About Aurora Forecast

Aurora Forecast (tonightaurora.com) answers one question — "can I see the aurora from where I am tonight?" — in the fewest possible seconds. Every city page computes a live, weighted answer from primary scientific data sources, states its reasoning in plain language, and stays honest when the answer is probably not. It is built and operated by an independent developer with a background in product engineering — every threshold, page and line of code is ours.

Why we built this

Most aurora apps do too much and answer too little. They show you a heatmap, a Kp graph, a cloud chart, a moon widget, a notification bell, and a "subscribe for $4.99" button. The thing they don't do is tell you whether to bother going outside tonight.

We started Aurora Forecast because people we knew were tired of opening four apps and still guessing — and tired of forecasts that inflated optimism to keep them engaged. The difference here is a single weighted score and a one-word answer (GO / MAYBE / LOW / NO) backed by a transparent five-factor breakdown: Kp vs your city's threshold, night-average cloud cover, moon illumination, the astronomical darkness window, and light pollution. Every factor card shows its number and its weight, so when the verdict is MAYBE you can see exactly which factor is holding the night back.

What we will not do

• Sell an account, tier, or feature behind a paywall.
• Send push notifications — web push apps get uninstalled for noise, and we keep no relationship to sell.
• Track you across the web. No accounts, no email wall, no advertising profile of our own, and we never sell personal data. Anonymous analytics and Google-served ads are the only tracking on the site — both are documented in full on the privacy page.
• Publish a "definitive" 14-day aurora forecast — that would be snake oil, and NOAA's own guidance agrees.
• Show a false GO to keep you coming back. A LOW night is labelled LOW.

Where the data comes from

Every number on this site is sourced from a primary science agency or an open data feed, computed client-side the moment you open a page:

• NOAA Space Weather Prediction Center — the 3-hourly Kp forecast, the observed Kp feed, and the 27-day Space Weather Outlook.
• Open-Meteo — hourly cloud cover for each city's exact coordinates (free public API, no key required).
• Standard astronomical algorithms — sun altitude (for the darkness window) and lunar illumination, computed locally in your browser.
• Public light-pollution data — a pre-processed Bortle class per city.

The full weighting logic, hard rules and known limitations are documented on the methodology page — deliberately, so that anyone can check our work or reproduce it.

Editorial standards

City guides, trip pages and explainer articles are written by us, fact-checked against the same primary sources, and reviewed on a schedule: thresholds and city data are re-validated each season, and articles are updated when the underlying science or data feeds change. If you spot an error, the contact page is the fastest way to get it fixed. When we don't know something — substorm timing, for example, which no public feed provides at minute resolution — the pages say so instead of guessing.

How each city's threshold is set

The single most consequential number on this site is a city's Kp threshold, and it is the one most aurora guides get wrong by treating latitude as if it were the whole story. Geographic latitude and magnetic latitude are not the same thing: Earth's magnetic poles are tilted and offset from its rotational poles, so two cities at the same parallel can sit in noticeably different positions relative to the auroral oval.

Our thresholds are derived from each city's magnetic latitude — corrected using the International Geomagnetic Reference Field model rather than read off a map — and then banded: cities above roughly 67.5° magnetic need about Kp 1, those between roughly 64.5° and 67.5° need about Kp 2, and the requirement steps upward from there as you move toward the mid-latitudes. Local factors then adjust the expectation: a city's Bortle class describes the light pollution it starts from, and coastal versus inland geography changes how often the sky is actually clear.

This is also why the answer for two neighbouring cities can differ. Fairbanks and Anchorage are 500 km apart on the map; their magnetic latitudes place them in different bands, and their thresholds differ accordingly. We would rather publish a number that looks conservative than one that sends someone outside on a night when nothing was ever going to be visible.

Who this is for

Three groups, mostly. People who live at high latitudes and want to know whether tonight is worth the cold; travellers who have booked a trip north and want to know their odds before they arrive; and the mid-latitude majority who only hear about the aurora when a storm makes the news and want to know whether their own sky is in play. The second group is the reason every city page carries its own season dates and local viewing notes — the useful question is never "will there be aurora in Norway" but "will there be aurora where I am, on the nights I am there".

How the site is maintained

This is a working tool, not a published-once brochure. Changes are logged here so that returning readers can see what moved and when.

• September 2026 — Site launched with the first set of city pages, the five-factor scoring model and the methodology page.
• 23 September 2026 — Rebuilt the first screen: every city page now opens with the live verdict, the five-factor breakdown and a seven-night outlook, computed on load.
• 2 October 2026 — Added Churchill (Canada) and Akureyri (Iceland) after re-deriving their magnetic latitudes from the geomagnetic field model. The site now covers 55 cities.
• 4 October 2026 — Rewrote every city page: aurora-season dates are now computed from each city's own coordinates rather than quoted per region, each page carries a computed darkness profile, and local viewing guidance was written per city. Kp thresholds and scoring were not changed in this update.

If something on this site is wrong, the fastest route to a fix is email. Corrections are applied and logged.

Limitations, honestly stated

We do not account for: substorm onset at minute resolution (no public source exists), local weather effects finer than the cloud model's resolution — inversions and sea fog can flip conditions in 30 minutes — or CME impacts whose arrival time is still uncertain while en route. The MAYBE and LOW labels exist precisely because of these gaps. Cloud forecasts degrade quickly beyond 48 hours, and the 27-day outlook is a planning window, not a promise. If a verdict on this site was wrong for your sky last night, tell us — that is the kind of report we want most.

Contact

Everything routes through one address — [email protected] — answered by a human within five working days.