The purpose of our first example is to investigate the accuracy of the solution of the constant- wave equation using the proposed low-rank scheme, in the presence of a sharp contrast in both velocity and . We use an isotropic two-layer model with in the top layer and in the bottom layer. The model is discretized on a grid, with a spatial sampling of along both and directions. An explosive source with a peak frequency of is located at the center. The reference frequency is . Wavefield snapshots are taken at . Figure 1a shows the acoustic case, in which the model has the velocity discontinuity but no attenuation (). For comparison, Figure 1b demonstrates the effect of homogeneous attenuation where . Both velocity dispersion and amplitude loss can be observed. In Figure 1c, we set in the top layer and in the bottom layer. The transmitted arrival exhibits less attenuation compared with that in Figure 1b. In Figure 1d, both velocity and remain the same as those in Figure 1c; however, the fractional power of Laplacians, , is taken to be the averaged value, which corresponds to the original implementation by Zhu and Harris (2014). To compare the results modeled by the two strategies, a middle trace at is extracted from both wavefield snapshots (Figures 1c and 1d). Figure 2 shows the two traces, along with their difference. Errors caused by using a constant instead of a spatially varying can be easily observed.
snap-0,snap-1,snap-3,snap-4
Figure 1. Viscoacoustic wave propagation in a two-layer model: (a) acoustic modeling with in top layer and in bottom layer; (b) same velocity as (a), homogeneous ; (c) same velocity as (a), in top layer and in bottom layer; (d) wavefield propagated using a constant averaged fractional power using same model as (c). |
---|
cross
Figure 2. Traces at extracted from wavefield snapshots and their difference. Red, long-dashed line corresponds to averaged ; blue, solid line corresponds to variable ; black, shot-dashed line is their difference. |
---|