✦ Aurora Forecast AI

How to Read Our Aurora Forecast: Score, Kp and Clouds

· Founder and developer of Aurora Forecast AI

Last updated:

On this page
  1. What the Aurora Score and its label mean
  2. The forecast card: status, summary and recommendation
  3. Best Viewing Window and Direction to Look
  4. Reading the hour-by-hour chart
  5. "Why this forecast?": score breakdown and factors
  6. The 3-Day Aurora Forecast and Hours of darkness
  7. Reading a country or region page instead of a city
  8. Reading the home page before you pick a city
  9. Three ways to turn this into tonight's decision
  10. What this page doesn't tell you

Every location page turns five inputs — Kp, NOAA's aurora model, cloud cover, darkness and moonlight — into one Aurora Score with a plain-language label. Read the label first, then the Best Viewing Window and hourly chart for timing, and the country or region table if your city's odds are poor. That order answers go, where, and when.

What the Aurora Score and its label mean

On a city page — Tromsø's, for example — the Tonight's Forecast card shows a score from 0 to 100 next to a status label — Unlikely, Low, Moderate, High or Very High. The label is the fast read: 0–19 Unlikely, 20–44 Low, 45–64 Moderate, 65–84 High, 85–100 Very High. It already accounts for Kp, NOAA's aurora model, cloud cover, darkness and moonlight together, weighted for that location's geomagnetic latitude. It is not a probability — a high score means most of the five inputs line up, not that a display is guaranteed. Exactly how those five inputs turn into the number — the weights, the darkness and season gates, and what happens when data is missing — is the methodology guide; this page only covers reading the result.

The forecast card: status, summary and recommendation

Under the score, the card prints a one-sentence summary and, for most statuses, a short explanation. At the bottom, a recommendation box tells you directly what to do. Each tile in the metrics grid — Max Kp and Cloud Cover are always there, with Moon and OVATION (NOAA's aurora-probability model) tiles when the location has that data — carries a small dot in the status colour. Many tiles also carry a "?" that opens a plain-language explanation of that reading; a tile without one simply has nothing to add beyond what's already printed, by design.

Best Viewing Window and Direction to Look

Two cards sit under the metrics. Best Viewing Window is the two-to-four-hour stretch tonight when the hourly score is highest; if no hour gets dark enough, the card says so instead of showing a time. Direction to Look is a small compass with Min Kp and Lat chips beside it. In the Northern Hemisphere it normally points north — south for aurora australis locations in the Southern Hemisphere — and at high latitude or inside the auroral oval it can read "Overhead or north" instead, because a strong display there can pass straight overhead. The "Lat" chip is the location's geographic latitude — the one on a map. It is not the number that sets Min Kp: that is the location's geomagnetic latitude, printed in the "where the odds are best" table on country and region pages, and what Kp means at your latitude explains why it decides the threshold. Once you know when and where to look, framing a shot is a separate skill — see northern lights phone photography.

Reading the hour-by-hour chart

The Hour-by-hour visibility chart draws one bar per hour from 22:00 to 04:00, taller where that hour's score is higher; the bars inside the Best Viewing Window are outlined and kept in colour while the rest of the chart dims. Opening a bar's "?" shows that hour's own Kp and cloud cover, plus the night's moon and OVATION readings, which are held constant from hour to hour — the same four numbers behind the nightly score. A short dip surrounded by taller bars is usually a cloudier hour, not a drop in geomagnetic activity — check that hour's own numbers before assuming the storm faded.

"Why this forecast?": score breakdown and factors

Below the chart, Why this forecast? is open by default on a city page and collapsed on the home page. Inside, the Score breakdown lists five rows — Geomagnetic activity, Aurora oval probability, Cloud cover, Darkness, Moonlight — each with a bar and a number, plus two summary lines (Aurora activity, Sky visibility) and the final Aurora Score. Activity and Visibility are explanations, not extra inputs — the Aurora Score itself is never built from them. Underneath, factor cards explain why a row landed where it did: an Aurora zone, Cloud cover or Darkness card each carries a short explanation and, on a "?", the full level legend. A Data sources list at the bottom links to NOAA and Open-Meteo directly. For the exact weights and formula behind these rows, see the methodology guide — this page stops at what each row means.

The 3-Day Aurora Forecast and Hours of darkness

Most city pages carry a 3-Day Aurora Forecast for {location} with up to three cards — Tonight, Tomorrow night, In 2 nights — each with its own Max Kp, Cloud Cover and Moon figures and a Best Viewing Window line; the block needs at least two nights of data to appear at all. Only the Kp and cloud numbers for the second and third night are real forecasts — the OVATION figure always reflects the latest snapshot, so tomorrow's score is softer than tonight's.

Further down, Hours of darkness in {location} lays out the twelve months as columns, each with that month's hours of astronomical darkness and, beneath the table, any span of polar night, midnight sun or nights that never get properly dark. This table answers a different question than tonight's score: an "Unlikely" status in June at a high-latitude city is usually the season, not a quiet geomagnetic night — the score's darkness gate takes the result to zero when the sun never gets properly dark (midnight sun), and its season gate discounts the rest of the bright months on top of that, regardless of Kp. And a low Min Kp threshold on the page doesn't mean every night works: how Kp translates by latitude, and Iceland's own geomagnetic latitude is a good example of how far that number can sit from the figures usually quoted for a place.

Reading a country or region page instead of a city

A country or region page has no single score — it compares its own cities instead. The block at the top, titled with the area's name and tonight ("Norway tonight"), shows the status of the best-placed city plus one to three sentences built from the data. Below it, a region lists every city and a country its top five, each row carrying Best Viewing Window, Cloud Cover and Min Kp — the same figures a city page shows for itself. A region also gets its coming nights, each card naming that night's best-placed city.

Further down, a "Where the odds are best in …" table ranks every child place by the Kp it needs, with columns Place, Min Kp, Geomagnetic latitude, Season, Aurora zone, and a region also gets a summary of what its cities span, in the shape of a "{Region} in numbers" block. On the USA page, that table is the honest answer to "can I see it in my state": Alaska sits near the top of it, and states further south sit near the bottom needing a real storm. Reading the table beats guessing from a map.

Reading the home page before you pick a city

Without a city in mind, the home page opens with Global Aurora Activity — a planet-wide status label and score, the current Kp and the maximum forecast for the next 24 hours, and a storm-level tile once NOAA has issued one — followed by its own Global Kp Forecast timeline. Below that, Best Chances Tonight links to locations where conditions are realistic tonight; each card carries a score and, when something is holding it back, a named main limiting factor — so you know which factor to check before writing off a whole region. On a quiet night the block never disappears: it states the reason, then shows popular destinations instead. Popular Aurora Forecasts below groups more locations by area, each with today's chance label.

Three ways to turn this into tonight's decision

A high Kp with a cloud problem. Suppose a city's status reads "Moderate," not "High," even though the Max Kp tile looks strong. Open Why this forecast? and check the Cloud cover row and its factor card — the card's own explanation (clear, partly cloudy, or overcast) will usually say clouds are the limit, not activity. On the region or country page, the "Where the odds are best" table can show a place with the same Min Kp and a clearer sky nearby — that's the "where."

A modest score with a real window. Suppose the status reads "Low" at 8pm, but the Best Viewing Window shows 23:00–02:00 and the hour-by-hour chart's tallest, outlined bars sit inside it. The card up top describes the whole night on average, not the best hour in it — the "when" is to check back, or be outside, during the window rather than the moment you happened to look.

An "Unlikely" that's the season, not a quiet night. Suppose a high-latitude city reads "Unlikely" on a June night with an entirely ordinary Kp forecast. Before assuming nothing is happening, check Hours of darkness for that month: at high latitude the summer sky may never get dark enough, and then the score's darkness gate takes the whole result to zero regardless of activity; the season gate discounts the remaining bright weeks. The same city with the same quiet Kp in October is a genuinely quiet night — the darkness table is what tells the two apart.

What this page doesn't tell you

This page describes the blocks, not the arithmetic behind them or the physics behind Kp. The exact weights, gates, refresh times and known limits — including that tomorrow's score is softer than tonight's, that cloud cover is a model forecast rather than a look at your actual sky, and that the moon's position above the horizon isn't yet accounted for — are in the methodology guide. What Kp physically measures, and why the same Kp means different things at different latitudes, is the Kp index guide. Neither the score nor this page is a promise: a high score means the conditions line up, not that aurora is guaranteed, and a quiet forecast can still surprise you. This guide, the methodology page and the Kp index guide are three of the site's guides — start there for the full list.

More guides