Search

The 150km long Longitudinal Valley (LV) in eastern Taiwan is located at the suture zone of Eurasian and Philippine Sea Plate, connecting two subduction systems in northeastern and southern Taiwan. In this area, considerable seismicity as well as crustal deformation are accommodated by the Longitudinal Valley fault system (LVF), which is mainly composed of high angle east-dipping thrust faults. To better understand the subsurface structure and the fault geometry of the southern LV, we conducted shallow seismic reflection survey across the LVF from Donli to Luyeh in 2017. Totally, there are four 4-km-long seismic reflection profiles. We used two mini vibroseis trucks with sweeping frequencies from 35-200Hz as sources and the common-midpoint spacing is 2.5m with maximum 30 folds. A strong east-dipping reflector relative to the bedrock reflection down to 1.5km deep can be observed after noise suppression. Some low reflectivity zones in the east side of the profiles could be related to the muddy Lichi mélange. It is interesting to note that in the eastern Luyeh profile, the syncline with the east-dipping axis implies the Lichi segment of the LVF and the orientation of the tectonic deformation. As a result, the southern LV displays an asymmetric sedimentary basin pattern with a deeper bedrock to the east. [ABSTRACT FROM AUTHOR]

Picking seismic events is a time-consuming task by just relying on manpower, especially in high seismicity regions. For instance, Taiwan, on February 2018, over 4,000 aftershocks were detected within 12 days after the Hualien earthquake. Although there are lots of papers propose various deep learning models that are suitable for dealing the problem, there is still a gap between the picking routine and the AI research. Thus, we design a toolbox that can be easily integrated into the regular work flow. In this study, based on PhaseNet developed by Zhu et al. (2018), the masking arrival-time label with a Gaussian distribution is the essential part for the training methodology. In addition, we implement the neural network model UNet++, which re-designs the skip pathways for better performance comparing to the original U-Net that is used by the PhaseNet. Here, we provide an open-source framework that can be easily installed and operated. This automatic picking toolbox is built upon Obspy and Keras with Tensorflow backbone, by reading S-file catalogs from SEISAN for generating labeled datasets. Besides, a Docker image is available for fast server deployment. As a result, in our scenario, most of the temporary stations are only contained single channel and mainly focus on P arrivals. We reduce the training set from input: three-components – output: P/S/Noise categories into input: Z-component – output: P category. The testing data is successfully identified by the model trained with 90,000 traces. Moreover, minor events missed by manually picking are detected by the model. [ABSTRACT FROM AUTHOR]

Autocorrelation of the ambient noise field and the seismic wave coda theoretically provide the near zero-offset reflectivity responses of the structure beneath the recording stations. The strong reflectivity in the stacked result implies the existence of the strong impedance of the subsurface layers. In this study, we investigate the crustal reflection characteristics beneath Taiwan by autocorrelating the ambient noise data and the regional/teleseismic coda waves. The broadband data are from the permanent broadband network BATS and the temporary deployments Formosa Array which aims to provide dense spatial sampling at the northern part of Taiwan. The autocorrelations are calculated by the phase autocorrelation and the amplitude autocorrelation with an additional spectral whitening at the pre-processing stage. Then the resulting stacks are computed using the time-frequency phase weighted stacking method. Similar results at the same location but from different recording periods and instrument types indicate the consistency and robustness of the subsurface signal. For the results of permanent stations which have years of data, seasonal variations of the amplitude of the signal can be observed. The priori information from receiver functions and tomographically inferred theoretical PmP two way times can help us to identify the signal from Moho, and most of the results show good correspondence with it, although some of the stations appear no clear Moho signal that can be attributed to gradual structural transition. Besides of explaining the single station result, for the dense array dataset, we can further trace the variation of the reflectivity responses along the profile we interest, which provides more constraint for illustrating the subsurface images. [ABSTRACT FROM AUTHOR]