CASE STUDIES
Improving Seismic Data Quality on Waka Field
1439 km2
Survey Acreage
New Zealand Offshore
Region
24 Gb
Volume size
Regular noise is present in Waka Square, as well as areas with faults that are not fully and clearly traced. Post-stack functions are normally used to improve signal to noise ratio. Commonly post-stack processing involves a long choice of workflows and procedures and its time consuming to select the optimal settings.
Post-processing workflows can be complicated and time consuming, so Geoplat AI offers simple and more sophisticated functionality to improve the post-stack seismic data quality, based on the application of Machine Learning techniques.
Post-processing workflows can be complicated and time consuming, so Geoplat AI offers simple and more sophisticated functionality to improve the post-stack seismic data quality, based on the application of Machine Learning techniques.
No need for
parameterisation
parameterisation
Our algorithms allow you to improve the fault zones display, apply structure-oriented smoothing, restore attenuated reflections and suppress seismic interference noise. You will be able to generate noise attenuated seismic volume after running Geoplat AI’s workflow.
Seismic spectrum QC demonstrates below that the shape of the spectrum remains mostly unchanged, and the vertical resolution has been slightly decreased, while the signal-to-noise ratio increased more than 2 times.
Delineation of Salt Bodies
(F3 Seismic Survey)
(F3 Seismic Survey)
386 km2
Survey Acreage
North Sea Offshore
Region
546 Mb
Volume Size
Seismic pattern analysis on F3 benchmark volume shows that it is highly complicated due to intensive development of faults and disjunctives in addition to the major salt signatures present in the area. The upper layers of the salt build up are flexed into a significant anticline structure.
Study of pre-salt and
post-salt sediments
post-salt sediments
The goal was to trace salt bodies in the entire survey area using our advanced Machine Learning techniques as a salt body detection tool. Below you can see the salt body attribute volume (salt dome’s structural model) which was generated using Geoplat AI's salt delineation workflow.
Using our ML algorithms in Geoplat AI we were able to build a detailed model of the salt body in the entire study area considerably saving time and processing resources compared to the conventional auto-tracking techniques.
Automatic Horizon Interpretation
(Opunake Area)
(Opunake Area)
266 km2
Survey Acreage
New Zealand Offshore
Region
10 Gb
Volume Size
The Opunake area has varying fault structure complicated by a considerable decrease in the seismic data quality in the lower part of the profile. Due to these challenges, the surface tracking results could potentially be ambiguous and the conventional horizon tracking methods could have taken months to complete.
LGT includes
fault zones
fault zones
In order to interpret horizons across the entire area, we used our ML reflective horizon tracing tool. It generates a Local Geological Time volume (LGT) which is a colour-coded attribute with each colour representing a separate horizon. This attribute also preserves fault zones and can be useful for tracking faults on the same survey as well.
Below you can see a selection of individual stratigraphic horizons on the LGT volume. With Geoplat AI we have generated and extracted all stratigraphic horizons within a few hours compared to the conventional horizon tracking techniques that could have taken much longer time to complete.
Paleo Channels Discovery
(Tui-3D Area)
(Tui-3D Area)
680 km2
Survey Acreage
New Zealand Offshore
Region
2 Gb
Volume Size
Key channel elements were identified on the seismic data, located at different depths, and consisting of sand with clay injections. The goal was to clarify the spatial position and build the model of the detected channels.
Identification of the channel body system, located at different depths and with varying vertical thickness
We used Geoplat AI geobody detection tool to detect paleo channels within the entire volume. Most of the channel systems located at different depths were identified using our 3D attribute volume. Our innovative paleo channel interpretation workflow allowed us to obtain a detailed structural model of the Upper Jurassic channel bodies quickly and efficiently running on a standard workstation. The obtained level of detail and calculation speed were much more sophisticated compared to the conventional attribute-based interpretation workflow.
Automatic Fault Tracking
(Canning Area)
(Canning Area)
5067 km2
Survey Acreage
NW Australia Offshore
Region
109 Gb
Volume Size
Canning Area has significantly complex tectonic structure. One of the most challenging parts of the interpretation workflow was the presence of noise in the lower part of the profile. The conventional ways of interpreting faults on this data are difficult and hard to get an accurate result as most of the seismic structural attributes are sensitive to stratigraphic features and noisy areas of the profile. We have generated fault probability attribute based on our ML technology to minimise these factors.
The goal was to build a structural model with detailed fault surface identification.
The goal was to build a structural model with detailed fault surface identification.
The fault probability attribute was generated for the area which allowed us to successfully track the entire network of sub-meridional and sub-latitudinal faults in detail. Geoplat AI’s ML fault extraction workflow demonstrated good results in the low signal-to-noise ratio areas as well, and highlighted all major fault structures on vertical cross sections and time slices as demonstrated below.
Detailed fault model
in a few hours
in a few hours
You can automatically track and extract fault surfaces for the whole volume once the fault probability attribute is generated. Geoplat AI identifies faults according to the given probability values. You can also determine fault directions, their spatial extent and dip angles. It took just several hours to generate structural fault model for the entire seismic volume including the calculation of fault probability attribute (to highlight major faults). This was a significant reduction in time compared to the conventional fault interpretation workflows that could have taken months to complete.