/
Spotlight

Seismic Geometric Decomposition Technology: Cayros’ Unique Tool

Tue, 07/14/2020 - 09:35

Cayros’ Seismic Geometric Decomposition (SGD) technology reveals internal subtleties that cannot be resolved by legacy seismic. Its proprietary workflows, which can delineate both the geometry and the vertical extension of the reservoirs, will yield new seismic volumes to support, optimize and validate well planning and positioning. More detailed surface interpretation, detailed fault mapping and rock property volumes can be accommodated using this process. Processes can run both in the PostStack and in the Gather domain, and the technology has been proven in more than 150 fields from many basins around the world. It has helped to better understand the lateral and vertical complexities of the reservoirs and to improve both well positioning and production efficiency.

These proprietary workflows include:  

PCE (Principal Component Enhancement): Spectral Decomposition, along with deep learning algorithms, are used to generate a conditioned version of the original seismic in which random noise is reduced, vertical resolution is increased and lateral reflector continuity is improved. The deep learning algorithm is supervised by logs that are generated by applying an optimization function that minimizes the error between synthetics and seismic traces at well locations. PCE increases the vertical resolution of the original seismic by structural oriented filtering of each of the main bandwidths of the spectrum. Only the components (or nodes) that contribute the most to the stack are kept (through deep learning). The neural network is supervised by pseudo logs obtained via an optimization function that tries to match the seismic response to the P-impedance logs.

VD (Vector Decomposition): In this case, a different approach for Spectral Decomposition is applied. The original seismic trace is reconstructed by using a combination of bandwidths based on 21 frequency vectors (or components). An optimization function is obtained by stacking only the vectors that contribute the most to resolve the vertical variations observed in the P-impedance log. The result is a seismic vintage that increases the vertical resolution of the input seismic (PCE versions are used in general) by honoring its native phase. Again, the main original onsets are preserved by the process, meaning that prior surface interpretations might be used.

SGD High Frequency Enhancement: This is the result of applying a series of equations to delineate the internal geometry of seismic traces. A high frequency pseudo reflectivity volume is generated by the combination of complex trace attributes from the Hilbert transform. Later a convolutional model is constructed, wavelet frequencies are selected based on the best match with synthetic seismograms from wells. These types of vintages in general are used to fine-tune the reservoir segmentation and to optimize curvature analysis.   

SGD Layering: This is a seismic version to be used for internal reservoir pattern recognition. It might help to support well correlation. The SGD High Frequency is used as an input to a thin layer sparse spark inversion. The result is a pseudo reflectivity that captures subtle internal changes from original seismic traces. Truncations and lateral thickness variation can be better delineated. 

SGD Synthetic Reflectivity: This is a process that uses either linear regression or deep learning algorithms to honor synthetic seismograms from key wells. It is a seismic volume with relative amplitudes that might support both well correlation and detailed surface interpretation. The synthetic wavelet frequency threshold is selected based on the minimum dominant frequency required to resolve main vertical variations observed in P-impedance logs.

 

In the Mexican market, SGD technology has been successfully used in over 20 fields. Cayros has a proven track record of impeccable projects developed in a timely manner through the application of this patented technology.