next up previous [pdf]

Next: Numerically blended synthetic data Up: Chen et al.: Deblending Previous: Comparison of sparsity-promoting transforms

Examples

In this section, in order to test the effectiveness of the proposed iterative framework defined in equation 5, we create three numerically blended examples with different kinds of shaping operations to demonstrate its applicability. The iterative framework in equation 5 is fundamentally a noise attenuation algorithm. The better an operator's ability to remove blending noise and preserve useful signal, the more appropriate it is for shaping. From this point of view, we first created two synthetic sections in different cases. One is a linear event case, with conflicting dip components, and the other is a hyperbolic event case, without crossing events. In order to deal with a more realistic and difficult situation, we created a more complex synthetic model (shown in Figure 1 and used previously for comparing the sparseness of different transforms) to test the proposed deblending algorithm. For the field data example, we use two common receiver gathers acquired by the OBN technique to generate a numerically blended data test.

The first three blended synthetic datasets (shown in Figures 3 to 11) show nearly perfect deblending results, particularly when the shaping operator is chosen as soft thresholding in the seislet domain. The third blended synthetic datasets shows an acceptable experiments are based on an OBN acquisition, as noted in the beginning of the paper. Thus, each seismic section in the examples corresponds to a common receiver gather. The first example is used simply for testing the denoising ability of each of the shaping operators in the case of crossing events. This example can also demonstrate the deblending performance for common offset gathers, where most seismic events are linear. For conciseness, we show the deblending performance for only one source.



Subsections
next up previous [pdf]

Next: Numerically blended synthetic data Up: Chen et al.: Deblending Previous: Comparison of sparsity-promoting transforms

2014-08-20