Statistical Methods to Distinguish Shallow Moonquakes From Impacts

One of the biggest challenges in lunar seismology is accurately classifying seismic signals to evaluate impact and seismicity rates. While past studies have used qualitative differences for classification, we introduce quantitative methods to distinguish the most energetic signals: shallow moonquakes and natural impacts. Our approach utilizes previously under-analyzed data from the short-period Apollo seismic instruments, which operated on the Moon between 1969 and 1977. First, we convert short-period spectrograms to smoothed probability density functions. Next, we use the Kullback–Leibler divergence as a metric to measure the differences in the spectrograms between the two types of event. Using this comparison metric, we find that shallow moonquakes are more similar to other shallow moonquakes than to impacts. By analyzing individual waveforms, we identify features that significantly differ between shallow moonquakes and impacts, such as spectral entropy and autocorrelation. These features, which have characteristic ranges of values for each class of event, can be used to categorize the signal without comparison to another event. We apply these statistical metrics to a set of previously unclassified high-frequency events and shallow moonquakes that were identified last year. We find that high-frequency events and newly identified shallow moonquakes have a variety of features. Many of the shallow moonquakes that were identified last year are consistent with those identified over 50 years ago and may have a range of source depths. Along with supporting reanalysis of the Apollo seismic signals, these statistical metrics may be useful for future analysis of lunar seismic data.