Spatiotemporal Seismicity Patterns and Strain Release in Active Magma-Poor Rifts, Resolved with a Machine-Learning-Enhanced Earthquake Catalog

Meritxell Colet, Folarin Kolawole, Rasheed Ajala, Felix Waldhauser, & Kaiwen Wang

Submitted September 7, 2025, SCEC Contribution #14641, 2025 SCEC Annual Meeting Poster #TBD

We address long-standing knowledge gaps on modes of strain release in regions of active tectonic extension where faulting and seismicity persist in the absence of voluminous volcanism, commonly known as ‘magma-poor rifts’. Examples of such rifts are found along East African Rifts’s Western Branch (e.g., Tanganyika, Rukwa, Malawi rifts) and along eastern California’s Walker Lane (e.g., Sierra Valley segment, northern Walker Lane). Here, crustal and upper mantle earthquakes and their space-time evolution may provide critical insight into the dominant deformation mechanisms that drive strain release. We explore the data-rich Tanganyika-Rukwa Rift Zone in East Africa, where two en-echelon, magma-poor rifts define the axes of active plate divergence, and where previous geophysical imaging has delineated the presence of blind lower crustal melts. We used data from the TANGA14 seismic array, which comprised 13 stations deployed from June 2014 to September 2015. First, we analyzed a previously published earthquake catalog containing ~2200 earthquakes, and performed declustering analysis to identify first-order seismicity space-time patterns, and their geometrical and geomechanical association with active faults. We identify 15 clusters of spatially isolated seismicity that can be categorized into distinct temporal scales and geometries (pipe and patch), indicating fluid-induced swarms and fault creep events, which manifest different modes of strain release in the extending lithosphere. Second, to further improve catalog completeness and resolve the fine-scale characteristics of these seismicity clusters, a high-precision earthquake catalog with improved magnitude completeness is required. To this end, we use machine-learning-based earthquake event detection to build an expanded catalog from the continuous waveform data, after which we apply relative relocation using HypoDD with cross-correlation delay time measurements. We then perform declustering on the machine-enhanced earthquake catalog to examine the refined clusters. The approach allows us to better resolve space-time seismicity patterns that inform the driving mechanisms of strain release in the extending lithosphere of the active rift zones.

Citation
Colet, M., Kolawole, F., Ajala, R., Waldhauser, F., & Wang, K. (2025, 09). Spatiotemporal Seismicity Patterns and Strain Release in Active Magma-Poor Rifts, Resolved with a Machine-Learning-Enhanced Earthquake Catalog. Poster Presentation at 2025 SCEC Annual Meeting.

Related Projects & Working Groups
Seismology