In-Depth Proprietary Technology

Compressive Seismic Acquisition (CS-A)

Compressive Sensing (CS) is a modern approach to efficiently collecting seismic data in the field, capable of reducing the number of sources and receivers needed to acquire high-quality data OR enhancing the data with no additional sources/receivers.

Seismic data is highly compressible, that is, seismic data can be stored more efficiently than by storing the data as raw numbers along each trace. If the same amount of seismic information can be stored using far fewer bits, then it should also be possible for the same seismic information to be acquired using fewer bits.

Using this idea, researchers have been able to dramatically improve imaging across many fields and applications, including increasing the speed of medical MRI, CT, and PET scans (e.g. Compressed Sensing MRI Review) and improving optical imaging (e.g. Nature - Compressive Sensing of Large Images).

We apply this technique to seismic data by intelligently determining, in advance, which sources and receivers are redundant, containing information already extractible from other traces. To do this, we use our patented 2D MC-map to determine a subsampled survey design optimally tuned for reconstruction using our sparsity-promoting EPOCS solver.

Fig. 1. In-Depth’s proprietary Compressive Seismic Acquisition (CS-A) workflow.

How it works:

  1. Our clients provide their ideal desired survey design, and we return an CS-optimized design using a subset of those points, up to a desired subsampling factor (e.g. 70% sampling).

  2. During acquisition, we update the CS-optimized design for any new obstacles in real time.

  3. After acquiring the data, we process and reconstruct so that the interpretation products are virtually indistinguishable from the data that would have been acquired using the client’s original survey design.

You can use this approach to cut costs OR enhance resolution/fold!

Fig. 2. Our patented 2D MC-map solver takes into account the presence of obstacles beforehand, clustering points to ensure that every location of your survey can be properly reconstructed.

Fig. 4. Stack comparison of Conventional (left), Cost Saving CS-A (center), and Resolution CS-A (right), with the same dataset as in Figure 3. The conventional data was acquired using a 50 m grid design. Cost Saving CS-A used the same grid with 50% subsampling, reconstructing the original signal while enhancing S/N. Resolution CS-A used the same number of nodes as the conventional design, but reconstructed to a tighter 25 m grid, improving resolution and S/N.

Fig. 3. Comparison of Conventional (left), Cost Saving CS-A (center), and Resolution CS-A (right). The conventional data was acquired using a 50 m grid design. Cost Saving CS-A used the same grid with 50% subsampling, reconstructing the original signal while enhancing S/N. Resolution CS-A used the same number of nodes as the conventional design, but reconstructed to a tighter 25 m grid, improving resolution and S/N.

Fig. 5. The effect of different subsampling patterns on MRI images of the brain in k-space (similar to fk-domain) and image space. (Lustig, Michael et al. “Compressed Sensing MRI.” IEEE Signal Processing Magazine 25 (2008): 72-82.)