Earthquake catalogs are fundamental cornerstones in the study of earthquake phenomena. They provide the origin time, location, and magnitude information that constitutes the basis for earthquake interaction analysis, as well as physics‐ and statistics‐based earthquake forecasting models. In this article, we locate the southern California seismicity using the state‐of‐the‐art probabilistic and nonlinear method NonLinLoc. We use only the P wavepicks to avoid introducing the velocity‐model and picking‐time errors of the S phase, which is harder to detect and thus less constrained. Using a subset of the best locatable earthquakes, we conduct a joint inversion using the VELEST software to obtain a minimum 1D velocity model and station corrections. We use the NonLinLoc method with this 1D velocity model and the inferred model uncertainties to obtain realistic location distributions for each event. We then perform a multifractal analysis to compare the resulting catalog with the current state‐of‐the art catalog obtained using double‐difference methods. Our results indicate that the absolutely located seismicity exhibits a spatial scaling regime that is more consistent with the surface fault traces and is much less clustered compared to its relatively located counterpart, which features an additional distinct spatial scaling break. We speculate that this discrepancy is likely to have significant effects on physical and statistical models that rely on consistent seismicity catalogs as their input data. We also find that there is a correspondence between the spatial scales detected in the multifractal analysis of seismicity and the scales of active faulting in the region.

You do not have access to this content, please speak to your institutional administrator if you feel you should have access.