Shaping the Future of Seismic Interpretation: View and Do

PreStack Pro, Sharp Reflection's unified approach to pre-stack data analysis, combines data processing, visualisation and interpretation in a single platform to greatly improve workflow efficiency and the accuracy of subsequent seismic interpretation.
This article appeared in November, 2020


Shaping the Future of Seismic Interpretation: View and Do

More data, more often, and in more detail. It is a simple premise, but one that has driven Sharp Reflections to become a market-leader in pre-stack data analysis software and services. Norway’s largest offshore operators now use its software in-house and the company’s international client list continues to grow.

The catalyst for this expansion has been a three-word mantra: ‘View and Do.’

The ‘view’ centres on visualising and quality control (QC). The ‘do’ refines the processing and analyses the data in the same workspace, improving workflow integration and minimising movement of data in and out of different specialist software applications.

Sharp’s data science mission is to create a one-stop platform for seismic data analysis.

And they are doing just that.

To Stack or not to Stack?

Most interpreters work with stacked data and never see the processed data in its final pre-stack form. As such, they may not be aware of the dense array of traces that are acquired to deliver the stacked image.

Processors have long appreciated the power of the stack, which reinforces signal and cancels noise to produce high-quality seismic images. However, while stacking is great for structural and stratigraphic interpretation, it must be used sparingly for seismic rock and fluid prediction, because traces are not truly redundant, with seismic amplitudes varying systematically as a function of distance between source and receiver. That information provides a so-called ‘DNA fingerprint’ that geophysicists must decipher. To preserve these subtle signals, more noise must be removed before stacking. These cleaner pre-stack volumes are then used to calculate a collection of angle stacks and to generate pre-stack seismic attributes that are even more sensitive to noise.

It sounds easy, but there is a catch: by removing too much noise, it is also possible to remove some or all of the primary signal, thereby compromising the data itself – which is why it is so important to get the approach just right.

A Best-in-Class Solution

Clients need their gathers to assess data quality and amplitude reliability. Sharp Reflections has developed software that was designed to give the interpreter the tools needed to make these critical decisions themselves.

For Sharp, three elements provide a best-in-class solution:

  • Scalable compute power: to handle the much higher trace count, multi-terabyte datasets required for large-scale pre-stack work;
  • Interactive visualisation: to quickly inspect pre-stack data anywhere within the 3D volume;
  • Calibration: so that processed gathers can be compared to synthetic seismic models created from logs at well locations.

Sharp’s unified approach, with data processing, visualisation and interpretation combined in a single platform, improves the workflow efficiency dramatically by making it possible to assess the impact of processing efforts directly on the interpreted results. Interpreters and processors can collaborate to optimise data for specific targets, helping to better delineate reservoirs and improve fluid detection.

Meeting Industry Challenges Head On

Sharp Reflections understands there are several challenges for the full adoption of pre-stack data which it has looked to address head on.

Firstly, pre-stack seismic survey datasets are large and growing exponentially. Regular workstations and software cannot always digest the data, severely limiting the ability to work with the full data stream. However, Sharp’s high-performance computing platform provides a scalable, highly interactive solution for even the largest of datasets.

As seismic becomes increasingly commoditised, typical data sources are often not being processed to the standards required for reliable fluid and rock property prediction. Interpreting on static, stacked images, that often mask data quality defects, can lead to incorrect ranking of prospects and inaccurate pre-drill decisions. PreStack Pro fully integrates rigorous QC and pre-stack data conditioning directly into the analysis workflow to deliver superior seismic reservoir characterisation results – there is no need to compromise on data quality. By working with pre-stack data, interpreters can clean and condition data before stacking, improving reliability and confidence in subsequent interpretations.

  • The PreStack Pro integrated model.

Working Together to Create Value

Large regional 3D seismic datasets are ideal proving grounds for the PreStack Pro technology. They offer significant volumes of data and cover large areas, often being used for regional-scale exploration work. However, as it is not possible to optimise data for any depth or reservoir, data is often processed conservatively, with a reliance on stacking to reduce noise.

The ION Picanha programme represents a genuine ‘Big Data’ challenge. To date, ION has reprocessed and depth imaged more than 50 interconnecting public-domain 3D surveys to provide a regionally calibrated and consistently imaged 3D dataset across the pre-salt basins in south-east Brazil. 

  • ION Picanha programme area (left, red polygon) and Sharp Reflections 17,000 km2 area for AVO-consistent gather conditioning study (right, red polygon).

Sharp and ION worked closely together to conduct rigorous quality control of the seismic processing, and to develop an optional processing sequence which is more suitable for quantitative work. The goal was to create a fast-track version of the conditioned seismic data that Sharp normally conducts on a bespoke basis for individual clients.

The approach can save clients time by eliminating the need for them to move incredibly large pre-stack datasets onto their in-house IT systems. Working in partnership to create shared value, ION provided access to the data and supplied the hardware, while Sharp contributed software and expertise.

Starting with ION’s raw migrated gathers, Sharp was able to generate AVO-consistent conditioned gathers and angle stacks with improved amplitude integrity, event continuity and resolution. These products were suitable for fast, regional AVO screening, as well as structural interpretation.

The data conditioning workflow reduced residual moveout and increased the signal-to-noise ratio through attenuation of remnant multiple and random noise. In turn, this improved the reliability of AVA attributes which are especially sensitive to these processing artefacts.

The results, when validated against available well information, also showed better matches to synthetic seismic models. In short, this meant that any volume and map attributes could now be computed over large areas with higher trust in data quality, allowing for confident exploration away from calibrated wells.

Sharp and ION are exploring other forms of business collaboration, including workshops to review the pre-stack data with clients and customised jobs in sub-volumes of the overall dataset. While these workflows currently require use of ION’s internal HPC computing resources, they might eventually be superseded by 100% cloud workflows in the future.

No compromise on data quality: Selected results from the Picanha pre-stack study illustrate the advantages of Sharp’s integrated approach to data quality improvement and assurance on a massive 17,000 km2 area.

Health Checks Via Volumetric and Map QCs

Visualisation goes beyond simple viewing of the offset gathers, by incorporating quality assurance maps that can be quickly generated at specific reservoir objectives.

The software allows the user to create attribute QC maps from either pre-stack gathers or angle stacks. These QCs are health checks that provide a detailed analysis of the data quality at the reservoir level. Insights are then gained into the types of issues the data can have (noise level, event flatness, bandwidth, etc.) and how they vary regionally. Using threshold criteria, the health checks highlight the AVO fidelity of the data and areas where the conditioning can improve the data quality. Easy to perform in PreStack Pro, these QCs can be repeated at different stages of the conditioning to ensure convergence on an optimal solution.

  • Noise Level QC map, extracted along the Upper Cretaceous horizon, before (left) and after (mid) conditioning. Lower values (warm colours) indicate noisy areas and higher values (cool colours) indicate clean areas. Health check metrics (right) are also assessed and summarised to provide insights on which aspects of the data can be improved.

The software provides instantaneous point-and-click synchronisation between QC maps and pre-stack gathers to help investigate causes of spatial variation in data quality at any location. Health checks also include localised QC at the well location by comparing synthetic to seismic attributes. Sharp conducts both micro and macro-level quality assurance, which helps drive processing decisions for down-the-line data analysis.

Interactive Noise Removal

Processed gathers often contain remnant coherent and random noise that interferes with key reservoir reflections. To counter this, Sharp offer high-quality algorithms that remove residual noise and enhance continuity of gathers and stacks. On many datasets, this second processing pass is required to coax out convincing differences between water and oil reservoirs updip.

Interbed multiples are particularly challenging, as they often manifest as strong events with very little differential moveout. PreStack Pro’s high-resolution Radon transform can accurately target these reflections, without the primary leakage seen with many other conventional transforms. It provides an interactive mute tool to confidently remove these events, as well as other forms of dipping or aliased noise remaining after data migration.

The software also boasts an array of image processing techniques for removing random noise from 2D and 3D datasets. 2D anisotropic diffusion filters reduce offset-dependent noise before stacking without distorting wavelets, while 3D dip/azimuth-steered median and diffusion filters dramatically reduce random noise on stacks and speed automated interpretation. PreStack Pro’s modern parallel compute engine has the capability needed to filter each offset class independently and reassemble as noise-cancelled gathers.

  • Pre-stack gathers before and after conditioning. Parabolic, downward-trending multiple noise is attenuated, revealing previously obfuscated flat primary reflectors underneath.

Innovative, Integrated Interpretation

Better data is a requirement, but not the end goal.

Sharp’s focus has always been for clients to drill the absolute best prospects and wells. To shorten the decision time, they remain committed to helping clients to understand how changes to gathers are reflected in the stack, volume attributes, and amplitude maps.

Traditionally, this would require transfer of data from the processing to interpretation system, creating bottlenecks and discouraging workflow iteration. Today, all of this can be done in PreStack Pro.

Workflows automate the generation of all standard interpretation volumes, so interpreters can see how changes to the gathers impact the stack. AVA attributes and other quantitative interpretation attributes are especially sensitive to gather quality and can be tracked through the processing sequence.

  • While there are minor data quality improvements observed on the far angle stack after conditioning, the impact on derived AVA attributes such as pseudo-Vp/Vs (lithology indicator) is much more appreciable. In the lower right panel, the reservoir sands (yellow colour at centre) are better delineated and more interpretable after false background anomalies, caused by noise and residual moveout, are removed via conditioning.

PreStack Pro offers a full set of tools to track events on stacked volumes and can be used to create amplitude maps and AVA attributes in just a few mouse clicks.

  • Spectral decomposition of the Upper Cretaceous section extracted from 17,000 km2 of conditioned stack data, highlighting paleo-depositional systems and their relationship to down-dip salt evacuation. This extraction can be performed in seconds over large areas using PreStack Pro.

Pre-stack horizons can then be created from these stacked interpretations and used to extract high-resolution amplitude attributes that utilise 100% of the data.

Sharp’s scalable compute solution also reduces the time required to generate compute-intensive attributes, such as those used for spectral decomposition. The computations are carried out interactively in map view, without the need to pre-compute and store full 3D volumes at key seismic frequencies.

Time is of the essence and PreStack Pro can take initial data to develop a portfolio of promising amplitude leads in just a few short days.

Simplifying the Lives of Clients

The core applications, processes, and analytics within most companies are fragmented. Modern data and analytics platforms are becoming increasingly unified in many industries, as part of a push for increased digitalisation and better workflow efficiency.

Sharp Reflections is riding this wave to create an integrated Big Data platform for advanced seismic analysis. Its software is driving accelerated uptake of full-fidelity seismic by making life easier for clients across the globe who work with these dense datasets.


Related Articles