Characterizing the Effects of Noise on Receiver Function Estimation

Abstract:

The receiver function method is firmly established as a seismological technique for investigating the wave speed discontinuity structure in the crust, upper mantle, and mantle transition zone beneath seismic stations. The theoretical basis for this method is that the conversion of compressional to shear wave energy can be characterized as a linear, time-invariant system in which P wave motion on the vertical (or L) component is mapped to SV wave motion on the radial (or Q) component record. This simple model allows for efficient calculation of receiver functions using various deconvolution techniques. Even with careful data quality control and selection, practical implementations can introduce artifacts in the guise of arrivals that appear coherent across epicentral distance. We calculate a set of high-frequency synthetic seismic data in reference Earth models with known discontinuities, using the axisymmetric spectral element method AxiSEM. We contaminate them with a series of noise realizations to characterize our ability to recover the input model through our implementation of the iterative time-domain deconvolution receiver function method, using our newly developed and publicly released Python package rflexa. When recovery shows signal-processing limitations, manifested as spurious discontinuities, we explore the effectiveness of filtering and stacking techniques to combat them. Our geophysical focus remains on improving our knowledge of mantle transition zone discontinuity structure beneath ocean islands, noisy environments with limited aperture and insufficient coverage.