Refining Moho Depth Estimates from Receiver Functions, introducing variable Weights for P-to-S converted Phases

In order to understand the crustal structure of the Earth and related geodynamic processes, accurate knowledge of the underlying mantle and crust and its properties, such as thickness or seismic velocities, is necessary. The depth of the Moho and relative bulk crustal velocities are often estimated using receiver functions and the Hk-stacking method. Hk-stacking employs a grid search to find the combination of H and vP/vS which best reproduces the arrival times of P-to-S converted phases. P-to-S converted phases (Ps, PpPs, PpSs+PsPs) are generated at interfaces with strong velocity contrast, such as the Moho. The Ps phase is usually the strongest, while attenuation and local scattering dampen the multiples. Therefore in the original Hk stacking, a higher weight is attributed to the first arrival and the multiples are considered less. Furthermore, usually an underlying vP is defined, These assumptions are a source of bias in the depth and relative velocity estimate of the Moho. In this study, we present an adaptive Hk stacking algorithm that expands the traditional two-dimensional grid search to five dimensions: H, vP/vS, vP, and the weights w1 and w2. The algorithm includes a misfit function that adjusts the weight distribution based on the observed amplitudes of the converted phases, reducing bias from user-defined parameters. We apply the algorithm to receiver functions of four seismic broadband stations in Central Italy. We found that different weight distributions can significantly affect the estimated vP and, consequently, the Moho depth. Our results indicate that accounting for variable weights and vP leads to a more realistic assessment of uncertainty. This improves the reliability of Moho depth estimates and vP/vS, which are key to improve our geological and geodynamic understanding.