US Aurora Season Climatology

Important: this is a typical year reconstructed from a closed record: planetary Kp from 1996-01-01 to 2026-09-07, crossed with a 2001–2020 cloud normal. It is not a forecast, not a probability for any dated night, and not a record of aurora sightings. The live city pages on this site show tonight's cloud, moon and darkness from current data; the two will not agree, and neither is broken when they differ.

Downloads

• JSON dataset
• CSV dataset — the flat city-month grain, 4,476 rows.

Methodology and limitations

For every city, 89,661 three-hourly Kp intervals from the GFZ Potsdam planetary Kp index (1996-01-01 to 2026-09-07) were tested against that city's own horizon threshold and against local astronomical darkness (sun geometric altitude at or below -18 degrees (astronomical twilight)), sampled every 10 minutes across each year of the record and bucketed by solar time so daylight saving never enters. A night counts as an opportunity when at least one dark interval reached the threshold. That count is then scaled by the local monthly mean cloud amount from NASA POWER SYN1DEG 20-year Meteorological and Solar Monthly & Annual Climatologies (January 2001 - December 2020) to give clear_opportunity_nights.

Thresholds mirror the same Kp ladder the forecast pages use. Moon interference is reported separately as moon_free_share — the share of dark hours with lunar illumination below 35% — and is not subtracted from the opportunity count.

Oval position is a model estimate, reported separately from the measured record. NOAA auroral oval: equatorward boundary near magnetic latitude 67 - 2 x Kp, elevation from a 110 km emission height and a 6371 km Earth radius. model estimate, reported separately; it never overrules the measured record in `verdict` — a location the model puts below the horizon can still have a Kp threshold crossing in the record, and Miami's 2024-05-11 entry is exactly that case.

Limits worth stating: cloud is a 20-year monthly mean, so it smooths away the run of clear nights that actually decides a trip; the Kp record includes two solar maxima and one deep minimum, so a single year can sit far from the average; and nothing here measures light pollution, terrain, or viewing direction. See methodology for how the live forecast differs.

Columns

• city, state, month: location and calendar month (jan–dec).
• dark_hours: mean hours of astronomical darkness on the nights of that month that have any. Zero means none at all — Alaska from May to August.
• opportunity_nights: nights in an average month with at least one dark interval at or above the city's horizon threshold, before cloud.
• cloud_amount_pct: monthly mean cloud amount from the 2001–2020 normal.
• clear_opportunity_nights: opportunity_nights scaled by (1 − cloud). The headline metric.
• moon_free_share: share of dark hours with the moon under the illumination cutoff.

The JSON carries the same rows plus per-city and per-state summaries: thresholds, magnetic latitude, best months, clear nights per year, the verdict band, and the last date Kp reached the local threshold.

Sources

• GFZ Potsdam planetary Kp index (CC BY 4.0)
• NASA POWER SYN1DEG monthly climatologies (parameter CLOUD_AMT)
• NOAA SWPC: Aurora Tutorial
• Aurora Forecast USA methodology

License, citation, and contact

Use is subject to the site terms; this dataset is not released under an open-data license. The underlying Kp index is published by GFZ Potsdam under CC BY 4.0 and must be credited separately.

Cite as: “Aurora Forecast USA by AuroraMe, US Aurora Season Climatology, version 2026-09-07, https://aurora-forecast-usa.com/data/us-aurora-season-climatology.” State that values describe a typical year from a 1996–2026 Kp record and a 2001–2020 cloud normal, not a forecast or an observation record.

Questions or corrections: support@auroraforecast.me.

States

Monthly figures for a state are those of its best-placed tracked city, named below. The clear-nights range spans every city we track in that state.

Cities