To constrain an earthquake’s source properties, the path‐ and site‐effect contributions to the seismic waveform can be approximated using another earthquake as an empirical Green’s function (EGF). An ideal EGF earthquake is smaller in magnitude than the mainshock and shares a similar focal mechanism and hypocenter. Here, we quantify how to optimally select EGF events using data from the spatially complex San Jacinto Fault Zone (SJFZ) in southern California. The SJFZ’s high seismicity rate allows us to test the EGF method for 51 target 3<Mw<5 mainshock events over a range of potential EGFs (>200 for each mainshock). We purposefully select a large population of inappropriate EGFs in order to identify thresholds and restrictions that optimize EGF selection criteria. For each mainshock/EGF pair, we compute the spectral ratio, fit the mainshock corner frequency, and measure the variability of these corner frequencies across the network. We assume a suitable EGF event will produce similar corner frequency estimates at every station. We discover that limiting hypocentral separation distances between mainshock and EGF events to <1 km (within ∼1–3 mainshock fault lengths) is an effective criterion in EGF choice. Surprisingly, separation distances of 2–14 km produce negligible changes in corner frequency variability, suggesting that EGF events at a 2‐km distance may be as poor a choice as EGF events at much greater distances. When EGF events within 1 km are not available, we suggest limiting EGF events to those with highly similar waveforms to that of the mainshock to ensure source similarity.
Online Material: Map of earthquake locations within our study region color‐coded to indicate the mapped spatial differential between the original catalog location in the ANZA catalog and the location in the Lin et al. (2007) relocated catalog.