Published seasonal earthquake peaks at six western US volcanic areas did not repeat in 2004-2025 ComCat data; more snow gave a larger July-December share at three volcanoes after 2003 only

Christiansen et al. (2005) and Saar and Manga (2003) reported seasonal earthquake rates at western US volcanic areas in data that ended near 2003. We tested the published peak half-years with USGS ComCat data for Mount St. Helens, Mount Hood, Mammoth Mountain, the Long Valley east south moat, Yellowstone Lake and Hebgen Lake, for 1984-2003 and for new data from 2004-2025, with three non-volcanic control areas. The statistic was the rate ratio of Gardner-Knopoff declustered events in the published peak half-year to the other half. A year-rotation null kept swarms and yearly totals. In 2004-2025 no volcanic area had a higher rate in its published peak half-year (one-sided p 0.07-0.97; Stouffer p = 0.80). In 1984-2003 our method found the signal only at Mammoth Mountain (rate ratio 1.39, p = 0.019, peak near 31 August), and it did not continue (0.88, 95% CI 0.61-1.35). A small late-August excess at Mount St. Helens disappeared for larger events and agrees with seasonal detection change. In 2004-2025, years with more 1 April snow water equivalent had a larger July-December share at Mount St. Helens, Mount Hood and Mammoth Mountain (permutation p 0.010-0.041; six-area Stouffer p = 0.009), but not in 1984-2003 or in control areas. Limitations: our own area boundaries, a live catalogue, few snow stations, and many related tests without adjustment.

seasonal seismicity · volcano seismicity · snowmelt · groundwater recharge · Mammoth Mountain · Mount St. Helens · Mount Hood · Yellowstone · Long Valley · ComCat · SNOTEL · out-of-sample test · replication