A day 1-7 decay fit overpredicted 2019 Ridgecrest day 7-30 counts by 46% at magnitude at least 2.5

We tested whether a decay rate fitted during days 1-7 predicts event counts during days 7-30 in a fixed USGS Ridgecrest catalog. The primary sample has preferred magnitude at least 2.5 within 100 km of the 2019 mainshock. The rate model sets the time offset and background rate to zero. The fitted exponent was 0.804. It predicted 565.9 events from 544 training events, but 387 events occurred in the test period. A benchmark with exponent fixed at one predicted 406.8. Fitting the exponent thus raised absolute count error from 19.8 to 178.9. The fixed exponent also had lower error in all 18 planned radius, magnitude, and fit-end combinations. These forecasts share events and are not independent trials. A Poisson simulation with parameter refitting gave a 95% predictive error band of 391.1-707.0 for the primary fitted model. The observed count was just below this band. A daily block weight sensitivity range for its expected count was 379.2-848.4 and included the observed count. Evidence against the rate model therefore depends on uncertainty assumptions. The split and benchmark were informed by a published endpoint analysis of this sequence. This is a retrospective audit, not a prospective skill estimate. One sequence, mixed magnitude types, catalog loss, omitted triggering, and model form limit inference. The package includes exact source fields, hashes, all forecasts and simulation draws, code, figures, independent checks, and an exact offline repeat.

Ridgecrest · aftershock count forecast · Omori decay · fixed exponent benchmark · retrospective forecast · catalog completeness · parameter uncertainty · reproducibility