Potentials and pitfalls of permafrost active layer monitoring using the HVSR method: a case study in Svalbard

Abstract

Time-lapse monitoring of the subsurface using ambient seismic noise is a popular method in environmental seismology. We assess the reliability of the horizontal-to-vertical spectral ratio (HVSR) method for monitoring seasonal permafrost active layer variability in northwest Svalbard. We observe complex HVSR variability between 1 and 50 Hz in the record of a temporary seismic deployment covering frozen and thawed soil conditions between April and August 2016. While strong variations are due to changing noise conditions, mainly affected by wind speed and degrading coupling of instruments during melt season, a seasonal trend is observed at some stations that has most likely a subsurface structural cause. A HVSR peak emerges close to the Nyquist frequency (50 Hz) in beginning of June which is then gradually gliding down, reaching frequencies of about 15–25 Hz in the end of August. This observation is consistent with HVSR forward modeling for a set of structural models that simulate different stages of active layer thawing. Our results reveal a number of potential pitfalls when interpreting HVSRs and suggest a careful analysis of temporal variations since HVSR seasonality is not necessarily related to changes in the subsurface. In addition, we investigate if effects of changing noise sources on HVSRs can be avoided by utilizing a directional, narrowband (4.5 Hz) repeating seismic tremor which is observed at the permanent seismic broadband station in the study area. A significant change of the radial component HVSR shape during summer months is observed for all tremors. We show that a thawed active layer with very low seismic velocities would affect Rayleigh wave ellipticities in the tremor frequency band. We compile a list of recommendations for future experiments, including comments on network layouts suitable for array beamforming and waveform correlation methods that can provide essential information on noise source variability.