Automatic 3D illumination-diagnosis method for large-N arrays: robust data scanner and machine-learning feature provider Chamarczuk, M.; Malinowski, M.; Nishitsuji, Yohei; Thorbecke, Jan Willem; Koivisto, E.; Heinonen, S.; Juurela, S.; Mężyk, M.; Draganov, Deyan ABSTRACTThe main issues related to passive-source reflection imaging with seismic interferometry are inadequate acquisition parameters for sufficient spatial wavefield sampling and vulnerability of surface arrays to the dominant influence of the omni-present surface-wave sources.Additionally, long recordings provide large data volumes that require robust and efficient processing methods. We address these problems by developing a two-step wavefield evaluation and detection method (TWEED) of body waves in recorded ambient noise. TWEED evaluates the spatio-temporal characteristics of noise recordings by simultaneous analysis of adjacent receiver lines. We test our method on synthetic data representing transient ambient-noise sources at the surface and in the deeper subsurface. We discriminate between basic types of seismic events by using three adjacent receiver lines. Subsequently, we apply TWEED to 600 hours of ambient noise acquired with ~1000-receiver array deployed over an active underground mine in Eastern Finland. We demonstrate detection of body-wave events related to mine blasts and other routine mining activities using a representative one-hour noise panel.Using TWEED, we successfully detect 1093 body-wave events in the full data set. To increase the computational efficiency, we use slowness parameters derived from the first step of TWEED as input to a support vector machine (SVM) algorithm. Using this approach, we detect 94 percent of the TWEED-evaluated body-wave events indicating the possibility to limit the illumination analysis to only one step and therefore increase the time efficiency at the price of lower detection rate. However, TWEED on a small volume of the recorded data followed by SVM on the rest of the data could be efficiently used for a quick and robust (real-time) scanning for body-wave energy in large data volumes for subsequent application of seismic interferometry for retrieval of reflections.

No abstract

Abstract. In NE Poland, Eastern European Craton (EEC) crust of Fennoscandian affinity is concealed under a Phanerozoic platform cover and penetrated by sparse, deep research wells. Most of the inferences regarding its structure rely on geophysical data. Until recently, this area was covered only by the wide-angle reflection and refraction (WARR) profiles, which show a relatively simple crustal structure with a typical three-layer cratonic crust. ION Geophysical PolandSPAN™ regional seismic programme data, acquired over the marginal part of the EEC in Poland, offered a unique opportunity to derive a detailed image of the deeper crust. Here, we apply extended correlation processing to a subset (∼950 km) of the PolandSPAN™ dataset located in NE Poland, which enabled us to extend the nominal record length of the acquired data from 12 to 22 s (∼60 km of depth). Our new processing revealed reflectivity patterns, which we primarily associate with the Paleoproterozoic crust formed during the Svekofennian (Svekobaltic) orogeny, that are similar to those observed along the BABEL and FIRE profiles in the Baltic Sea and Finland, respectively. We propose a mid- to lower-crustal, orogeny-normal lateral flow model to explain the occurrence of two sets of structures that can be collectively interpreted as kilometre-scale S–C′ shear zones. The structures define a penetrative deformation fabric invoking ductile extension of hot orogenic crust in a convergent setting. Localized reactivation of these structures provided conduits for subsequent emplacement of gabbroic magma that produced a Mesoproterozoic anorthosite–mangerite–charnockite–granite (AMCG) suite in NE Poland. Delamination of thickened orogenic lithosphere may have accounted for magmatic underplating and fractionation into the AMCG plutons. We also found sub-Moho dipping mantle reflectivity, which we tentatively explain as a signature of the crustal accretion during the Svekofennian orogeny. Later tectonic phases (e.g. Ediacaran rifting, Caledonian orogeny) did not leave a clear signature in the deeper crust; however, some of the subhorizontal reflectors below the basement, observed in the vicinity of the AMCG Mazury complex, can be alternatively linked with lower Carboniferous magmatism.

Passive seismic experiments have been proposed as a cost-effective and non-invasive alternative to controlled-source seismology, allowing body–wave reflections based on seismic interferometry principles to be retrieved. However, from the huge volume of the recorded ambient noise, only selected time periods (noise panels) are contributing constructively to the retrieval of reflections. We address the issue of automatic scanning of ambient noise data recorded by a large-N array in search of body–wave energy (body–wave events) utilizing a convolutional neural network (CNN). It consists of computing first both amplitude and frequency attribute values at each receiver station for all divided portions of the recorded signal (noise panels). The created 2-D attribute maps are then converted to images and used to extract spatial and temporal patterns associated with the body–wave energy present in the data to build binary CNN-based classifiers. The ensemble of two multi-headed CNN models trained separately on the frequency and amplitude attribute maps demonstrates better generalization ability than each of its participating networks. We also compare the prediction performance of our deep learning (DL) framework with a conventional machine learning (ML) algorithm called XGBoost. The DL-based solution applied to 240 h of ambient seismic noise data recorded by the Kylylahti array in Finland demonstrates high detection accuracy and the superiority over the ML-based one. The ensemble of CNN-based models managed to find almost three times more verified body–wave events in the full unlabelled dataset than it was provided at the training stage. Moreover, the high-level abstraction features extracted at the deeper convolution layers can be used to perform unsupervised clustering of the classified panels with respect to their visual characteristics.