Method: how Aurorix turns space weather into a viewing forecast

Every number on Aurorix, where it comes from and the rules we add: NOAA Kp and OVATION, MET Norway cloud cover, geomagnetic latitude and the darkness limit.

Diese Seite ist auf Englisch geschrieben.

What each number is

Kp index now: the latest one-minute estimated Kp from NOAA SWPC (planetary_k_index_1m). It is a nowcast estimate, so it can still move before the three-hour value is final.
Kp chart: NOAA SWPC's Kp series (noaa-planetary-k-index-forecast), which joins observed values, the estimate for the current block and NOAA's own prediction for the next three days.
Aurora chance: the OVATION value for the one-degree grid cell nearest to the place, from NOAA SWPC's 30-minute aurora forecast.
Cloud cover: total cloud area fraction from MET Norway's Locationforecast 2.0, for the place coordinates.
Sun altitude: calculated in your browser with the NOAA Global Monitoring Laboratory solar equations.

The OVATION chance

OVATION (Oval Variation, Assessment, Tracking, Intensity, and Online Nowcasting) is an empirical model developed at the Johns Hopkins University Applied Physics Laboratory. It takes the solar wind speed and the interplanetary magnetic field measured at the L1 point, about 1.6 million km from Earth, and calculates the electron and proton precipitation that causes aurora.

NOAA derives an aurora viewing probability by assuming a linear relationship with the modelled intensity, which is why Aurorix reads the grid value as a percentage. The forecast lead time is the 30 to 90 minutes the solar wind needs to travel from L1 to Earth. When the solar wind data are unusable, NOAA drives the model with the current Kp instead and the lead time is zero.

The value is for the sky directly above the grid cell. NOAA notes that aurora can be seen from as much as 1,000 km away when it is bright and conditions are right, so a low value does not rule out a low glow on the northern horizon. The value also says nothing about clouds or daylight: the model has no weather and no sun.

Geomagnetic latitude and the Kp a place needs

Aurora follows the Earth's magnetic field, not the geographic grid. Aurorix converts each place to geomagnetic latitude with the centered-dipole formula and the IGRF geomagnetic pole for 2026, 80.9 N and 72.8 W. NOAA gives a rule of thumb for the oval: its equatorward edge is near 66 degrees geomagnetic latitude at Kp 0, moves toward the equator by about 2 degrees for each Kp step, and reaches 48 degrees at Kp 9.

Aurorix rounds up to the first whole Kp at which that edge has reached the place. A place at 64 degrees needs Kp 1; a place at 56 degrees needs Kp 5. This is an average relationship. NOAA says it holds only approximately, and a dipole is a simplification: corrected geomagnetic coordinates can differ from the dipole value by several degrees in some regions, which is up to two Kp steps. Treat the figure as a guide to the right order of magnitude, not a threshold.

Dark enough

Aurorix calls the sky dark when the sun is more than 12 degrees below the horizon, which is the end of nautical twilight. Aurora can be seen in civil twilight when it is bright, but 12 degrees is a safer line for a forecast. The altitude is geometric and ignores refraction, which changes it by less than a degree near the horizon.

The darkness table on each destination page counts the hours of the 15th of every month with the sun below that limit. It is the ceiling before cloud and Kp are applied: in midsummer the northern places have no dark hours at all.

Cloud cover

MET Norway publishes Locationforecast 2.0 under the CC BY 4.0 licence. Aurorix stores the cloud_area_fraction for the next 72 hours, as a percentage of the sky, and holds each value until the next forecast step. The first days are hourly; later steps can be three or six hours apart.

Total cloud cover is a blunt measure: it does not say whether the gap is over the north, and thin cloud can still show bright aurora. Aurorix treats 40 percent or less as clear enough. That threshold is our own editorial rule, not a NOAA or MET value.

The hourly strip

Each cell is one hour, shown in the local time of the place. An hour is classified in this order:

Too light: the sun is above −12 degrees.
No forecast: a cloud or Kp value is missing for that hour.
Dark but cloudy: cloud cover is above 40 percent.
Dark and clear, Kp too low: the forecast Kp is below the Kp the place needs.
Dark, clear, Kp high enough: all three conditions hold.

How fresh the data are

A scheduled GitHub Actions job downloads the NOAA and MET Norway files every hour and commits them with the time they were fetched and the exact source URL. The site is rebuilt from the latest commit, so a page can be a few hours behind NOAA. The "Data fetched" line shows the fetch time and, in your browser, how long ago that was. After three hours the page says the snapshot is ageing; after 24 hours it warns that it may no longer match the sky.

If a download fails, the previous file is kept and nothing is invented. The value stays on the page with its original timestamp.

Limits you should know

The chance is not tonight's chance. OVATION looks 30 to 90 minutes ahead; the Kp forecast looks up to three days ahead and has far less skill at the long end.
Moonlight, light pollution, mountains and your exact spot are not modelled.
Cloud cover is for the place centre. Weather can be very different 30 km away, which is often why guides suggest driving to a clearer valley.
Geomagnetic latitude is a dipole approximation. Longitude matters too: the oval is not centred on the geomagnetic pole at every hour, and substorm activity moves aurora north and south of the average position.
Only six northern-hemisphere places are covered. The southern hemisphere has its own OVATION grid, which Aurorix does not read.

Tests

The geomagnetic pole maths, the Kp-to-latitude rule, the NOAA storm scale, the solar altitude equations, the OVATION grid lookup and the snapshot provenance are covered by unit tests. Each test names the NOAA or IGRF page it is checked against.