Performance evaluation of the Malta seismic network

Abstract

At the beginning of the project, the permanent Malta Seismic Network (MSN) consisted of 6 permanent stations, 3 being located in Malta, 2 in Gozo and 1 in Comino. The MSN is operated by the Seismic Monitoring Research Group (SMRG) within the Department of Geosciences at the University of Malta. For a 12-month period between August 2017 and August 2018 a temporary network of 6 additional stations were installed in various locations around the Maltese archipelago as part of the FASTMIT project. This extension to the permanent Malta Seismic Network (MSN) increased the number of recording stations to 12, offering an unprecedented opportunity for working with seismological data. The performance of each station in terms of its background level of noise was evaluated. This consisted of an analysis of noise (both of natural and anthropic origin) in terms of seasonal variations by considering 15-day periods in winter and in summer for each station. Via computation of probability density functions of power spectral density (PSDPDF) the level of background noise and its origin, in different frequency bands, was analysed. Furthermore, noise recordings were also used to perform a horizontal-to-vertical spectral ratio (HVSR) where several amplification peaks were identified for stations underlain by a significant thickness of Blue Clay. This was used to present tentative time-delay corrections related to non-uniform topography and geology. Initially, the performance of the extended MSN as a beamforming array was assessed for a limited number of well-located teleseismic and regional Mediterranean events of large magnitudes, by comparing the locations derived from the software package Generic Array Processing (GAP) to that provided by international bulletins such as the EuropeanMediterranean Seismological Centre (EMSC), for which results were promising. However, the main focus of this project was the potential of the MSN and the extended network to be used as a steered, small-aperture array for locating regional earthquakes. Using data archived by the SMRG, seismic activity up to 150 km from the Maltese archipelago was analysed by investigating 21 events which were only recorded locally in addition to 6 events which were also located by neighbouring networks affiliated with the Istituto Nazionale Geofisica e Vulcanologia (INGV). All these local events had already been located using singlestation polarisation algorithm (SSL) and where possible, using the multiple SSL technique, incorporated within LESSLA. This technique is known to suffer from back-azimuth (BAZ) estimation. GAP was used to investigate whether the array could yield improvements in the BAZ. Seismic traces recorded at the 12 stations were used to perform a grid search for slowness, ranging between – 25 s / ° and + 25 s / °. Using beamforming, a value for optimal BAZ was obtained at maximum beam power using the program slow2d. Generally, the standard deviation for BAZ for these low-magnitude events was below ± 10 °, with the standard deviation typically depending on the quality of the seismic traces, the magnitude of the event, and the epicentral distance from the reference station WDD. The epicentral solutions were obtained using the epicentral distances previously calculated by the SMRG and available on the online portal, derived using an S-P time calibration curve. The performance of the network when including only the 6 permanent stations, and also when introducing tentative corrections compensating for varying geology and topography, was also analysed for any effect on BAZ estimation. The locations obtained using GAP were also compared to other methodologies such as least square fitting techniques using elocate. The results indicate an improved ability to calculate BAZ for small earthquake activity around the Maltese Islands, especially to the south-east of the archipelago, when using GAP. Moreover, it can also be concluded that seismicity in the study area considered is more clustered than originally thought and is tentatively linked to active faults to the south-east of the archipelago.