We have successfully inferred the focal mechanism solutions for 16 events in the Mw 6.3 Agustus 2012 Palu-Koro earthquake sequence, with magnitudes ranging from Mw 3.8 to Mw 6.3. Quality of moment tensor used here was adopted from Clinton et al. (2006) and Langston et al. (1982). Two of the solutions were classified as high-quality (A-quality) where station used at least 4 with VR 62-73% (VR for A-criteria > 60%), DC 56-81% (more than 50%), and CN 3-8 (below 10). The remaining 14 solutions were classified as moderate-quality (B-quality) where station used at least 3 with VR 41-59% (VR for B-criteria 40-60%), DC 56-81% (more than 50%), and CN 3-8 (below 10).
The newly obtained focal mechanism solution supplements the existing GCMT solution catalog, which only contains the solution for the Mw 6.3 mainshock within the same geographical area and observation period. Tables 2 and 3 present the centroid moment tensor solutions for the Palu Earthquake Mainshock, with comparative analyses of solutions from other agencies, as well as events in the aftershock sequence. More detailed information is provided in Appendix: Table S1.
To test the robustness of the Bayesian Inversion, we first applied the algorithm to determine the centroid moment tensor (CMT) solution for the significant Mw 6.5 August 2012 Palu event and compared the result to solutions from other agencies. By utilizing the Bayesian method, we showed that the DC value was greater than solutions from other agencies (Table 2). The fault plane orientation, suggested as relatively north-northwest to south-southeast, aligns with the geological characteristics of Sulawesi Island and the Palu-Koro fault system (geological information provided by Beaudouin et al. 2003).
The results suggest that non-DC processes are insignificant. Jost and Hermann (1989) suggested that this non-DC component can be interpreted in two forms, i.e., two double couple (DC) that occurs simultaneously during an earthquake or through a compensated linear vector dipole (CLVD) mechanism (the component of the moment tensor that describes the volume change that occurs during an earthquake, this is related to the expansion or contraction of the earthquake source zone). The occurrence of tectonic earthquakes in regions distant from volcanic sources, such as Palu-Koro, appears to align with the first mechanism. The decomposition of the 2012 earthquake indicates that the two DC mechanisms are a major moment tensor with a left strike-slip fault mechanism and a minor moment tensor with a normal fault mechanism. (Fig. 4 and Table 4).
The Bayes-ISOLA code calculated uncertainty values for each result and visualized them using a histogram. The more centered the histogram, the more representative the results are compared to the original physical conditions. Uncertainty values were examined for various parameters, including distance, DC value, magnitude, and time shift. The largest uncertainties were observed for distance and depth, with maximum values of 7.74 km and 4.94 km, respectively. The maximum uncertainty in magnitude was 0.48. These results are good considering the wide range of regional BMKG stations used. As an illustrative example, the processing results for the event with the best solution are presented in Fig. 5 (VR value above 50, DC above 70, clear uncertainty histogram, good station distribution), while the inversion results for the other events can be found in the Appendix (Fig. S1-S16).
Our analysis revealed that the majority of the centroid moment tensor (CMT) solutions converge near the Palu basin intersection (Fig. 6). To interpret the geometry of the fault, we selected fault planes from the CMT data listed in Table 5 that met the following criteria: a rake value between -20 to 20 degrees, indicating a left strike-slip mechanism (adopted from Houksson 1990); a strike value that was relatively north-south or at least northwest-southeast; or the strike value with the highest probability distribution as output by the Bayes-ISOLA analysis (for this criterion, Earthquake No. 5 used as examples in Fig. 7).
The overall histogram for the 16 CMT Solutions is presented in Fig. 8. From the analysis of the three fault parameters, we found that most CMT solutions have strikes that trend predominantly towards North or Northwest, indicating the dipping direction will point between Northeast-East. Most of the dip inclinations were observed to be within the 80° to 90° range (Tend to be vertical). The rake is quite consistent with the regional geology of Palu-Koro which is characterized by a left-lateral strike-slip fault, where half of the solutions generated have this mechanism (rake value between -20o to 20o).