Apollo Lunar Seismic Data Disturbance Mitigation and Phasor Walkout Method Assessment for Searching Planetary Free Oscillations

    Planetary seismic data are essential for studying planetary interiors and dynamics, yet acquiring high-quality recordings in harsh extraterrestrial environments turns out to be far more challenging than on Earth. In both the Apollo Passive Seismic Experiment on the Moon and the Seismic Experiment for Interior Structure (SEIS) on Mars, considerable transient disturbances—often referred to as “glitches”—span a wide range of frequencies, complicating the search for potential low-frequency signals of planetary free oscillations and gravitational wave responses. To address this issue, we propose an automated workflow for detecting and removing strong transient disturbances in Apollo seismic data with deep learning, thereby enhancing the recovery of weak long-period signals. We also examined two approaches for removing disturbance: (1) directly muting transient segments and applying Fourier transforms, and (2) treating these segments as data gaps and applying the Lomb–Scargle periodogram to uncover weak low-frequency signals. Synthetic tests show that even with ~80% of the data contaminated, most low-frequency peaks can still be recovered. Moreover, our workflow recovers the recently discovered temperature-related long-period signals in Apollo data without relying on stacking or clustering techniques, highlighting its vast potential in revisiting low-frequency components of Apollo seismic data.

    Complementing this disturbance-mitigation framework, we also evaluate the phasor walkout method, which determines whether a spectral peak originates from a true signal or merely from noise. The core assumption of this method is that harmonic signals will generate linear walkout patterns at their true frequency peaks, while random noise will produce irregular, random walkout paths. However, our findings indicate that random noise can contain a considerable amount of frequency peaks with deceptively linear phasor walkout patterns. Although noise in planetary seismic data are not random Gaussian noise, and artifacts observed in simple random noise may or may not arise in actual lunar or Martian data, this finding nonetheless highlights the need for extra caution when interpreting phasor walkout results in planetary seismic data.

    In summary, our study offers both an effective strategy for strong-disturbance removal—enabling the search for weak low-frequency signals—and an assessment of the phasor walkout method, raising awareness of risks in potential misinterpretations. These insights not only open new avenues for re-examining legacy Apollo data but also provide gentle reference in detecting planetary free oscillations and other low-frequency seismic signals.