Tomographic Imaging Of Velocity Structure In Upper Crust Based On Correlated Inversion Of V-P And V-S

Wang Xiao,Zhou Xiao-Peng,Zhang Xin-Yan,Bai Zhi-Ming,Teng Ji-Wen

DOI: https://doi.org/10.6038/cjg20151011

2015-01-01

Abstract:Presently seismic tomographic methods based on first arrivals of deep seismic sounding traveltime data have been widely used to image the upper crustal velocity structure. However, generally there are some difficulties in the inversion process using conventional methods: (1) How to construct reliable upper crust models using Pg and Sg data; (2) How to quantitatively estimate the reliability of reconstructed models; and (3) How to decrease the uncertainty and distortion validly.Based on the idea that the V-P and V-S anomalies from a same geological body are closely related each other, we develop a correlated inversion method: V-P and V-S inversion are carried out alternately, in which the corresponding V-P/V-S model and initial V-S (or V-P) model are updated using the obtained V-P (or V-S) model after each inversion. The slowness perturbation resulted from the traveltime residuals is weighted and projected to the grid-points on the ray trace based on the updated correlation information, thus the velocity models are updated. Forward and inversion processes are respectively based on finite-difference calculations of travel times and the back-projection inversion method.Correlated inversion of V-P and V-S models and sequential inversion technique are incorporated into the tomographic process to ensure the reliability of inversion results and avoid the distortion of the resulting V-P/V-S model. Comparison with the conventional inversion results using the code of Ammon and Vida le (1993), both of the derived V-P and V-S models by our approach have been obviously improved, which are clearly closer to the original theoretical ones. While those constructed using the code of Ammon and Vida le (1993) have been seriously distorted. Moreover, the maximum V-P/V-S error based on the conventional approach is up to 0. 2, while the maximum V-P/V-S error with our approach is only 0. 02 similar to 0. 03.Tests using two typical models show that the technology used in inversion tomography of the velocity structure within upper crust, which can effectively improve the reliability of the inversion results, can avoid the uncertainty and distortion resulted from the conventional single V-P and V-S inversion process on a large scale. This method has been successfully applied to the western Tibetan plateau to reconstruct the velocity structure of upper crust beneath the Zhada-Quanshuigou deep seismic sounding profile. The derived models reveal the remained upper crustal velocity structure features resulted from plate collision in the western Tibetan plateau. Two similar obvious low-value anomalies of V-P, V-S and V-P/V-S with the maximum depth 8 similar to 10 km are present beneath the Zhada and Shiquanhe basin, which are probably the effect of the subduction of Indian plate beneath the Eurasian continent since Cenozoic or the closure of the Mesozoic Bangong-Nujiang oceanic basin. The disclosed Gar basin has a sedimentary thickness of 3 similar to 4 km, where the sedimentation and deep processes may be due to the gradual uplift of the Ayilariju Mountains and the strike-slip movement of the Karakoram fault.