This is a follow-up to a previous post about statistical anomalies between Cassini Saturn radio data and Earth magnetometer stations. The main criticism was solar wind as common driver. We ran the controls. Here’s what happened.

What we tested:

We had seven INTERMAGNET magnetometer stations and full-year 2017 OMNI solar wind data (1-minute IMF Bz, Bmag, velocity, density). Every test was re-run with solar proxies removed or with OMNI as an explicit control variable.

Day-of-week anomaly

Solar wind does have a day-of-week pattern (Kruskal-Wallis H=97 for Bz), which initially looked like it explained the SKR/ground DOW structure. Two things make it incomplete. First, solar wind magnetic field and velocity peak on different days: Bz peaks Monday (same as SKR, lag=0d), velocity peaks Friday (same as six of seven ground stations, lag=4d). Bz reflects HCS sector structure, which responds to sector crossings near-instantaneously. Velocity reflects energy flux, which propagates. They are reading different things, and the velocity/ground co-peak at +4 days is consistent with a signal that takes roughly that long to arrive.

Second, when we extract the 7-day Fourier phase from SKR in rolling 28-day windows across 2017, it drifts monotonically across the year (Spearman r=+0.484). OMNI’s phase is stable (r=+0.103). Saturn’s rotation is 10.7 hours, incommensurate with 168 hours, so any Saturn-modulated weekly pattern should precess as the year progresses. SKR does exactly that, while OMNI stays locked to the Sun’s sector boundaries. Note: with 41 overlapping windows the parametric p=0.0013 is optimistic; the meaningful observation is the contrast between the two r values, not the p-value alone. Two different clocks.

Geometric correlation, strongest result

We computed the PPO-band (10.4–11.0h, Saturn’s main magnetospheric oscillation period) cross-correlation between all 21 pairs of our 7 ground stations in 14-day sliding windows. We then correlated each pair’s time series of window correlations against Earth-Saturn distance for that week.

Results:

  • 8/21 pairs significant by permutation test (10,000 permutations), chance expectation 1.1
  • Best pair YKC-MEA: r=+0.665, p<0.0001 (permutation), p_parametric<0.0001
  • Control: same test against OMNI Bz activity level, r=+0.010

When Earth is closer to Saturn, auroral-latitude magnetometer pairs co-oscillate more strongly at Saturn’s rotation period. Solar activity level predicts none of this.

PPO partial correlations

Removing OMNI solar proxies via OLS and re-computing SKR vs ground station correlations in the PPO band, all 7 stations either hold or increase after solar removal. YKC, raw r=+0.188 -> residual r=+0.199. OTT: raw r=-0.281 -> residual r=-0.348. If solar wind were the common driver, regression out should reduce these. It doesn’t.

What we can’t explain with solar wind:

The geometric result is the cleanest, it’s controlled against simultaneous solar activity, permutation-tested against the station pair data itself, and the Saturn distance predictor (r=+0.665) simply doesn’t correlate with OMNI (r=+0.010). These can’t both be true if the mechanism is heliospheric activity driving both systems.

We’re not claiming to know the mechanism. We’re reporting that “solar wind explains it” is not an answer given these controls.

Would genuinely welcome anyone who works with SKR or Pc5 pulsation data taking a look at the geometric result specifically.