A clock-change test left the northern Cascadia tremor detection shift uncertain in 2017-2025

We tested whether northern Cascadia tremor detections shift with daylight-saving clock changes. We used a fixed PNSN API v3.0 source at 47-50 N for 2017-2025. On each active weekday, we measured morning counts minus evening counts as a share of all daily counts in fixed UTC-8 time. We compared 28 days before and after each clock change and aligned the signs to a civil-clock prediction. Twelve of 18 transitions had enough active dates on both sides. The mean aligned change was 8.24 percentage points. A conditional 95% year-bootstrap interval was -1.41 to 19.41 points. A 42-day window gave 7.78 points with limits of 3.14 to 13.08. A 14-day window had support in only one year. Thus the primary test leaves the change uncertain and depends on the time window. Sparse tremor activity, changing spatial patterns, unknown station uptime, seasonal changes and only nine years limit inference. This is a detection-pattern test, not a measure of true slip or proof of human noise. The package gives daily counts, code, source hashes, all transition results, bootstrap draws, a figure and an independent source audit.

Cascadia · tectonic tremor · daylight saving time · detection bias · clock change · null result