Observations of Long-term Subsidence in an Induced Permafrost Warming Experiment Using Distributed Acoustic Sensing (DAS)

Increasingly warming trends are causing degradation of permafrost landscapes. This alteration induces surface deformation that can pose a threat to civil infrastructure such as roads, pipelines and buildings. Hence, it is crucial to understand timescales and mechanisms involved in ground deformation related to permafrost thaw. The aim is to provide an observational basis that could be used to detect critical areas and develop early warning systems. Here, we focus on the feasibility of Distributed Acoustic Sensing (DAS) to detect and monitor subsidence as a result of permafrost thaw at depth. DAS is typically used to detect dynamic strain in the seismic bandwidth. However, recent studies successfully apply DAS to monitor quasi-static strain variations at very low frequencies and long periods (>1000 s). By temporally integrating strain rates measured by DAS to strain, we can derive the ground's long-term response to deformation. In this way, DAS provides a unique means of continuously recording subsidence with high spatial resolution (1 m) over large distances (10s of km). With that goal, a continuous, 4000 m long 2D DAS array was deployed in 9 crossing profiles across a permafrost region undergoing a controlled thaw experiment at the Fairbanks Permafrost Experiment Station in Alaska. Data was acquired for the entire duration of thawing, from August 5^th to October 4^th, 2016. The experiment, with dimensions of 10.5 m × 12.7 m, accelerated permafrost degradation by ca. two decades, deepening the permafrost table from 4 to 5.5 m. Analysis of the temporal variation of strain shows that DAS recorded the onset and long-term evolution of subsidence related to permafrost alteration. Subsidence initiates ca. 20 days after start of thawing, manifest in the DAS dataset as an abrupt decrease in strain. A rapid decrease occurs for a period of another 20 days after which deformation slows down and stabilizes. Results are validated against two-point Electronic Distance Measurements (EDM) and terrestrial based LiDAR images. They also correlate with data acquired using other fiber optics sensing techniques such as Distributed Strain Sensing (DSS). Our study shows the feasibility of DAS to detect subsidence related to permafrost degradation and how it could constitute a powerful tool for long-term monitoring of environmental processes.