Appropriateness of the McNamara and Buland's (2004) methodology for computing frequency-dependent seismic power

TL;DR

This study evaluates the McNamara and Buland (2004) seismic power spectral density methodology, finding it suitable for applications requiring averaged seismic energy over frequency bands, but with limitations in high-resolution spectral detail.

Contribution

It compares McNamara and Buland's method with Welch's method across different stations, clarifying its appropriate use cases in seismic analysis.

Findings

Both methods produce similar trends in seismic power time series.

McNamara and Buland's method is suitable for averaged spectral energy calculations.

Slight differences exist in absolute power values between the methods.

Abstract

The methodology developed by McNamara and Buland (2004) for computing Power Spectral Densities (PSDs) has gained popularity due to its low computational cost and reduction of spectral variance. This methodology is widely used in seismic noise studies and station performance evaluations and is implemented in tools like ISPAQ, MUSTANG, and PQLX. However, concerns have been raised about its appropriateness in certain contexts, particularly when high-resolution spectral detail is required. This study evaluates McNamara and Buland's methodology by comparing it with Welch's method across three Alaskan stations with differing microseism conditions. When calculating seismic power across a band of frequencies--for example, the 5-10s secondary microseism--we find that both methodologies produce time series with nearly identical trends, albeit with slight differences in absolute power values. Our results demonstrate that McNamara and Buland's methodology is fully appropriate for certain applications, specifically ones that rely on averaged seismic energy over a frequency band as opposed to a single discrete frequency.