2-D Acoustic Full Waveform Ambient Noise Inversion

We present a 2-D ambient noise full waveform inversion technique based on noise

cross-correlation sensitivity kernels. These kernels are constructed through the adjoint state

method, using a time-domain finite-difference solver to simulate both forward and

adjoint acoustic wavefields. Both the ambient noise source distribution and velocity structure

are treated as unknown. The inversion for source, and then structure parameters is

carried out sequentially. This sequential inversion is based on waveform energy misfit in

the case of noise source and cross-correlation travel time misfit in the case of velocity

structure. The present study focuses on applying this ambient noise full waveform inversion

methodology at local scales.

We use this approach to image the velocity structure beneath the Lonar crater in India.

This basaltic impact crater has close geological analogs on the Moon, and its internal structure

provides an important benchmark for assessing geometric models of crater formation.

We compare the results from our inversion with those obtained using conventional ambient

noise interferometry, which relies on Green’s function retrieval.