Stability of Seismicity Parameter Estimates: RTL and Gutenberg–Richter b-Value

Abstract

We present the results of examining the variability in the estimates of the complex parameter of seismicity RTL and the b-value of the Gutenberg–Richter frequency–magnitude relationship. Both parameters are frequently used in practical applications such as anomaly detection and earthquake forecasting. The calculation of these parameters involves a number of adjustable algorithm settings and alternative statistical estimation methods. The choice of a specific settings can be crucial to the calculation results, potentially reflecting algorithmic artifacts. In this paper, we explore the stability of these seismicity parameters with respect to the tuning coefficients in corresponding algorithms, using earthquake catalogs from Northern California and Kamchatka. Our results show that the empirical coefficients of the RTL algorithm have a characteristic effect on the calculation results: beyond certain threshold coefficient values, the variability of the results becomes negligible. A similar effect is observed when varying the spatial window for both the RTL and b-value calculations. We estimate the scatter of the results calculated under varying algorithm settings and analyze how this scatter varies across several distinct spatiotemporal regions within seismically active regions.