Surface wave tomography for southeastern Asia using IRIS-FARM and JISNET data

Abstract

Previous seismic studies for southeastern Asia around Indonesia have used global seismic data and thus included few regional broadband seismic stations. To better understand seismic structure of this area, we temporarily deployed Japan–Indonesia Seismic NETwork (JISNET) which was composed of 23 broadband seismic stations in central to western Indonesia since 1998 under the Super Plume Project of Japan. In this study we analyse regional JISNET data and global Incorporated Research Institute of Seismology (IRIS)-Fast Archive Recovery Method (FARM) data, in order to obtain detailed images of 3D S-wave velocity (Vs) of this area. We use vertical component LP seismograms of FARM data recorded from 1989 to 1993 (13471 seismograms) and vertical component of broadband JISNET data recorded from 1998 to 2000 (1782 seismograms). We select seismic events with body wave magnitudes greater than 5.5, and take hypocenters and centroid moment tensor solutions from the Harvard CMT catalogue. Phase velocity anomalies of fundamental mode Rayleigh waves (period from 34 to 197 s) relative to Preliminary Reference Earth Model (PREM) are determined by measuring phase differences between the synthetic waveforms and the observations. Global distribution of Rayleigh wave phase velocities are determined by block inversions using damped least squares with a finer block size of 5°×5° for the Indonesia region than the surrounding area. Firstly, we compare the results obtained from the inversion using the FARM data alone (called “FARM inversion”) and those from the inversion using both the FARM and the JISNET data (called “JISNET + FARM inversion”). For the fundamental Rayleigh waves of short periods, phase velocities obtained from the JISNET + FARM inversion show greater degrees of negative anomaly throughout Indonesia than those from the FARM inversion. The clearest difference is seen for the region around Kalimantan and Java: slower phase velocities of less than 1% relative to PREM at periods of about 70 s are obtained by the FARM inversion, whereas including JISNET data requires lower phase velocity anomalies of 2.5%. Secondly, we compare the resolution kernels of both data set. Around Java, Sumatra, Kalimantan, and Sulawesi the resolution of the JISNET + FARM inversion is substantially better than that of the FARM data alone, with the peaks of resolution kernels up to five times as high as those for the FARM inversion. The best resolution improvement is obtained for west Kalimantan to Java Sea. Thirdly, we invert the phase velocity perturbations for the modes to obtain the upper mantle 3D S-wave velocity (Vs) structure of the study area. The JISNET + FARM inversion tends to give larger (up to four times) negative Vs anomalies when compared with the FARM inversion. The Vs heterogeneities concentrate in the uppermost mantle shallower than 200 km throughout Indonesia. Beneath Sumatra, Java, and Kalimantan, the uppermost 100 km of the mantle is slower than PREM by about 2–3%. A prominent high Vs anomaly (2–4%) around the depths from 100 to 200 km extends from northern Australia, through the Banda Sea, to east Sulawesi and Halmahera. The subducted Indian plate is visible as a nearly 1% high Vs anomaly.