The simple objective of the DUG Deblend inversion algorithm is to explain all the input data as unblended records so that when re-blended they will accurately reproduce the input. We have developed an industry-leading inversion-based algorithm that performs exceptionally well to recover both weaker and stronger parts of the wavefield even in the most extreme of circumstances. In addition, DUG Deblend is able to handle seismic interference so that multiple acquisition campaigns can operate simultaneously, further increasing efficiency.
This land data set from onshore Egypt was acquired in a blended manner with up to 30 vibroseis units in operation at any one time. Each vibroseis unit worked on a separate source line, each over 20 km long resulting in a highly variable separation of sources. Subsequently, a source line was reacquired in a non-blended manner, absent of interfering shot energy to serve as a benchmark for deblending.
The results (right) compare the blended input data on the left, the DUG Deblend result in the centre and the benchmark, unblended data on the right. The non-blended dataset provides a powerful comparison highlighting the efficacy of our deblending solution. The deblended result could be argued to be an improvement over the unblended result due to the increase in signal-to-noise ratio that comes naturally from inversion-based deblending.
The slices (left), taken through cross-spreads filtered to less than 6 Hz demonstrate the quality of signal recovery even at the very lowest frequencies. The success of the comparison lead independent observers to conclude that the DUG Deblend result was “almost indistinguishable from the perfect benchmark”.
Efficient OBN acquisition requires a multi-vessel, multi-source set-up resulting in data that requires deblending. In this example from a very large OBN survey in the North Sea, three vessels each with three sources were shooting within 15 km of each other. There are at least 15 distinct shots present in each record. DUG Deblend is able to effortlessly handle the most complex of blended data sets to uncover the weakest of signals – such as the diving wave energy visible on the far offsets of these records.
In addition to multiple sources and multiple source vessels, this data set also included significant seismic interference from nearby acquisition campaigns. This interference is simply another form of blended energy and as such can be included as part of the deblending process using acquisition shot timing information from the nearby vessels. DUG Deblend is able to simultaneously handle intrinsic blended energy as well as extrinsic interference.