Seismicity and Seismological Models of the Virunga Volcanic Province and Kivu Rift, Democratic Republic of Congo

The monitoring of geological hazards, such as earthquakes and volcanic eruptions, is the primary and most important task of volcano observatories. These hazards are observed and studied from a wide range of scientific disciplines. The work presented in this thesis is principally based on a local seismic network, which is often viewed as an essential monitoring infrastructure. While such monitoring is not an easy task for any observatory worldwide, it remains a major challenge for those in developing countries for a number of reasons, including recurrent socio-political instabilities. The Goma Volcano Observatory (GVO) is one of these.

The GVO is located in the Kivu Rift region in the bordering region of the Democratic Republic of Congo (DRC) and Rwanda, in the Western branch of the East African Rift. The active volcanoes Nyamulagira and Nyiragongo in the Virunga Volcanic Province (part of the Kivu) threaten the city of Goma (DRC) and neighbouring agglomerations. Urbanization in the direct vicinity of the volcano undergoes sustained rapid growth, and the region counts more than 1 million inhabitants today. Recent Nyiragongo’s flank eruptions, which occurred in 1977, 2002 and 2021, caused major disasters and casualties. Moreover, destructive earthquakes of moderate magnitude can also affect this tectonically active region.

After several attempts to set up an operational seismic network in the region, an initiative involving local and international scientific institutions was successful over the past decade. For the first time, a dense and telemetered broadband seismic network, called KivuSNet, provided real-time data in the region. This local network became the main seismic network of the GVO for daily routine monitoring work between 2013 and 2022.
This work presents the lessons learned from about 7 years (July 2015 – June 2022) of continuous seismic monitoring in the Kivu rift.
In the first part of this thesis, we have investigated the relevance of using the recorded seismic amplitude at each available station to assess the seismic activity in the Virunga volcanic Province (VVP). Under certain conditions, the spatio-temporal variability of volcanic seismicity can be monitored from the analysis of the seismic amplitude recorded across the network. The chosen approach is based on the amplitude at each station as well as the amplitude ratios between pairs of stations, which has proved to be efficient to track high-frequency volcano-tectonic seismicity (> 2Hz) at other volcanoes. In our case, we took advantage of the existence of shallow tremor sources (i.e., continuous seismic perturbations) of volcanic origin at low frequency (< 2 Hz) to successfully apply this technique for the first time in this context. Coupled with other geophysical datasets, the same approach was also well suited over the period 2018-2022 for interpreting the variations of seismic amplitudes in terms of volcano-related processes before, during and after the 2021 flank eruption at Nyiragongo. These results have direct implications in the monitoring of the Virunga volcanoes since they can provide guidance to GVO on the use of similar amplitude-based tools, which are easy-to-implement and relevant to use in case of seismic data scarcity using only a few stations.
In the second part, our aim was to fill the gap in the poor knowledge of the seismic velocity structure and seismicity patterns in the region. We used the seismological information derived from KivuSNet to obtain robust 1D seismic (P- and S-wave) velocity models for the Kivu Rift region. A first local magnitude scale was also proposed using the same data set. We recommend the use of these models as new routine models at GVO and as starting point for further works in the region. Both the methodology and the results were extensively detailed, which allowed a comprehensive comparison between this new model and existing ones. Then, a seismic catalogue of nearly 50,000 detected events (volcanic and tectonic) over 7 years (2015-2022) was located. The main characteristics (spatial distribution, count, magnitude, spectral frequency) were discussed with a special emphasis on Nyiragongo and Nyamulagira and how this new knowledge can help the Goma Volcano Observatory in improving its monitoring tasks. This analysis must be seen as a first promising outcome, which should encourage further research in the region on this large topic.