We ran the live Moonbow Index backwards over seventy-odd years of reanalysis weather and the river gauge's daily record: every night within two days of a full moon since 1950, scored the way tonight is scored. Here is what the moon has been up to.
A hindcast, not a sighting log. Reanalysis weather (ERA5) is smoother than what actually happened over the gorge, and it lacks the fog and rain-chance inputs the live score uses, so absolute numbers run a little rosy. The comparisons between months, hours, and years are the point.
Which month, really
| Month | Nights | Mean index | ≥ 60 ("worth the drive") | ≥ 85 ("go now") | Typical river | Typical cloud |
|---|
The odds on any given window
How the nights fall. The pile at zero is overcast; the pile at the top is the moon, the river and the sky all showing up on the same night.
Nights relative to the full moon. The park's "two either side" holds up, with the second night after full running noticeably weaker as the moon rises later and later.
Which hour
On nights scoring 40 or better, the hour the Index peaked. The model favors the late evening because it fades the score as the moon climbs past 55°; the park says midnight to 2 AM. That disagreement is exactly what field reports will resolve, and this chart will change when they do.
Year by year
The all-time nights
Twenty-five nights since 1950 the instrument would have called perfect: full or near-full moon, clear sky, big river, still air. If you were at the overlook on one of these, we'd like to hear about it.
| Night | Index | Moon | River (cfs) | Cloud |
|---|
And, for balance, the full-moon nights the sky simply refused, ranked by how much river went to waste:
| Full moon | Index | River (cfs) | Cloud |
|---|
Weather: ERA5 reanalysis via the Open-Meteo historical API (CC BY 4.0), hourly, for 36.84°N 84.34°W. River: USGS 03404000 daily mean discharge. Moon: computed. Generated . Regenerate with tools/climatology.js.