Arjun Datta, Aileni Mahesh

We present ANSEICCA, an open-source package for forward and inverse modelling of seismic ambient noise cross-correlations at local scales, where the effects of Earth's sphericity are negligible. The package implements a nonlinear finite-frequency inversion technique wherein measurements of cross-correlation energy are used to invert for the spatial distribution of ambient noise sources, under the assumption of a fixed Earth structure model. It is seamlessly integrated with other open-source Python packages for seismic wave propagation modelling, including a C-based numerical solver for acoustic modelling in 2-D media. It is a unique package insofar as the inversion is based on finite-frequency sensitivity kernels, but executed without the adjoint method. Instead, speed and computational efficiency are achieved by parallelising the code. Moreover, the Hessian-based optimization ensures convergence in a relatively small number of iterations (~10), compared to purely gradient-based methods. We introduce the structure of the package in detail, describing both the serial and parallel versions of the code. Performance benchmarks show that ANSEICCA affords compute times of the order of a few minutes per iteration of inversion, with typical local-scale seismic array geometries.