Comparing performance of multi-frequency bands Occam’s receiver function inversion to standard linearized receiver function inversion

Abstract

Receiver function (RF) is a characterized waveform that sensitive to the sub-surface velocity structure near the seismic station. In the geophysical problem, an inversion scheme has been used for determining the structure model that corresponded to the observed data. Because RF inversion is a non-linear problem, the standard method such as linearized inversion suffered in the non-uniqueness of the results. Thus, using a stochastic optimization algorithm (e.g. Monte Carlo or Generic Algorithm) is the most interesting trend for solving the problem. However, their computation cost is high and the efficiency is strongly depended on user setting. In this work, Occam’s inversion algorithm, which is popular in the electromagnetic method, and multi-frequency bands receiver function have been integrated to improve the efficiency of receiver function inversion. The propagation method has been applied for the calculation of seismogram in this work. The synthetic cases, which are thin low velocity zone model (TLVZ), broad low velocity zone model (BLVZ), thin high velocity zone model (THVZ) and broad high velocity zone model (BHVZ), were used to compare the performance of this newly implemented algorithm and the standard linearized inversion. Single frequency band (SFO) and multi-frequency bands Occam’s inversion (MFO) have been tested for each case. For, TLVZ, THVZ and BHVZ, the MFO inversion provided better fitting results than SFO in most cases. Comparing to linearized inversion of CPS330, the MFO results still provide better performance. For the smooth low velocity structure in BLVZ model, both linearized and MFO can recover a significant structure of the true model. More concisely, the linearized inversion can recover the absolute velocity of the true model better than MFO. By the way, the MFO provides the best data fitting for the BLVZ. In summary, the MFO receiver function inversion that implemented in this study provides a new improved tool for a seismologist to investigating the subsurface structure.