Temporal stability of surface wave dispersion extracted from ambient seismic noise using three-component beamforming

Accurate subsurface seismic velocities are crucial for drilling exploration wells, exploring geothermal resources, or locating seismic events. Due to their dispersive nature and prevalence in ambient seismic noise, surface wave velocities can be used to obtain shear velocities beneath seismic arrays. Localized shear velocity anomalies indicate the presence or absence of fluids; Temporal variations in shear velocities can indicate changes in fluid content or poisson ratio over time, i.e. during geothermal operations.

 

Three-component ambient noise beamforming (B3Am) detects waves passing over seismic arrays and retrieves their propagation direction, propagation speed, and polarization parameters. Since B3Am analyzes the wavefield at discrete frequencies, with the frequency band limited by the array size, and in short time windows, typically ten times the period, results can be stacked over short time periods and temporal analysis of dispersion curves, backazimuth, and polarization parameters becomes possible. However, before interpreting temporal variations caused by physical changes in the subsurface, the overall variability needs to be estimated to accurately estimate uncertainties. Changing noise source fields do not impact the B3Am results directly but could cause deviations in absolute parameters. Furthermore, seasonal changes in water content in sediments could introduce seasonal variations not related to geotechnical activities.

 

Using an existing dataset recorded with 23 broadband seismometers deployed over a range of ten months in an area of about 15 km in diameter in Germany, we analyse the seismic noise wavefield and estimate the temporal stability of surface wave dispersion curves. We calculate probabilistic power spectral densities, investigate their variability over time, and compare them to the wavefield composition, i.e. body to surface wave ratio, computed with B3Am. We plot backazimuths and velocities against frequency for all wave types for selected days. We compare different stack lengths in a bootstrapping-type analysis to see the minimum number of detections, i.e. recording length, needed for accurate results. We see that for frequencies below 1 Hz, five days of continuous noise recordings produce a stable dispersion curve. However, we see large seasonal variations between results in Autumn and Spring to results in Winter. These variations can be attributed to changes in the noise field, both of natural and anthropogenic origin. Finally, we derive expected uncertainties and provide insights about the impact for depth inversions.