Topography effect on ambient noise tomography: a case study for the Longmen Shan area, eastern Tibetan Plateau

Ambient noise tomography (ANT) is a widely used method to obtain shear wave velocity structure in the crust and upper mantle. Usually, the topography is assumed to have negligible effect on the resulting models. This, however, might not be proper in regions with large topographic variation, such as plateau edges, submarine slopes and volcanic islands. In this study, we use synthetics from waveform-based numerical simulation to quantify the topography effect on ANT in the Longmen Shan area, eastern Tibetan Plateau margin. Three kinds of models are used in forward simulation to obtain theoretical waveforms, including Case1: the layered model, Case2: the layered model with topographic variation and Case3: the flattened model of Case2. The final inversion results show that the bias of ANT is negligible in the blocks with relatively flat topography, such as the interior regions of the Tibetan Plateau and the Sichuan Basin. However, for the Longmen Shan boundary zone with significant topographic variation (similar to 4 km), the shear wave velocity image has an obvious negative bias that can reach up to -4 per cent. The maximum depth of bias is similar to 5 km, which is mirrored with the maximum topographic elevation difference of the region, and the average bias disappears as the depth decreases to the surface (0 km) or increases to three times of the maximum influence depth (similar to 15 km). The horizontal distribution of the tomographic bias is almost linearly related to the topographic elevation difference with a slope of -1.04 and a correlation coefficient of 0.90 at maximum influence depth. According to this first-order correction formula and the decreasing trend of average bias with depth, the topography effect on ANT can be suppressed to a certain extent.