Exploration and Quality Control of Large-scale Distributed Acoustic Sensing Data to Study Permafrost Degradation in Arctic Alaska

When permafrost thaws, it can cause meter-scale changes in surface elevation that cause irreversible damage to infrastructure and facilities, as well as the potential for release of carbon dioxide and methane that strongly reinforces global warming. Permafrost temperatures in Arctic Alaska have increased by 1-3°C in recent decades and are projected to continue to increase. We do not fully understand how permafrost's in-situ geophysical and geomechanical characteristics change with climate, region, and time at annual and decadal scales. This prevents building and maintaining resilient civil infrastructure in the Arctic. As part of a larger study aiming to understand and forecast variations in characteristics of degrading permafrost, we installed 2-km Distributed Acoustic Sensing (DAS) cable in Utqiaġvik, Alaska. DAS continuously records the ground's strain rate as a dense seismic array. The methodology developed in this research aims to provide transformative and cost-effective geotechnical monitoring in Polar regions.

Before we use any data for seismic imaging, we must characterize vibration sources over multiple frequency bands (e.g., wind, ocean waves, and anthropogenic noises) and investigate data quality of this many-terabyte dataset. We found a strong correlation between the DAS data's power spectrum density and wind speed, but our analysis suggests that much of this signal may be wind driving other vibration sources that are detected (rather than direct wind measurement). The success of ambient noise interferometry, by which we aim to retrieve signals mimicking controlled-source seismic data, requires good quality data containing environmental passive seismic noises. Our preliminary ambient noise interferometry results frequently show strong repeating signals, suggesting the need for novel pre-processing techniques. Therefore, we developed software tools to check DAS data quality and remove inappropriate signals automatically. We show a new data management strategy to reduce the intermediate ambient noise interferometry results, which previously took many terabytes of space. Eventually, we perform velocity analysis via surface wave dispersion to determine degrading permafrost's shallow seismic velocity structure.