Constraining lowermost mantle structure with PcP/P amplitude ratios from large aperture arrays

Abstract:

Observations of weak short-period teleseismic body waves help to resolve lowermost mantle structure at short wavelengths, which is essential for understanding mantle dynamics and the interactions between the mantle and core. Their limited amount and uneven distribution are however major obstacles to solve for volumetric structure of the D” region, topography of the core-mantle boundary (CMB) and D” discontinuity, and the trade-offs among them. While PcP-P differential travel times provide important information, there are trade-offs between velocity structure and core-mantle boundary topography, which PcP/P amplitude ratios can help resolve, as long as lateral variations in attenuation and biases due to focusing are small or can be corrected for.

Dense broadband seismic networks help to improve signal-to-noise ratio (SNR) of the target phases and signal-to-interference ratio (SIR) of other mantle phases when the slowness difference is large enough. To improve SIR and SNR of teleseismic PcP data, we have introduced the slant-stacklet transform to define coherent-guided filters able to separate and enhance signals according to their slowness, time of arrival and frequency content. We thus obtain optimal PcP/P amplitude ratios in the least-square sense using two short sliding windows to match the P signal with a candidate PcP signal. This method allows us to dramatically increase the amount of high-quality observations of short-period PcP/P amplitude ratios by allowing for smaller events and wider epicentral distance and depth ranges.

We present the results of measurement of PcP/P amplitude ratios, sampling regions around the Pacific using dense arrays in North America and Japan. We observe that short-period P waves traveling through slabs are strongly affected by focusing, in agreement with the bias we have observed and corrected for due to mantle heterogeneities on PcP-P travel time differences. In Central America, this bias is by far the stronger anomaly we observe after correcting for geometrical spreading, radiation pattern of the source, reflection and transmission coefficients at interfaces and intrinsic attenuation using the AK135 and QL6 1-D models. In areas not influenced by slabs, we generally see lower amplitude ratios than predicted.