The 2004 source-shape switch reduced calculated GCMT time widths in a Mw 5.8 to below 7.5 audit

We audited assumed source durations in the Global CMT catalog for 1976-2020. We selected 8,367 records with moment magnitude from 5.8 to below 7.5. We removed cube-root moment scaling and calculated the time standard deviation for each recorded source shape. Between 1998-2003 and 2004-2009, median scaled half-duration changed by -2.43%. Median scaled time width changed by -31.01%, with a conditional 95% three-year block interval of -32.03% to -29.89%. For the same half-duration, the boxcar-to-triangle switch alone reduces width by 29.29%. An earlier fall in scaled half-duration also occurs near 1992-1993. These are catalog measurements of assumed functions. They do not show a change in actual earthquake rupture duration. Current catalog documentation already states that half-duration is assigned from moment. We add an aggregate audit across eras and a shape-aware calculation tool. One revised catalog, selected magnitude limits, exploratory era choices and unmeasured source errors limit interpretation. Code, aggregate tables, source hashes, a numerical audit and a figure support reproduction. External event records remain outside the package.

Global CMT · source duration · catalog audit · source time function · earthquake dynamics