Persistent shallow microseismicity near a glacier in southwestern Switzerland (Arolla VS) revealed by enhanced earthquake catalogs

On October 05, 2021 an Mw4.0 earthquake struck 6 km south of the village of Arolla, near the tongue of the Arolla Glacier. Almost one year prior to this earthquake, an M3.5 event occurred on November 08, 2020 in the same location. Both earthquakes were followed by a few aftershocks that were detected and located by the Swiss Seismological Service (SED). The unusually shallow depth of 1-2 km of these earthquakes, indications for a mostly thrust-type mechanisms within a region characterized by a predominantly extensional stress regime, and unusual high CLVD (50-70%) components of SED’s routine moment tensor solutions raised questions regarding the triggering mechanism. To understand and explain the possible existence of shallow thrust earthquakes in the area, we perform a thorough seismotectonic analysis that is based on enhancing the existing earthquake catalog of the SED and complementary moment-tensor solutions computed by multiple algorithms. The original SED earthquake catalog contains 83 earthquakes that occur between January 01, 2020 and December 31, 2021 and locate ~5 km around the two mainshocks. Using a deep learning based algorithm (EQTransformer), we detect additionally 253 events, thus the new catalog contains 4 times more earthquakes than the original SED bulletin. Absolute locations for the additional earthquakes are obtained using the probabilistic NonLinLoc method in combination with a recently updated Vp and Vs crustal 3D velocity model. In addition, we compute local magnitudes (MLhc) using SED’s standard procedure, in order to compile a homogeneous catalog consistent with the SED bulletin. The enhanced catalog events are used as templates for a match filtering scheme, which increased the number of detections by at least one magnitude order. Last, we relocate the final catalog using the double difference method towards obtaining a high resolution enhanced earthquake catalog. Spatially, the main cluster shows an intense seismic activity, stretched in N-S direction that matches the strike of the fault planes derived from moment tensor inversion. An additional cluster, that is not present in the SED bulletin locations, is identified next to the area were the aftershock activity of the two main events locates. Furthermore, the enhanced catalog shows a smother temporal evolution with more background events than previously recorded. Overall, we explore the possibility of fluid driven microseismicity that might be related to the nearby glacier. With our study we emphasize the importance of enhanced earthquake catalogs using both machine learning pickers and template matching algorithms. These approaches lead to unravel prior unmapped structures and improve our understanding of the seismotectonic regime in the study area.