.
Forecasts worth reading
- The 3-day forecast from SWPC gives the expected Kp for each 3-hour block. On 4 October 2026 it showed a greatest expected Kp of 5 (G1) for 4 to 6 October.
- The 27-day outlook, issued on Mondays, gives the largest expected daily Kp. It relies on features of the sun coming back around as it rotates, so it has little skill for one particular night.
- OVATION, also from SWPC, is "a 30 to 90 minute forecast of the location and intensity of the aurora". It is the one to check when you are already dressed.
Our Tonight page puts the NOAA Kp forecast next to an hourly cloud forecast for each place and compares Kp with the threshold in the table above. SWPC puts the best viewing "within an hour or two of midnight".
Solar cycle 25 and winter 2026/27
The sun's activity rises and falls in cycles. According to SILSO at the Royal Observatory of Belgium, which keeps the international sunspot number, the smoothed monthly number for the current cycle peaked at 161 in October 2024. We did not find a formal NOAA statement declaring the maximum.
SWPC's prediction for winter 2026/27 is a smoothed sunspot number of about 88 from October to January (range roughly 70 to 100) and about 86 by March 2027. That is roughly half the peak. The cycle is clearly declining but far from its minimum, which the same prediction puts at about 8 by December 2030.
In plain terms, expect fewer big storms than in 2024. That matters most to people hoping to see aurora from much further south, such as the northern United States. Under the oval it matters much less, as the FMI line above says. The G1 forecast of 4 October 2026 was attributed to a coronal hole, a kind of source often described as typical of the declining phase, though we could not confirm that outlook in SWPC's own wording.
For choosing a month, see when to go. For why we put so little weight on promises, see how we calculate the numbers.