Spatially Locating Time-Lapsed Relative Velocity Changes in Yellowstone

In this study, we examine temporal subsurface relative velocity variations in Yellowstone. This project region has a much larger spatial scaling, as well as a much longer, less abrupt, temporal change than most of the other active settings in which ambient noise temporal velocity variations are measured. By examining the time lag of the ambient Noise Correlation Function (NCF) for a given station pair, we can extract relative velocity variations. These changes represent either subsurface velocity changes (in active settings) or changes due to the seasonal variation in the ambient seismic field. We improve the temporal and spectral stability of the NCFs through a new filter (the adaptive covariance filter - ACF), which allows us to evaluate the main Rayleigh wave arrival rather than coda. By making use of the main arrival the spatial location of these changes can be better understood. We examine all of the available data near the Yellowstone National Park area, including the USArray and the NOISY networks. Using the ACF, we calculate robust NCFs for all station pairs (overlapping in time). We then obtain time-lag estimates of the main arrival for various calendar times throughout the data availability. These lags are then compared to the lags observed at all other spatial station pairs. Using these many combinations of station pairs and lags, we build and perform an inversion, where these relative velocities are compared to a background velocity at subsequently different times. The spatial location of the velocity changes can be observed over time, and it can be ascertained as to whether these observed velocity variations are due to subsurface changes (such as the inflationary uplift from 2004-2010 in Yellowstone), or whether the observed changes are more indicative of the regional changes in the ambient seismic field.