Abstract

The prediction of seismic wave fields between stations using machine learning offers great potential for geophysical monitoring, particularly in remote areas or in regions with sparse sensor coverage. We introduce a novel encoder-decoder deep learning architecture that successfully learns the transfer function between seismic stations. By learning the complex signal transformations, this method enables accurate predictions of how seismic signals alter as they travel from one station to another. Notably, high quality predictions are achieved using only two days of data consisting solely of ambient seismic noise. Extending the approach from individual station pairs to entire seismic arrays, Virtual Seismic Arrays are introduced as a powerful proof of concept. By training the algorithm on all station pairs within an array, a set of predictive models is obtained that collectively form the Virtual Seismic Array. This enables the reconstruction of full-array recordings from a single reference station, even after physical sensors are no longer present. In the secondary microseism frequency band, beamforming analysis validates the effectiveness of Virtual Seismic Arrays by showing a high degree of agreement between the original and predicted waveforms. This novel application of encoder-decoder networks for modelling transfer functions has the potential to enhance seismic monitoring, while reducing the need for continuous sensor coverage. By reconstructing signals at multiple stations from a single reference station, the approach enables ongoing array functionality in remote regions while reducing costs and maintaining array capabilities.

Abstract

The Arabian Plate is formed by major earthquake-generating plate boundaries: extension and oceanic spreading the south and west along the Aden Ridge & Owen fracture zone and the Red Sea, respectively, compression and subduction in the east along the Makran and the Zargos fronts, and lateral motion in the north and northwest along the East Anatolian Fault and the Dead Sea Fault system. The geology of eastern Saudi Arabia is dominated by thick sedimentary sequences of the Arabian Platform with the world’s largest oil reservoirs. In contrast, the Arabian Shield of western Saudi Arabia is formed by an amalgation of Precambrian terranes, dotted with Cenozoic volcanism and affected by Red Sea rifting during the past ~30 Mio years. Interplate & intraplate earthquake activity and. widespread Cenozoic volcanism on the Arabian Peninsula need to be reconciles with the long-term tectonic evolution of the Arabian Plate with its current surface geology. And despite its abundant oil & gas resources, the Kingdom of Saudi Arabia is exploring the use of geothermal energy.

The Earth structure underneath Saudi Arabia remained elusive until 2008 when the Saudi Geological Survey began installing a dense seismic network across the Kingdom. Early studies provided an initial understanding of crustal and upper-mantle structure based on small datasets (e.g. Mooney et al., 1985; Julia et al., 2003; Al-Damegh et al., 2005; Chang and Van der Lee, 2011), and for small regions of interest (e.g. Hansen et al, 2007, 2013; Koulakov et al 2015, 2016). Using a comprehensive dataset based on the dense Saudi National Seismic Network (SNSN) and measurements from neighboring countries, we developed three-dimensional model of the lithosphere and asthenosphere below Saudi Arabia.

In the first part of my talk, I discuss results from collaborative research at KAUST to discuss the shear-wave velocity structure of crust and upper mantle, potential asthenospheric flow, and the thickness of the mantle-transition zone underneath the Arabian Plate. These deep-seated structures have profound effects on the current volcanism in western Saudi Arabia and determine the heat-flow near the Earth surface, pinpointing to regions that may be conducive to harness geothermal energy.

The second part of my talk focuses on large strike-slip earthquakes along the East-Anatolian Fault and the Dead Sea Fault system. The devastating earthquake sequence of February 2023 helps to understand rupture dynamics and resulting near-fault shaking effects of super-shear ruptures on complex-geometry faults. At the southern end of the Dead Sea Failt the gulf of Aqaba has hosted the last major magnitude M > 7 event in 1995, but nearly segments have not ruptured for nearly 500 years. What does that mean for the Kingdom’s major infrastructural developments in northwestern Saudi Arabia, like NEOM and its associated mega-projects like “The Line”, and “Trojena”? In my talk, I will show results from dynamic rupture simulations of the 1995 Nuwaiba earthquake and other M > 7 scenario earthquakes in the Gulf of Aqaba to illustrate shaking and tsunami hazard for the region.