Oral Presentation

Distributed acoustic sensing (DAS) allows us to more easily collect long-term data to study earthquake processes at fine scales across large regions, so DAS data rates are often thousands of times higher than traditional seismometer data rates. This is a problem when archiving and sharing data: our public data archives lack the infrastructure currently to support widespread use of DAS data. This is also a problem in analysis: the trace-by-trace nature of many seismology software packages is not well-suited to dense seismic array data, and most software aimed at dense seismic array data primarily focuses on active source exploration seismology (i.e. different from the passive seismic methods often used in earthquake seismology).

For some applications in earthquake seismology, it can be appropriate to simply downsample data in space and time to reduce data quantities enough for trace-by-trace analysis. However, for applications in urban areas (with many local noise sources), or where there may be small-scale geohazards, downsampling loses too much useful information. Thus, we are motivated to design more efficient algorithms for passive seismology that speed up analysis, that are robust to approximation errors introduced by lossy compression, that can operate directly on lossy-compressed data without decompressing, and that can take advantage of data products. This talk will give a high-level overview of several of these new algorithms for beamforming, pattern matching of events, and ambient noise interferometry. Further, current community needs related to data standards and software interfaces for DAS data will also be highlighted.