The reported shallow-share rise persisted after three space-time exclusion rules in global M5.5+ ComCat data for 2000-2025

We tested whether nearby event clusters account for a rise in the share of earthquakes at reported depths of 15 km or less. We reused 12,848 global ComCat records at preferred M >= 5.5 and compared 2000-2012 with 2013-2025. After removal of exact depths of 10, 33 and 35 km, the unthinned rise was 9.75 percentage points. A largest-first rule that excludes events within 100 km and 30 days of a retained event left a rise of 8.54 points. Its conditional 95% three-year block interval was 2.96 to 15.52 points. Fixed 20-degree cell weights gave 8.51 points. Two other rules left pooled rises of 8.49 and 7.00 points. Thus these exclusion rules did not remove the reported depth change. They do not identify independent earthquakes or establish physical shallowing. The main thinning reduction was only 1.21 points, and its paired interval included zero. Assigned depths, preferred-origin changes, magnitude selection, long event sequences and conditional resampling remain limits. Code, aggregate evidence, source hashes, uncertainty draws, an independent count audit and a figure are included. External event records remain outside the publication.

earthquake depth · ComCat · catalog bias · space-time thinning · event clusters · sensitivity analysis