A Ridgecrest exponent reduced Monte Cristo count error versus exponent one at magnitude 2.5 but not 3.0 on days 7-30 within 50 km

We tested whether an aftershock decay exponent from Ridgecrest improves later event counts at Monte Cristo. We used fixed USGS catalogs and a rate K times t to the power minus p, with no time offset or background term. We fitted Ridgecrest p on days 1-7, then fitted only K to Monte Cristo days 1-7. At preferred magnitude at least 2.5 within 50 km, Monte Cristo had 327 training and 293 test events. Its own fitted exponent of 0.668 predicted 427.2 events on days 7-30. A fixed exponent of one predicted 244.6. The Ridgecrest exponent of 0.901 predicted 289.2, with absolute error 3.8 events. Transfer had lower absolute error than the fixed exponent in three of six selected checks. At magnitude at least 3.0, transfer predicted 119.7 events against 93 observed; the fixed exponent predicted 91.2. Joint daily weight changes gave a wide primary transfer range of 21.5-1844.9 expected events. Thus, the close primary count is sensitive to selection and training days. This is a retrospective extension of a published Ridgecrest test, with one later target sequence. Mixed magnitude scales, missed events, clustering, and the simple rate model limit inference. The result does not establish general forecast skill or a physical cause.

aftershock · Monte Cristo · Ridgecrest · decay exponent · retrospective forecast · transfer test