Towards 4D mapping velocity changes with coda wave interferometry: methodology and applications

Abstract:

Understanding at what depth changes in seismic properties (wavespeed dv/v) occur might be a powerful indicator of subsurface tectonic deformation, magmatic activity, and susbsurface hydrology. To infer these velocity changes, seismologists measure phase shifts in coda waves of repeating earthquakes and/or ambient noise correlation functions. Coda waves are a poorly understood mix of surface and body waves such that the mapping of the velocity to depth is non-trivial. It is common practice to assume that low frequency measurements are sensitive to greater depths. The spectrum of dv/v measured over a wide range of frequency opens a new window to map dv/v measurements to depth. We develop a new method, referred as Wavelet Transform Stretching (WTS, Yuan et al. 2020), inspired by the recent study of Mao et al. (2019) develop the use of the wavelet transform to explore time-frequency variations of the phase shifts. We demonstrate WTS to be more stable and efficient.

Through series of numerical experiments and sensitivity tests, we explore the sensitivity of the spectrum of dv/v with background velocity and characteristics of perturbed layers. Dv/v is always measured as being a large underestimation of the imposed perturbation. We find that the spectrum exhibits a characteristic peak frequency that is sensitive to the depth of the velocity: it is high for shallow perturbation and low for deep perturbations. In simple models, we can derive a simple scaling between this characteristic frequency and the perturbation depth. We demonstrate its applicability into two settings: a sinking sedimentary basin in Jakarta and the active volcanic setting of Long Valley, California.