Detecting sources of shallow tremor at neighboring volcanoes in the Virunga Volcanic Province using seismic amplitude ratio analysis (SARA)

Volcano monitoring requires simple techniques to rapidly identify the cause of volcanic unrest. The so-called RSAM (real-time seismic amplitude measurements) technique, used in many observatories, is a good example of extracting information from seismograms with minimal processing. Built on a similar principle, the more recent seismic amplitude ratio analysis (SARA) technique allows locating migrating seismicity at high frequency (> 2 Hz, e.g., due to dike intrusions) under certain assumptions. However, such analysis generally requires a dense distribution of stations close to the seismic sources (depending on the magnitude) and/or station sites undisturbed by human activity. In a more straightforward and qualitative approach, computing amplitude ratios between station pairs can also allow for the detection of temporal and (2D) spatial changes of volcanic activity. In this work, we adopt such a simplified approach of SARA in order to characterize seismic tremors originating from two open-vent neighboring volcanoes, Nyiragongo and Nyamulagira, in the Virunga Volcanic Province (VVP) in the Democratic Republic of the Congo (DRC). In contrast with previous studies, we focus here on the low-frequency band (0.3–1 Hz), free from anthropogenic noise and sensitive to shallow volcanic tremors linked to intermittent or permanent intra-crater eruptive activity recorded through the large-aperture local network. We apply for the first time the SARA methodology for volcanic sources predominantly generating surface waves and propagating over long distances. The analysis is performed on more than two years of continuous seismic data. Seismic amplitude analysis in this frequency band is strongly influenced by the short-period microseisms originating from nearby Lake Kivu. Despite this diurnal to seasonal amplitude variability, SARA successfully detects continuous volcanic tremor activity and its arrest at both volcanoes. In light of these findings, we discuss the applicability of the method to the continuous, real-time detection, and characterization of long-period shallow volcanic tremor sources in this region.