A three-day gap rule gave opposite summer changes in tremor detections and episode starts at 47-50 N in 2010-2024

We tested whether a reported fall in summer tremor counts also occurs in regional episode starts in northern Cascadia. We used a fixed public PNSN API v3.0 snapshot at 47-50 N. We joined active days until three inactive days separated them. In 2010-2015, 31 of 155 episode starts were in July-September (20.00%). In 2016-2024, 71 of 263 were in those months (27.00%). The early-minus-late change was -7.00 percentage points. Its exact independent calendar-phase p-value was 0.09265. The detection fraction assigned to summer episode starts fell from 57.31% to 26.77%. Summer episodes had 5.37 times as many detections as other episodes in the early period and 0.99 times in the late period. All six tested gap and count rules gave opposite directions for start and detection-weighted fractions. Thus a fall in summer detection counts need not mean a fall in summer episode starts. This is a catalog measure, not proof of physical event starts or hydrological control. Short records, regional grouping, catalog changes, and unknown station uptime limit inference. The package gives daily evidence, episode tables, code, source hashes, exact phase curves, uncertainty draws, and checks.

Cascadia · tectonic tremor · episode starts · seasonality · event weighting · catalog limits