Receiver Function Principal Component Analysis and Its Application to Study the Crustal Structure of Central Sichuan Basin, SW China

The receiver function (RF) technique is an effective method of studying crustal structures. For a single station, features of the average 1-D crustal structure is usually derived by stacking all the radial RFs of that station, whereas small-scale variations (such as dipping discontinuities) or anisotropy of the crustal structure can be constrained through analysis of the waveform dependence on back azimuth of both radial and tangential RFs. However, the common presence of noise in real data often makes it difficult to extract information about structural variations directly from the RF records. In this study, we proposed a new method to achieve this goal by applying principal component analysis (PCA) to RF analysis for individual stations. This method (named RF-PCA) consists of two basic steps: 1) the set of RFs for one station is converted into a set of principal components that are unrelated with each other and reflect different features of the original RF dataset. 2) different principal components are selected according to the target of the study and used to reconstruct a new RF dataset to facilitate structural investigation. Our synthetic tests show the first (i.e., the largest) principal component of radial RFs reflects the common features of the radial RFs and thus can be used to recover the average 1-D crustal structure. The dominant principal components of radial RFs after removing the first one as well as the first principal component of tangential RFs reflect the dependence of the RFs on back azimuth and therefore can be used to constrain the small-scale variations or anisotropy of the crustal structure. Synthetic modeling results suggest our new method is effective in extracting crustal structural features, even in case of high noise levels of the RFs.

By applying the new RF-PCA method to real RF data, we investigated the crustal structure beneath a broadband seismic station (s233) in the central Sichuan Basin, SW China. Our result reveals the presence of two near-parallel dipping discontinuities within the crust beneath s233, probably suggesting a strong stratification of the basin crust. Compared with other seismic observations and geological data, we interpret the shallow dipping interface as the top boundary of the Precambrian crystalline basement, while the deep one corresponding to the Conrad discontinuity.