The Northern Lights Guide

Everyone in this business explains how the aurora works. Almost nobody publishes how likely you are to see it in a particular place, in a particular month, using a number you can check.

That gap is what this guide is for. The physics is here because you need some of it to read a forecast, but the pages that matter most are the ones with data behind them, and the data is public, free and decades old. It has simply never been convenient for anyone selling tours to put it on a page.

Across eleven seasons of reanalysis, this is how often the sky is actually clear enough to see an aurora, counted as nights with at least two hours below 50% cloud cover between 18:00 and 02:00.

Three things fall out of that table, and none of them appear in the standard advice.

December in Tromso is the worst month of the season, at 16.8% of nights, and it is peak season. The universal recommendation to travel in midwinter looks at darkness and ignores cloud. During the polar night the darkness is at its theoretical maximum and the sky is at its annual worst. The hours were never the problem.

September and March beat midwinter nearly everywhere, and cost less. There is a physical reason as well as a meteorological one: near the equinoxes, geomagnetic storms are close to twice as common as at the solstices.

Reykjavik is the consistent one, holding between 31% and 41% all season while the two Arctic destinations swing by more than double. If you have a single night booked months in advance, that consistency is worth more than a high ceiling.

One operator in Tromso advertises 60 to 70% cloud-free nights between late November and mid February. For that same period the reanalysis gives between 16.8% and 23.5%. A factor of about four.

We do not think that is dishonesty. They are almost certainly counting something different, over a period they do not state. Which is the whole problem: a percentage with no stated metric and no stated source cannot be checked by anybody.

So the rule we hold ourselves to is that every figure on this site carries three things: where it came from, exactly what was counted, and what it does not tell you. The thresholds in that table, the 50% cloud and the two hours, are our choices rather than natural facts, and we say so on every page that uses them.

The same rule is why there are no prices anywhere in this guide yet. We can see that identical-looking tours in one city range severalfold in price, and a sample of ten listings is not enough to publish a median from. When the catalogue supports a proper calculation, it will appear with the sample size and the date attached. Until then, saying nothing is more useful than saying something unverifiable.

1. Book more than one night. At Tromso's December rate of roughly one usable night in six, a single booking is a coin toss weighted against you, and no guarantee wording changes the arithmetic.
2. Prefer a tour that can move. A chase that will drive three hours to find a gap in the cloud does more for your odds than any promise in a contract.
3. Consider the shoulder season. September, October, February and March generally have better sky than midwinter, cost less, and sit closer to the equinoxes.