Scale-dependent WSPR response to geomagnetic storms: frequency dependence and baseline-invariant absolute loss

Abstract

We test the hypothesis that geomagnetic storms (Kp ≥ 5) suppress WSPR spot counts on HF amateur bands. Using openly available data from wspr.live (hourly aggregated, 3 bands: 20 m, 40 m, 15 m) and GFZ Potsdam Kp index, covering 18 months (April 2024 – September 2025, 39 456 WSPR hours, 198 storm hours), we find significant storm-induced spot reductions on all three bands: 20 m: −27.7%, 40 m: −27.2%, 15 m: −38.7% (block permutation p < 0.0001 for all, 10 000 iterations, block = 24 h). The effect is scale-dependent: the drop nearly doubles from Kp = 5–6 (~−18%) to Kp ≥ 6 (−42.8%), consistent with non-linear ionospheric response to geomagnetic forcing. This gradient replicates on an independent held-out period (January–March 2026): at Kp = 6–7, training and held-out drops are within 89% to 95% of each other. The frequency dependence supports a dual-mechanism interpretation: MUF reduction dominates on 15 m (−38.7%), while D-layer absorption affects 20 m / 40 m (−27.7% / −27.2%). The day/night asymmetry (40 m drop stronger at night in 9/10 training months) is consistent with intermittent auroral and polar cap absorption. A key result is baseline invariance: the absolute spot loss on 40 m is identical between training and held-out periods (−18 300 spots/h in both), despite 38% growth in the WSPR network. This confirms that the storm effect is physically real, not a network-size artifact, and indicates a structurally constrained population of vulnerable propagation paths. We conclude that WSPR can serve as a distributed ionospheric sensor for detecting and characterising the scale-dependent effects of geomagnetic storms on HF propagation.