Cloud and Digitalization

Revolutionizing Seismic Data Transfer from Sensor to Client

wave imagewave image
PGS has adopted the cloud and digitalization to revolutionize the methods by which we obtain, process, and utilize seismic data and accelerate the flow of data from sensor to customer.

Transforming a traditional industrial player into a cloud-native, data-driven energy data company is a monumental journey, one that PGS embarked on in 2020. Faced with ageing on-premises equipment that required substantial capital expenditure, the company found itself at a crossroads, especially coming out of a challenging economic downturn.

In this context, a pivotal decision was made to leverage the need for equipment renewal as an opportunity to look at our entire seismic data flow (from sensor to customer) and overhaul our IT and high-performance computing (HPC) landscape. The primary goal was to create a sustainable, cost-effective, and scalable solution that would not only future-proof the company but also deliver added value to our customers.

The ultimate objective was the establishment of a sustainable and future-proof data ecosystem, ensuring the optimal flow of high-quality subsurface data with maximum efficiency, economy, and minimal environmental impact.

The following article will provide more insight into the key projects and explore the changes we had to apply to the underlying platforms.

Accelerating Seismic Data Transfer from Sensor to Cloud

Traditional data flow (top) vs. optimized data flow (bottom). The traditional flow contains many data duplications and physical data handover points. The PGS optimized data flow transfers the subsurface information as soon as possible via satellite into the cloud for QC, processing, and final delivery.

The initial phase in the seismic data journey in PGS generally starts after the seismic data has been recorded by the streamer spread and the onboard recording system. After initial quality control (QC) of both the data and the associated navigation information, it is stored on the vessel until it is physically shipped to ours or our client’s onshore facility for further data conditioning and imaging (top row in the illustration above). Historically, this workflow has remained unchanged due to insufficient satellite bandwidth to facilitate immediate data transfer of the seismic data, despite having internet connectivity via satellite on all vessels for decades.

However, this paradigm has now been disrupted with the arrival of Low Earth Orbit (LEO) satellite technology. LEO satellites provide a compelling alternative to traditional geostationary (GEO) satellites, offering enhanced data transmission capabilities and reduced latency and enabling the optimized flow of data shown in the illustration above as the bottom row. 


Left, LEO and GEO orbits relative to earth. Right, Starlink (front) and VSAT dome (back) satellite antenna onboard the Ramform Vanguard seismic vessel.

Leveraging the cutting-edge capabilities of LEO satellites, PGS conducted a series of tests in 2023. These tests demonstrated the technology’s exceptional effectiveness, with PGS successfully transmitting full-integrity 4D seismic data from two surveys directly to the cloud. This eliminated the need for physical data transfer, revolutionizing the speed and efficiency of seismic data transfer. The data delivery time was reduced from an average of nine days to just one day (see graphic below).

Following these successful tests and recognizing the transformative potential of LEO satellite technology, a strategic decision was taken to implement a base Starlink (LEO) service level alongside our existing VSAT service (GEO). For the time being, we treat both services as complimentary to each other based on available bandwidth, latency and commercial terms depending on the use case.

The successful validation of the new satellite technology provides a huge opportunity for PGS and the seismic industry, both regarding how we provide clients with the acquired data but also on how we enable them to follow the project during the acquisition phase. It has the potential to perform some of the tasks onshore and gives the onboard crew a better connection to the onshore world.

The future success of this technology in our industry depends on how we, as individuals, companies, and as an industry, adapt our traditional processes. Nevertheless, PGS embraces this technological change and is committed to driving sustainable seismic data transfer, ultimately enhancing the efficiency and effectiveness of seismic operations worldwide.

Key Benefits of Sending Seismic Data Via Satellite

Key benefits of sending seismic data via satellite link rather than physical media transport.


A New Dawn for Handling Huge Imaging Workloads

After the seismic data have been acquired offshore and moved onshore the next step is to produce a 3D image of the subsurface. This typically requires multiple steps using different algorithms, often with significant use of high-performance compute (HPC). The HPC workload typically comprises large datasets (>1 TB and <100 TB) but a small number of files within each project (< 1000). Before 2019, our imaging software stack, comprising 300+ algorithms, utilized around 200 000 cores of Cray hardware and a 70 PB online parallel file system. Many of the most HPC-intensive codes were written specifically for this on-premises hardware and orchestrated by a heavily homegrown and highly customized job management system as well as proprietary and self-supported 3D data visualization.

Initially, our cloud approach aimed at a hybrid environment, prioritizing on-premises compute and utilizing cloud compute for excess capacity via a ‘Lift and Shift’ strategy. This meant trying to move on-premises applications to the cloud without redesigning them, which proved unsuccessful in terms of user experience and cost.

A fundamental reassessment was essential, extending beyond technical considerations to encompass our overall approach to the project. We started to work closer with our partner Google, reviewing our project set-up as well as our strategic short- and long-term goals.

The goal was to decommission the Cray platform by mid-2022, focusing on cloud scalability, storage optimization, and platform independence. The move to Google Kubernetes proved transformative, giving the scalability and stability we needed (see graphic below).

The use of GKE (Google Kubernetes Engine) instances in late 2022 and 2023 for our cloud-based imaging. The graph shows the spikiness of usage, at peak points we are using far more capacity than is ever available to us on-premises. However, outside these times we do not pay for any unused capacity.

Key Benefits of Moving Processing to the Cloud

  • Scalability and flexibility: In autumn 2022 we managed to sustain a peak of 1.2 million vCPUs across 12 GKE clusters, three times greater capacity than we previously had access to, allowing us to compete for larger jobs and run tasks in parallel. We can now routinely push through the 1 million vCPU barrier.
  • Improved turnaround time: Tailoring compute for each large run, enjoying virtually unlimited capacity to run jobs at scale and in parallel, and leveraging the latest software and hardware stack significantly reduces compute time, from weeks to hours in extreme cases.
  • Reduced exposure: Outsourcing tasks such as procuring equipment, maintaining the computer center and negotiating power supply minimized our risk exposure associated with running a HPC compute center.
  • Greener compute: Our selected data center with the preferred cloud provider runs on 100% renewable energy, contributing to our sustainability ambitions.
  • Access new levels of geophysics: Cloud scalability facilitates the development of new algorithms, such as those using elastic wave propagation, by providing short-term access to compute resources otherwise unaffordable for long-term use.

Illustrating Success

  • A large reverse time migration (RTM) on an OBN dataset which would have taken 6-8 weeks using our entire legacy on-premises compute capacity took 14 days in the cloud.
  • A multi-azimuth streamer RTM which would have taken 40 days using our entire compute capacity was completed in 26 days in the cloud.

Accelerated Data Access with Cloud-based Seismic Storage

The final leg of our journey involves delivering the final imaged data to our clients, either as part of contracted work or through our MultiClient data library for widespread access. Traditionally, this data transfer has occurred via physical media (e.g. tapes), involving multiple phases of global logistics (as illustrated in the first illustration in this article). This created a high risk of data loss, required many points of data duplication and was extremely time-consuming.

Recognizing an opportunity to revolutionize subsurface data management, PGS, as the owner of one of the world’s largest MultiClient libraries, in 2019 started to move our MultiClient library to the cloud. The goal was to eliminate data duplication, streamline data management through automation, and expedite data delivery. Even though the cloud offers nearly infinite storage capacity with ubiquitous, secure, and granularly controlled access, like our processing in the cloud story above, a mere lift-and-shift approach wouldn’t have addressed all the problems, like consistent data conditioning, client trust, or reduction of data duplication.

By late 2022, PGS implemented fully functioning and cloud-native data ingestion pipelines, automating the reading, conditioning, ingesting, and indexing of post-stack and prestack data from cloud storage. A well- documented connection endpoint (API) and Software Development Kit (SDK) were made available to clients for 24/7 trace and metadata access. This empowered our sales department to explore diverse commercial models with clients for accessing seismic data.

High-level process flow for making seismic data available in PGS’ cloud solution, providing users fast and secure cloud-based access to contextualised and quality-controlled subsurface data.

Key Benefits of Cloud-based Data Storage and Access

  • Improved turnaround time: Reducing access time from weeks to less than an hour 
  • Direct access to data in the cloud at scale: Data is available via Open Subsurface Data Universe (OSDU)-aligned APIs for further sharing, collaboration, and interoperability within a client’s own environment or within the industry.OSDU is becoming an open-source industry standard for storage, retrieval, and sharing
  • New commercial offers: This solution allows us to condition, ingest and connect our client’s proprietary data, providing a state-of-the-art managed cloud-based solution for storage, retrieval, and seismic data sharing. This enables more rapid and informed exploration and development decisions.

A Sustainable Future-Proof Data Ecosystem 

Confronted by aging infrastructure and economic hurdles, PGS seized the opportunity to revolutionize our seismic data flow, information technology, and high-performance computing landscape to transform into one of the industry’s leading cloud-native, data-driven energy companies.

A key learning was that the challenge extended beyond mere technical platform upgrades. It necessitated a profound shift in our workforce mindset. By embracing new project methodologies and technologies, integrating greater business participation into projects, scrutinizing and revising numerous workflows, and persuading stakeholders that the cloud provides substantial business potential.

Embarking on this initiative, we envisioned an exceptionally ambitious future, uncertain of its technical feasibility at the outset. The initiative revolutionizes the methods by which we obtain, process, and utilize seismic data optimizing data flow while minimizing environmental impact.