MIGRATION TESTS

To test the migration result I chose the same two synthetics used in the previous section. Figure 10 shows the result of migrating the North Sea synthetic using at maximum four reference velocities. The top picture shows the result of using a conventional approach with four reference velocities plus salt. The bottom picture shows the result of using Lloyd's method with a maximum of four velocities. The Lloyd result reduced the number of reference velocities needed at some depth levels, reducing the overall migration 30% faster and higher quality than the result with four reference velocities at each depth. At location `A' the left edge of the salt structure is better placed. At locations `B' and `C' the bottom of the salt is better focused.

 

amoco-mig
Figure 10 Comparison between regularly sampling the velocity range (top panel) and using Lloyd's algorithm for velocity selection.At location `A' the left edge of the salt structure is better placed. At locations `B' and `C' the bottom of the salt is better focused.

Figure 11 shows the correct (left) and selected (right) velocity for the Sigsbee synthetic. Note how the salt velocity was correctly selected and the rest of the model well represented. Figure 12 shows the result of migrating the Gulf of Mexico synthetic. In this case for the standard approach I chose five+salt reference velocities. The Lloyd's algorithm selection method resulted in a 35% cost saving in compute time. The top picture shows the conventional approach while the bottom picture shows the result of using Lloyd's method for velocity selection. The bottom of the salt reflection, `A', is better focused. At `B' the right canyon is more accurately positioned and better focused. At location `C' we see how the point diffractors are collapsed using the Lloyd velocity selection method while using the more standard approach they are not collapsed.

 

zig-chosen
Figure 11 The correct (left) and selected (right) velocity for the sisbee synthetic. Note how the salt velocity was correctly selected and the rest of the model well represented.

 

zig-mig
Figure 12 Comparison between regularly sampling the velocity range (top panel) and using Lloyd's algorithm for velocity selection. The bottom of the salt reflection, `A', is better focused. At `B' the right canyon is more accurately positioned and better focused. At location `C' we see how the point difractors are collapsed using the Lloyd velocity selection method while using the more standard approach they are not collapsed.