Overview

AuroraMe calculates an Aurora Score from 0 to 100 for any location on Earth by combining five real-time factors: geomagnetic activity (Kp index), cloud cover, moon phase and illumination, hours of darkness, and magnetic latitude. The score is a heuristic ranking of viewing conditions, not a statistically calibrated percentage chance. It produces a status — from "Unlikely" to "High" — that tells you whether it's worth going outside tonight.

Unlike services that only show the Kp index, AuroraMe accounts for the local conditions that determine whether you can actually see aurora. A Kp 7 storm means nothing if your sky is overcast.

Data Sources

AuroraMe predictions are built on trusted, real-time scientific data from multiple providers:

NOAA Space Weather Prediction Center (SWPC)

The NOAA SWPC is the primary source for geomagnetic data. We ingest:

Planetary Kp index — 3-hour geomagnetic activity measurement derived from 13 ground-based magnetometer stations at mid-latitudes worldwide (SWPC Planetary K-index)
OVATION aurora model — empirical model that maps aurora probability onto a geographic grid, updated every 5 minutes based on real-time solar wind measurements from NOAA's DSCOVR satellite at the L1 Lagrange point, 1.5 million km sunward of Earth (SWPC Aurora Forecast)
3-day Kp forecast — SWPC's official Kp prediction for the next 72 hours, issued daily
27-day outlook — solar rotation-based recurrence forecast

GFZ German Research Centre for Geosciences

GFZ Potsdam serves as our fallback Kp source. If NOAA data is unavailable, AuroraMe automatically switches to GFZ within 30 seconds via a circuit breaker mechanism (3 failures trigger a 30-second cooldown before retrying NOAA).

Cloud Cover and Weather

Hourly cloud cover forecasts are sourced from meteorological services and updated every hour for each tracked location. Cloud data is critical — overcast skies completely block aurora visibility regardless of geomagnetic activity.

Astronomical Calculations

Moon phase and illumination — calculated from standard astronomical algorithms. A full moon (100% illumination) washes out faint aurora, while a new moon (0%) provides ideal dark-sky conditions
Darkness windows — sunset, twilight phases (civil, nautical, astronomical), and sunrise times computed for each location daily. Aurora requires at minimum nautical darkness

Magnetic Latitude (MLAT)

Each location's magnetic latitude is computed from a genuine AACGM-v2 grid generated at 110 km altitude for the model epoch, then bilinearly interpolated at runtime. Where AACGM-v2 is undefined near the magnetic equator, AuroraMe explicitly falls back to a dipole approximation. Magnetic latitude determines the minimum Kp required for aurora visibility at a given location — this differs from geographic latitude because Earth's magnetic pole is offset from the geographic pole.

The 5-Factor Prediction Algorithm

Our algorithm evaluates five factors for each location and time window. Each factor produces a normalized score, and the weighted combination determines the overall visibility status.

Factor 1: Kp Index (Geomagnetic Activity)

The Kp index (0–9) measures global geomagnetic disturbance. We compare the current or forecasted Kp against the location's minimum Kp threshold (derived from magnetic latitude). The further the current Kp exceeds the threshold, the higher the Aurora Score.

Kp Threshold Formula

Required Kp ≈ (67 - Magnetic Latitude) / 2

Example: Tromsø at 67° magnetic latitude needs Kp 0–1, while London at 52° magnetic needs Kp 7+.

Factor 2: Cloud Cover

Cloud cover is evaluated on a percentage scale (0–100%). Aurora occurs at altitudes of 80–300 km, far above clouds. Any cloud layer blocks the view entirely. We classify coverage as:

0–20% — Clear skies, excellent viewing conditions
20–50% — Partly cloudy, aurora may be visible through gaps
50–80% — Mostly cloudy, significantly reduced visibility
80–100% — Overcast, visibility drops to near zero

Factor 3: Moon Phase and Illumination

Bright moonlight reduces contrast, making faint aurora invisible to the naked eye. However, research shows that bright aurora (IBC Class III and above) remains visible even under full moonlight. Our model adapts the moon penalty based on aurora brightness:

New moon (0%) — Ideal conditions, even faint aurora visible
Quarter moon (50%) — Minor impact, moderate aurora still clearly visible
Full moon (100%) — Faint displays washed out, but strong aurora (Kp 5+) still clearly visible

Factor 4: Hours of Darkness

Aurora requires dark skies. We compute exact darkness windows using the location's coordinates and date. Our algorithm adapts the darkness threshold based on aurora brightness — strong aurora can be detected earlier in twilight than faint displays:

Nautical darkness and beyond — Best conditions (sun more than 12° below horizon)
Nautical twilight — Acceptable for moderate to strong aurora
Civil twilight or daylight — Aurora not visible, even during storms

This factor is especially important at high latitudes during summer months (midnight sun) and the equinox transition periods. For example, Tromsø has 24-hour daylight from May through July, making aurora viewing impossible regardless of geomagnetic activity.

Factor 5: Magnetic Latitude

Magnetic latitude determines the baseline probability — locations within the auroral zone (65–70° MLAT) see aurora frequently at low Kp levels, while mid-latitude locations require rare strong storms. This factor sets the "difficulty level" for each location.

How Factors Combine

The five factors are weighted and combined to produce a single Aurora Score (0–100) and status label:

Status	Aurora Score	What It Means
High	50–100	Strong Kp, clear skies, dark conditions — go outside now
Medium	25–49	Conditions are favorable but one factor is suboptimal
Low	10–24	Worth checking if nearby, but don't travel far
Unlikely	0–9	Conditions not suitable (daylight, overcast, or Kp too low)

The current weights are domain-informed heuristic rules. The evaluation pipeline can record verified, opt-in sighting feedback and report each model version with threshold accuracy, Brier score, and observed-rate calibration error. Feedback collection is not yet large or representative enough for a fitted calibrator, so AuroraMe reports this output as a score rather than a probability.

In addition to the five core factors, our algorithm incorporates real-time solar wind data (IMF Bz component and solar wind speed) from NOAA's DSCOVR satellite at L1. When the interplanetary magnetic field turns strongly southward, our model can predict aurora intensification 15–60 minutes before traditional Kp-based forecasts — giving you an early warning advantage.

Update Frequency

Forecasts are refreshed at different intervals depending on the data source:

Data	Update Interval	Source
Kp index (current)	Every 15 minutes	NOAA SWPC
Solar wind (Bz, speed)	Every 5 minutes	NOAA DSCOVR (L1)
OVATION aurora probability	Every 5 minutes	NOAA SWPC
Cloud cover	Hourly	Meteorological services
Moon phase / darkness	Daily	Computed from ephemeris
3-day Kp forecast	Daily	NOAA SWPC
27-day outlook	Daily	NOAA SWPC

Coverage

AuroraMe provides forecasts for 1,000+ cities across 50+ countries, concentrated in regions where aurora is most frequently observed:

Arctic regions — Norway, Sweden, Finland, Iceland, Greenland, Svalbard
North America — Alaska, Northern Canada, Northern US states
UK and Ireland — Scotland, Northern England, Wales, Ireland
Russia — Murmansk, Arkhangelsk, Siberian cities
Southern hemisphere — Tasmania, New Zealand, southern Argentina (Aurora Australis)

Each location has a dedicated forecast page with real-time data, 12-hour detailed forecast, 27-day outlook, and historical monthly patterns.

Forecast Evaluation and Limitations

Aurora forecasting has inherent limitations due to the nature of space weather:

Current conditions (0–3 hours) — Most reliable. Based on real-time solar wind measurements from NOAA's DSCOVR satellite at L1
Short-term (3–24 hours) — Good accuracy for Kp direction (rising/falling), less precise on exact values
Medium-term (1–3 days) — NOAA's official 3-day forecast is useful for broad activity planning but less reliable for exact Kp magnitude and timing
Long-term (27 days) — Based on solar rotation recurrence; useful for trip planning but not for specific night predictions

Cloud cover forecasts follow standard meteorological accuracy — highly reliable for 0–6 hours, good for 6–24 hours, decreasing beyond that.

Why Forecasts Sometimes Miss

Coronal mass ejections (CMEs) can arrive earlier or later than predicted, and their magnetic orientation (Bz component) only becomes known when the solar wind reaches the L1 monitoring point — about 15–60 minutes before reaching Earth. A CME with southward Bz triggers aurora; northward Bz does not. This is why "last-minute" alerts from real-time data are often more accurate than multi-day forecasts.

Open Forecast Resources

Publishers and researchers can cite the machine-readable methodology JSON or embed the live NOAA aurora map widget. Attribution should link to this methodology page.

<iframe src="https://auroraforecast.me/embed/aurora-map" width="640" height="720" loading="lazy" title="Live aurora forecast map"></iframe>

Sources and References

NOAA SWPC — Planetary K-index
NOAA SWPC — OVATION Aurora Forecast Model
NOAA SWPC — Geomagnetic Storm Scale (G1–G5)
GFZ Potsdam — Geomagnetic Kp Index
AACGM-v2 2.7.1 — altitude-adjusted corrected geomagnetic coordinates, generated for the stated model epoch
NASA CCMC — OVATION Prime Model Documentation
Newell, P.T., Sotirelis, T., & Wing, S. (2009). "Diffuse, monoenergetic, and broadband aurora: The global precipitation budget." Journal of Geophysical Research, 114, A09207
Newell, P.T. et al. (2014). "OVATION Prime-2013: Extension of auroral precipitation model to higher disturbance levels." Space Weather, AGU — the operational aurora forecast model used by NOAA SWPC
Matzka, J. et al. (2021). "The Geomagnetic Kp Index and Derived Indices of Geomagnetic Activity." Space Weather, AGU — definitive reference for the Kp index by its custodians at GFZ Potsdam
NCEI — Wandering of the Geomagnetic Poles — magnetic latitude calculations account for the ongoing drift of Earth's magnetic north pole