Deep learning for quality control of receiver functions

Abstract

Receiver function has been routinely used for studying the discontinuity structure in the crust and upper mantle. The manual quality control of receiver functions, which plays a key role in high-quality data selection and accurate structural imaging, has been challenged by today’s booming data volumes. Traditional automatic quality control methods usually require tuning hyperparameters and fail to generalize to low signal-to-noise ratio data. Deep learning has been increasingly used to deal with extensive seismic data. However, it generally requires a manually labeled dataset, and its performance is highly related to the network design. In this study, we develop and compare four different deep learning network designs with manual and traditional quality control methods using 53293 receiver functions from three broadband seismic stations. Our results show that a combination of convolutional and long-short memory layers achieves the best performance of ∼91% accuracy. We also propose a fully automatic training schema that requires zero manually labeled receiver function yet achieves similar performance to that using carefully labeled ones. Compared with the traditional automatic method, our model retrieves ∼5 times more reliable receiver functions from relatively small earthquakes with magnitudes between 5.0 and 5.5. The average waveforms and H-κ stacking results of these receiver functions are comparable to those obtained by manual quality control from earthquakes with magnitudes larger than 5.5, which further demonstrates the validity of our method and indicates its potential for making use of smaller earthquakes in the receiver function analysis.

A New Method to Estimate Slab Dip Direction Using Receiver Functions and Its Application in Revealing Slab Geometry and a Diffuse Plate Boundary Beneath Sumatra

“…As shown in Figure S4 in Supporting Information S1 and Figure 3g, polarity changes, decreases in amplitude, and presence of strong anisotropy will significantly affect the two‐lobed waveform variation in back azimuth, which can be noticed by visual inspection. Visual inspection or AI‐based sophisticated analysis on RF waveforms (e.g., Gong et al., 2022) is thus recommended as a necessary procedure for data quality‐control.…”

Crustal Imaging with Noisy Teleseismic Receiver Functions Using Sparse Radon Transform

“…This may render the modeling step intractable, hampering robust estimation of the crustal parameters and the subsequent interpretation of crustal composition (Zandt & Ammon, 1995;Stankiewicz et al, 2002;Audet et al, 2009;He et al, 2013). For this reason, seismic analysts usually employ a variety of quality control procedures to select high-quality receiver functions, either manually or in an automated manner, e.g., using a combination of attributes from deconvolution, waveform features, and stacking statistics (Yang et al, 2016), or through supervised machine-learning models (Gong et al, 2022). Previous studies have also made several modifications to grid-search algorithms in an effort to improve the constraints from the low-amplitude reflections, including, but not limited to, using cluster analysis and semblance weighting (Philip Crotwell & Owens, 2005;Eaton et al, 2006), varying weighting factors for different phases (Vanacore et al, 2013), and performing moveout corrections preceding the grid-search (Rivadeneyra-Vera et al, 2019).…”