Geological and tectonic implications obtained from first seismic activity investigation around Lembang fault

The Lembang fault located at northern part of populated Bandung basin is the most conspicuous fault that potentially capable in generating earthquakes. The first seismic investigation around Lembang fault has been done by deploying a seismic network from May 2010 till December 2011 to estimate the seismic activities around that fault. Nine events were recorded and distributed around the fault. Seven events were likely to be generated by the Lembang fault and two events were not. The events related to the Lembang fault strongly suggest that this fault has left-lateral kinematic. It shows vector movement of Australian plate toward NNE might have been responsible for the Lembang fault kinematic following its initial vertical gravitational movement. The 1-D velocity model obtained from inversion indicates the stratigraphy configuration around the fault composed at least three layers of low Vp/Vs at the top, high Vp/Vs at the middle layer and moderate Vp/Vs at the bottom. In comparison with general geology of the area, top, mid and bottom layers may consecutively represent Quaternary volcanic layer, pre-Quaternary water-filled sedimentary layer and pre-Quaternary basement. Two eastern events related to minor faults and were caused by a gravitational collapse.


Backgorund
The Bandung Basin is a plateau surrounded by several mountains and active volcanoes. The basin morphology formed due to tectonic and volcanic activity during the Quaternary. Thick lake sediments were deposited in the basin during the Late Quaternary. These deposits might conceal several buried faults. The Lembang fault is located at the northern part of the Bandung basin where The Bandung City is located. Administratively, this basin is located in West Java province and surrounded by several mountains, which are associated with mapped faults (Figure 1). Bandung is a densely populated city in Indonesia. It is surrounded by several faults that may potentially be earthquake sources. Regarding to disaster risk reduction, it is important to reveal the seismicity in this area. We studied seismic activities around the Lembang Fault, the most conspicuous fault in the basin.
The formation of the Lembang Fault was explained by Dam [1]. In the Early-Middle Quaternary, the west-east trending highland with the oldest volcanoes of the Burangrang-Sunda complex (including Tangkubanperahu Volcano), the volcanic ridges and peaks in the northeastern Lembang area, and most of the volcanic terrain between Bandung and Sumedang were formed. Following the build up of the Sunda volcano, a gravitational collapse, due to the loading of enormous amounts of volcanic deposits on ductile marine sediments, caused thrust faults and diapiric structures in the near surface strata of the northern foot slopes (Van Bemmelen, [2]). Rifting associated with catastrophic sector failure eruptions destroyed the volcanic cones, while the depressurization of the main magma reservoir led to normal faulting and the formation of the Lembang fault. This fault, with an impressive scarp, was studied by Tjia [3], who concluded both older dip-slip and younger strike-slip displacement had occurred.
Recent paleoseismological study shows several evidences of near past activities of the fault. This study concluded that within last 2 kyrs, the Lembang fault has been capable in producing earthquake of~6.8 and 6.6 magnitude at about 2 and 0.5 kyrs BP respectively [4]. Accordingly, the fault may be capable in triggering earthquakes of comparable magnitudes in the future.
The Bandung Basin as seen in (Figure 1) will act to amplify seismic waves if the Lembang fault generates an earthquake. The level of amplification is depended on sediment thickness. The sediment structure has been investigated using the microtremor array method [5], which showed that the deepest basement reaches about 3.5 km. Seismic wave amplification in the Bandung Basin was simulated by Afnimar [6] using Haskell's method.
Although paleoseismological study of the Lembang fault shows evidences of significant faultings at the past, the seismicity around the Lembang fault is generally very low and mostly not sensed by people. In July 21, 2011, a M 2.9 earthquake and in August 28, 2011, a M3.3 earthquake (BMKG report) were those among others sensed by people and brought spotted damages to houses in the vicinity of the fault zone, and were recorded by the local seismic network around the fault. Until now, detail seismic investigation of the Lembang fault has not been done. In this study, we investigate it using hypocenter relocation (including 1-D velocity determination) and focal mechanism analysis.

The data
A temporal seismic network ( Figure 2) has been deployed around the Lembang fault by BMKG from May 2010 till  December 2011. Four Taurus-Nanometrics seismometers are installed at Lembang (LEM), Cimenyan (MYN), Parongpong and around Cibodas. In the middle of 2010, two seismometers were moved from Parompong and Cibodas to Tangkubanperahu Volcano (TKP) and Ciater (ATR), due to avoid noises from farming activities. During these two years, the network has recorded about 15 earthquakes. This research will only analyze events those that originated from the area around Lembang. One example of the recorded seismic waveform is shown in (Figure 3). The P wave arrival picking is based on its onset that is clearly visible on the seismograms. It is more difficult to find the S wave onset, especially in the seismogram from station MYN. Fortunately, the horizontal components of seismograms from stations LEM, TKP and ATR show clear S-wave onsets. The picking of S waves from these three stations could be used as a guidance to find the S wave phase on the seismogram from station MYN. There are one or two phases observed at station MYN before the S wave arrival. These phases are probably produced from reflections due to complicated structure north of this station.

Methods
The first step that should be done on earthquake analysis is earthquake location determination. An earthquake location includes a geographical position, a depth and an origin time. The origin time can be determined using what is called a Wadati diagram [7]. The result from the Wadati diagram is one input of the gradient inversion method that is usually used to locate one event. This is the reason for this method is often used as a single event determination (SED). The velocity structure used in this step is guessed from geological structures around the Lembang fault. This inversion method was first introduced and applied by Geiger [8] and called the Geiger method of earthquake location. The result of the SED method should be recalculated due to the structure heterogeneity around the Lembang fault. A joint Hypocenter Location (JHD) method was first proposed by  Douglas [9] to accommodate the residual time at all stations (station correction) caused by velocity heterogeneity of station locations. Kissling et al. [10] extended the JHD method by including a 1-D velocity model as a parameter in inversion.
To estimate the earthquake mechanism, the moment tensor inversion developed by Kuge [11] is applied in this research to estimate the earthquake mechanism. The optimum moment tensor solution is reached by best fitting between observation and synthetic waveforms through inversion process. The synthetic waveforms are calculated by the extended reflectivity method developed by Kohketsu [12]. The velocity structure where that synthetic waveform calculated is 1-D velocity model result from the JHD. The observation velocity waveforms are cut from P-wave onset to S-wave pulse (5 to 10 s windowing) and are filtered in 0.075-0.25 Hz using SAC. Sometime for small event, the velocity waveforms of an event are integrated to get the displacement waveforms to reduce the ringing pattern. We show the original outputs related to the highest magnitude event ( Figure 4) and to the lowest one ( Figure 5). The fitting between observation waveforms and synthetic ones for all events are varies that is identified by variance value (see Figures 4 and 5). Although there is variation of variance values, the synthetic waveforms still cover the trend of the observation ones. The observation waveforms recorded at station TKP look like noisy and can not fit well with synthetic waveforms. Even some events, for example the event in (Figure 5), the observation waveforms can not be identified at this station. The reason should be caused by the structure heterogeneity around Tangkubanperahu Volcano.

Earthquake location and focal mechanism
The relocated SED hypocenters obtained using the JHD method for all events listed in (Table 1) and their double-couple solutions of moment tensor results listed in (Table 2) are plotted in (Figures 6 and 7). Most of them seem to have relationship with the Lembang fault. Only two very shallow events (at a depth less then 5.0 km) are located to the east part and therefore it is unlikely to have relationship with the Lembang fault.
The events related to the Lembang fault strongly suggest that this fault has left-lateral kinematic with    Figure 6). Because the hypocenters for these events lie at some distance from this surface trace, they appear at first glance to be unrelated to the Lembang fault. But the strikes of the fault planes designated (I) in (Table 2) are quite consistent with the strike of the Lembang fault. Their vertical distribution along the cross section AB indicated in (Figure 7) also aligns well with a possible westward extension of the Lembang fault, assuming the near-vertical dip that is consistent with the estimated fault planes. For these reasons, we interpret these events to be related to the Lembang fault. This implies that the Lembang fault extends at least 10 km further westward than would be inferred from its surface trace. Consequently, there should be a fault line extends slightly westward from the end point of Lembang fault. This line could be connected to the existing Lembang fault line and morphologically unexposed (dash-line in Figure 6), or it is a different segment of Lembang fault. Based on regional mapping of morphological features, Horspool et al. [15] notified that at the west-end, the fault line is slightly hooked southward showing a horsetail shape. At the south end of this horsetail shape, another possible fault line stretches almost parallel to the Lembang fault where to the north of this line, events 2, 3, 4, 5, 8, 9 are distributed. Therefore, we simply interpret that those events were related to this line that is probably another segment of Lembang fault. From these events, we also can state the geometry of Lembang fault. The average strike is about 277°which is not so different with surface trend line of 282°, the dip is about 85°and the rake is about 35°.
Result of GPS measurement of Australian plate slip azimuth yielded an average direction of~N20-21°E [13]. As the general trend of Lembang fault is N282°E (Figure 8 (Figure 9).
Events those occurred at the eastern part of Lembang fault are distributed in an area where a graben structure had developed during cataclysmic eruption of Sunda Volcano at about 0.2-0.18 Ma [16]. A pair of E-W oriented faults at the north and south bordered this graben [16]. The south border is then recognized as the east segment of Lembang fault. This initial geological structure influences further local tectonic  Figure 6 Focal mechanisms of all events. The number above each beachball is the event number listed in (Table 2). Red dots are epicenter distribution and the red triangles are seismic station locations. The red solid-line is the exposed surface trace of the Lembang Fault, while the red dashed-line is a possible extension of the Lembang Fault and the blue dashed-line is another hypothesized segment of the Lembang Fault.

Velocity structure
The 1-D structure including P and S wave velocities (Table 3) obtained from the JHD method is presented as graphics of Depth vs. Velocity ( Figure 10). Interpretation is given in this figure. Three layers can be distinguished from Vp graphic i.e. high Vp values at depths deeper than 6 km with an exception of that at 18 km, moderate Le mb an g Fa ult Figure 8 Horizontal components of pressure (outward-pointing arrow couple) and tension vectors (inward-pointing arrow couple). Figure 9 A modified schematic geometry of the Lembang Fault. Vp at depths from~6 to~0.75 km, and low Vp at depth shallower than 0.75 km. The range of high, moderate and low Vp values are higher than 4.0, 3.0-4.0 and 2.0-3.0 km/s respectively. Two layers can be distinguished from Vs graphic i.e. high Vs at depths deeper than 6 km and low Vs at depths shallower than 6 km. The range of high and low Vs values is higher and lower than 2.5 km/s respectively. Two layers can be distinguished from Vp/Vs graphic i.e. high Vp/Vs at depths deeper than 0.75 km and low Vp/Vs at depth shallower than 0.75 km. The range of high and low Vp/Vs values is above and below 1.5, respectively. Accordingly, the stratigraphy configuration around the fault can be deducted composing of three layers i.e. high Vp/Vs with high Vp and Vs at the bottom (less than 6 km), high Vp/Vs with moderate Vp and low Vs at the middle (6 -0.75 km), and low Vp/Vs with low Vp and low Vs at the top (less than 0.75 km). The top boundary of this layer is probably at a depth between 0.5 to 1 km indicated by prominent decreases of Vp/Vs and Vp. It is assumed that this boundary is at 0.75 km. The lower boundary is probably at a depth between 5 to 8 km depths as indicated by prominent decreases of Vs and Vp. This lower boundary is assumed at 6 km. Vs are relatively low in the top and middle layers with a subtle fluctuation.
Low Vp/Vs with low Vp in the top layer may correlate with large aspect ratio of water content in pores of rocks. Takei [17] report that the water-filled pores have In comparison with general geology of the study area (Figure 11), the top layer should represent Quaternary volcanic layer. The middle and bottom layers should represent Tertiary sedimentary layer or basement according to Satake and Harjono [5].
The station corrections obtained from the JHD method are tabulated in (Table 4). The negative station correction at LEM is observed to occur on bedrock and the positive ones are those on sedimentary or weathering rocks. This means the waves arrived earlier at stations on bedrocks than those on sediment or weathering rocks. Pujol [18] obtained minus value correction related to high velocity anomalies and vice versa from data of Loma Prieta, California, mainshockaftershock sequence. Our result shows similar indication to that of Loma Prieta. The minus value at LEM is related to outcrop of igneous rock (high velocity anomalies) along Lembang fault. The plus ones at TKP, MYN and ATR are related to the volcanic zone (low velocity anomalies).

Conclusions
From this investigation, stratigraphy of the study area has been revealed based on Vp, Vs and Vp/Vs, consisting of three layers. In a perspective of aspect ratio of water content, the top layer with low Vp/Vs, low Vp and low Vs is composed of rocks with largest aspect ratio of water content. The bottom layer with high Vp/Vs, high Vp and high Vs is composed of rocks with smallest aspect ratio of water content. In comparison with general geology of the area, the top layer should represent the Quaternary volcanic layer, and the middle and bottom layers should represent the Tertiary sedimentary layer.
The source mechanism of earthquakes along the Lembang fault is left-lateral faulting. All western events are probably related to a new segment of the Lembang fault. This new segment is maybe developed by pressure of Australian plate indicated by horsetail feature. Two shallow eastern events are related to the minor faults and caused by a gravitational collapse.