← AuroraChaser

Data sources

These are all the data sources and services we use, with their licences and credits. Most of the data is open, and we are grateful to everyone who shares it. Our server also stores cloud observations from MET Norway (some from FMI and SMHI stations), road warnings from Statens vegvesen, disruption notices from Entur and the Hp30 index from GFZ Potsdam. We keep them only to test the forecasts and features we may add, and no screen or score uses them yet.

Sources and licences

SourceWhat we use it forLicence
NOAA Space Weather Prediction Center (SWPC)The OVATION aurora model, the Kp index, real-time solar wind, the 3-day and 27-day forecasts and hemispheric powerPublic domain (U.S. Government work)
FMI IMAGE magnetometersMagnetic activity at Kilpisjärvi, Masi and Kevo, shown as local aurora activity and part of the live score since app version 1.0.3CC BY 4.0
NASA DONKI (CCMC)Warnings of coronal mass ejections (CMEs) on their way to EarthPublic domain (NASA / CCMC)
MET Norway LocationforecastHourly weather and cloud forecasts for each viewing spotNLOD 2.0 / CC BY 4.0
MET Norway weather stations (Frost)The share of mostly clear night-time observations on our region pages, and the typical night cloud of each region that the app uses when it has no forecast, 2001–2020CC BY 4.0, based on data from MET Norway
CARTOThe map tiles in versions of the app before 1.1.0, and in later versions any tile our own server cannot deliver. The map's dark style in every version© CARTO, under the CARTO basemap terms. The style under the BSD 3-Clause licence
OpenMapTilesThe layout of our own map tiles, which we build from OpenStreetMap data for app version 1.1.0 and laterCC BY 4.0, © OpenMapTiles
Natural EarthCoastlines, lakes and borders on our own map tiles when you zoom far outPublic domain
OpenStreetMapMap data in the map tiles, and the roads behind the drive times from town to each spot. The drive times are estimates at about 80% of the speed limit, rounded up to 5 minutes© OpenStreetMap contributors, ODbL
Statens vegvesen NVDBThe national road database, to check that the parking at our viewing spots is public and kept open in winter, where it has records, and the convoy speed on the last stretch to North CapeNLOD 2.0 (credit below)
EnturPlanned bus times, to show which viewing spots you can reach by bus and leave again in the evening, and the ferry crossing times in the drive times to Kjerringøy and HavnnesNLOD 2.0 (contains data from Entur)
VIIRS Nighttime Lights (VNL)Satellite night-light data, from which we calculate each spot's light pollution (the artificial brightness of the northern sky, where faint aurora usually shows on quiet nights; on Svalbard, all around)CC BY 4.0 (credit below)
WorldPopOur copy of the VNL yearly composites, resampled and with gas flares removedCC BY 4.0
NASA Black MarbleNightly night-light data for Svalbard, where the VNL composites stop at 75° N, calibrated against VNL at mainland townsNASA open data (CC0)
Copernicus DEM GLO-90Terrain heights, to leave out light that mountains block© DLR e.V. and © Airbus Defence and Space GmbH, provided under COPERNICUS (credit below)
Statens vegvesen road camerasRoad cameras shown in the app, normally served through our own serverNLOD 2.0 (credit below)
KHO / UNISAll-sky and outdoor cameras at the Kjell Henriksen Observatory, Svalbard. Link only.© Kjell Henriksen Observatory / UNIS
PANOMAXThe Nordkapp panorama camera. Link only.© PANOMAX GmbH
yr.noThe Setermoen sky camera. Link only.© yr.no / contributors
Firebase Cloud Messaging (Google)Delivering our push notifications© Google

Webcam images belong to their operators. Webcams marked “link only” open on the operator's own website, and the app does not show their images. The app links to other operators in the same way, for example Kystnor, SkylineWebcams, harbour authorities, ski resorts and regional tourism sites.

Score calibration

We fitted the aurora score's numbers on these records. The app and this site contain only the fitted numbers.

SourceWhat we use it forLicence
FMI all-sky camera intensitiesCalibrated aurora brightness at Kilpisjärvi, Abisko and Sodankylä, behind how likely you are to see aurora and how bright it should beCC BY 4.0
THEMIS all-sky imagersKeograms from cameras in Canada and Alaska (University of Calgary and UC Berkeley), with UCalgary's cloud classification, to measure how much more often aurora shows somewhere in the sky than overhead, and how that changes with latitudeCC0 (NASA data; acknowledgement below)
FMI weather observationsCloud observations at the camera sites, for the cloud part of the score and the 14-day chance forecastCC BY 4.0
SMHI weather observationsCloud observations at Abisko and Katterjåkk, for the same fitsCC BY 4.0
GFZ Potsdam Kp indexThe Kp index we fitted the score on, also used each month to adjust the 27-day outlookCC BY 4.0
KHOnet2026, SvalbardAurora classifications from the Kjell Henriksen Observatory's camera, for the score on SvalbardCC BY 4.0
Ny-Ålesund all-sky cameraLabelled camera frames, for how moonlight affects the score on SvalbardCC BY 4.0
Tromsø AI camera classesAurora and cloud classes from the all-sky cameras at Tromsø and Skibotn, for how much aurora can still be seen through gaps in the cloudCC BY 4.0
MET Norway archiveArchived weather model forecasts (MEPS, 2016–2026; AROME-Arctic for Svalbard, 2015–2026) and airport weather reports (METAR), for the cloud part of the score and the 14-day chance forecastNLOD 2.0 / CC BY 4.0
ECMWF open dataCloud forecasts 2025–2026, to measure how many nights ahead a cloud forecast still helps the 14-day chance forecastCC BY 4.0, © ECMWF
ERA5 reanalysis (Copernicus)Hourly low, medium and high cloud, 2015–2026, to correct the cloud amount of the longer-range forecasts in the 14-day chance forecastCC BY 4.0 (credit below)
NOAA GEFS, processed by dynamical.orgEnsemble cloud forecasts 2020–2026, for the same measurement as the ECMWF dataCC BY 4.0 (NOAA's own data is public domain)

We also used methods from the literature. Magnetic coordinates follow Shepherd (2014, AACGM-v2). The eye's threshold and the brightness of the night sky follow Crumey (2014), Patat et al. (2006), Krisciunas and Schaefer (1991) and Kasten and Young (1989).

Where each number comes from

This table covers every number in the app and on this site. The score and the chances come from our model, so they are estimates.

NumberBuilt fromMeasured or modelledUpdates
The aurora score (0–100), in the app and on this siteNOAA SWPC (the OVATION aurora model, the Kp index and the 3-day forecast for the hours ahead), FMI magnetometers (the live score, mainland only), MET Norway's cloud forecast, the sun and the moon (calculated), and light pollution measured from satellite night-light dataModelled: your chance of seeing aurora with your own eyes in that hour, combined with how bright it should be, fitted on all-sky camera recordsEvery 5 minutes on our server (the score this site shows). The app shows that score while it is fresh, and otherwise calculates it on the phone with the same model
Kp (app)NOAA SWPC's real-time estimate of the planetary Kp index. In the first 90 minutes of each 3-hour block, the last full 3-hour value counts when it is higherMeasured (NOAA's estimate from ground magnetometers)Every minute
Clear (app)MET Norway's forecast for the spot: low, medium and high cloud, precipitation and the weather symbol. Snow and sleet count as an almost fully covered sky. Without a forecast, the month's average, marked EstimatedModelled: the chance of a clear enough view this hourHourly
Dark (app)The positions of the sun and the moon (calculated) and the spot's measured light pollutionModelled: how dark the sky is after twilight, moonlight and light pollution. It does not depend on aurora activityWith the score
Clear sky forecast, next 24 hours (app)MET Norway's hourly forecast for the spotModelled: the Clear chance, hour by hourHourly
14-day chance forecast (app)NOAA SWPC's 3-day forecast and 27-day outlook for aurora activity (the outlook adjusted each month against GFZ Potsdam's Kp), MET Norway's cloud forecast up to about 10 days ahead and the month's cloud average after that, and the sun and the moon (calculated)Modelled: the chance of seeing aurora at least once if you look during the night's dark hoursWhen the forecasts change: MET Norway hourly, NOAA's 3-day forecast twice a day, the 27-day outlook weekly

When an alert goes out

AlertGoes out whenMeasured or modelledChecked
Aurora active nowThe FMI magnetometers at Kilpisjärvi, Masi and Kevo show that a substorm (the magnetic field changing by 60 nT a minute or more) has just begun. Your spot must be dark enough for aurora (the sun at least 12° below the horizon), with its low and medium cloud both forecast at 40% or less. At most one an hour per region, and not on Svalbard.Measured (the magnetometers); dark and clear from the sun and MET Norway's forecastEvery 5 minutes
Good chance tonightYour spot's score reaches 60 (or the higher level you set) on live data, from the OVATION model and MET Norway's forecast. An Estimated score sends no alert. Once per spot per night (noon to noon), and not after a “Major aurora event” alert the same night.Modelled (the score)Every 10 minutes
Major aurora eventYour spot's score reaches 80, and NOAA's northern hemispheric power is 50 GW or more. If that power reading is missing, the current Kp must be 5 or more instead. Otherwise a score of 80 sends “Good chance tonight”.Modelled (the score), with a model estimate (hemispheric power, from NOAA SWPC's OVATION model)Every 10 minutes
Tonight's forecastAt 18:00 on your phone's clock, with tonight's outlook for your spot, and its best hour and score when they come from our server. It skips nights when no hour is dark enough to score above 0 (the midnight sun and light summer evenings).Modelled (the score's hourly forecast for tonight)Once a day

Alerts follow your settings and quiet hours. By default, quiet hours are while the sun is up over your region. “Tonight's forecast” arrives even during quiet hours.

Credits and citations

Here are the credits some licences ask for, in their own words, and the papers and datasets we cite.

  • VIIRS Nighttime Lights. This product was made utilizing VIIRS Nighttime Lights (VNL) data produced by the Earth Observation Group, Payne Institute for Public Policy, Colorado School of Mines.
  • WorldPop. The VNL annual composites we used were resampled, with gas flares removed, by WorldPop, University of Southampton.
  • NASA Black Marble. NASA's VIIRS products VNP46A2 and VNP46A4. Román, M. O. et al. (2018), Remote Sensing of Environment 210, 113–143.
  • Copernicus DEM GLO-90. © DLR e.V. 2010-2014 and © Airbus Defence and Space GmbH 2014-2018 provided under COPERNICUS by the European Union and ESA; all rights reserved.
  • Statens vegvesen (road database and road cameras). Contains data under the Norwegian licence for Open Government data (NLOD) 2.0 distributed by Statens vegvesen.
  • FMI all-sky camera intensities (CC BY 4.0). van de Kamp, M. (2026), Aurora intensities from All-Sky Camera data, Finnish Meteorological Institute, Zenodo, doi.org/10.5281/zenodo.19365681; described in van de Kamp, M. et al. (2026), Journal of Space Weather and Space Climate, doi.org/10.1051/swsc/2026026. We use only this published release, not FMI's camera images.
  • THEMIS. We acknowledge NASA contract NAS5-02099 and V. Angelopoulos for use of data from the THEMIS Mission; S. Mende and E. Donovan for use of the ASI data, the CSA for logistical support in fielding and data retrieval from the GBO stations, and NSF for support of GIMNAST through grant AGS-1004736.
  • FMI weather observations. Source: Finnish Meteorological Institute open data.
  • SMHI weather observations. Source: SMHI open data.
  • GFZ Potsdam Kp index. Matzka, J. et al. (2021), Geomagnetic Kp index, GFZ Data Services. doi.org/10.5880/Kp.0001
  • KHOnet2026. Partamies, N. and Syrjäsuo, M. (2026), UNIS/KHO. doi.org/10.5281/zenodo.19653026
  • Ny-Ålesund all-sky camera. Frames from the Light observatory Ny-Ålesund (Johnsen, G. et al., UiT/NTNU, NIRD Research Data Archive), with human labels by Ge et al. (2026).
  • Tromsø AI camera classes. Automated aurora and cloud classifications from the Tromsø AI camera project (Nanjo et al. 2022, Scientific Reports 12, 8038), as compiled by Ge et al. (2026, Zenodo, CC BY 4.0, doi.org/10.5281/zenodo.20446304).
  • ECMWF open data. © European Centre for Medium-Range Weather Forecasts (ECMWF).
  • ERA5 reanalysis. Hersbach, H. et al. (2020), The ERA5 global reanalysis, Quarterly Journal of the Royal Meteorological Society 146, 1999–2049. Hersbach, H. et al. (2023), ERA5 hourly data on single levels from 1940 to present, C3S Climate Data Store. Contains modified Copernicus Climate Change Service information 2026. Neither the European Commission nor ECMWF is responsible for any use that may be made of the Copernicus information or data it contains.
  • NOAA GEFS. NOAA NWS NCEP GEFS data processed by dynamical.org from NOAA Open Data Dissemination archives.

Region darkness and clear-sky data (open data)

We calculate the dark hours, dark-season dates, polar night and midnight sun periods on our region pages from the sun's position at the centre of each region. The share of mostly clear night-time observations comes from MET Norway weather station records, 2001–2020 (based on data from MET Norway, CC BY 4.0). You may reuse all of it under the Creative Commons Attribution 4.0 licence (CC BY 4.0). Please credit “AuroraChaser (aurorachaser.app)” and, for the clear-sky figures, “based on data from MET Norway”.

You can download the dark hours per region and month (CSV) and the season, polar night and midnight sun dates per region (CSV). Dates are given as month and day for the 2026/27 season and shift by about a day between years.