We propose a new approach for source inversion with automatic criteria for the station selection process and using a multiobjective optimization scheme combining teleseismic body‐wave and regional surface‐wave observations. To illustrate the approach, we determine the source parameters of the 12 September 2016 5.4 South Korea earthquake, as well as its foreshock (same day) and its main aftershock on 19 September 2016.
Available teleseismic (short period and broadband) and regional (broadband) seismograms from the Incorporated Research Institutions for Seismology International–Federation of Digital Seismograph Networks (IRIS‐FDSN) webservice are selected (see Data and Resources). First, all teleseismic data with signal‐to‐noise ratio (SNR) above 2 in the 0.3–1.5 Hz frequency band are selected and a cepstrum function is blindly applied to help detect reflected wave arrivals inside a P coda. This method allows the identification of records with weak P‐coda and clear depth phase (pP, sP) arrivals and can be used to select an optimal set of teleseismic seismograms to be used for the inversion. Second, regional seismograms are selected according to the fit between their observed and theoretical dispersion curves extracted from local 1D velocity model. Depending on the epicentral distance, two frequency bands are proposed for regional data: 15–50 s for epicentral distances smaller than 2° and 30–80 s otherwise.
Finally, we explore the space parameters from exhaustive grid searches for the two objective functions and for teleseismic and regional data, respectively. We show that the seismic moment parameters can be exclusively derived from and can be fixed for . However, the depth parameter is not sensitive to and can be simultaneously inverted from both. The selection of the optimal solution satisfying both cost functions uses the optimum Pareto front that allows a decrease in epistemic uncertainty. We finally discuss the advantage of this approach for improving the characterization of moderate earthquake sources in areas with limited regional instrumentation.