“Learn to identify and accept change early”

KEY SEISMIC SOLUTIONS LTD
INTERVIEW COORDINATED BY SATINDER CHOPRA

Tobin Marchand is an experienced geophysicist who specializes in the processing of both conventional P-wave and multicomponent surface seismic data. In 2005, after getting a master’s degree in applied mathematics from McGill University in Montreal he arrived in Calgary. After some onboard training and several field deployments, he started processing seismic data at VeritasDGC (now CGG). In 2011, he joined Key Seismic Solutions Ltd. in Calgary, where he assisted in finalizing the multicomponent product line and was the Multicomponent Group Leader for several years. Tobin is now the Vice-President Technology at Key Seismic supporting the development of new initiatives and the refinement of existing technologies.

The RECORDER requested Tobin for an interview to share his experiences and views on a wide range of topics, to which he sportingly agreed, much to our delight. Following are excerpts from that interview.

Tobin, let us begin by asking you about your educational qualifications and your work experience.

My post-secondary academic studies started with a joint honours bachelor’s degree in Mathematics and Physics at McGill University in Montreal. I then completed a master’s degree in applied mathematics in 2005, also at McGill, with the goal to teach calculus at the Quebec collegiate level. In the fall of 2005, my wife and I moved from Montreal to Calgary in favour of better economic prospects. After the move, it became apparent that without an education degree or a doctorate degree, it would be very challenging to secure a position teaching calculus outside of the Quebec educational system. I then pivoted and started applying to any job posting that was related to my scientific interests. The following February, I interviewed for a junior position in the seismic processing department at VeritasDGC (now CGG) and was hired on at the start of March 2006. At the time, VeritasDGC had an amazing training program that included introductions to various aspects of seismic processing (data reformat, frontend processing, seismic processing & reservoir characterization) and rotations with field crews in North-East British Columbia and on the Alaskan North Slope that provided me exposure to field acquisition. After nine months of onboard training and several field deployments, I settled into a processing group and began to accumulate the experiences that kick-started my career.

You started working for VeritasDGC (now CGG) in early 2006 as a seismic processor, and then switched over to the processing of multicomponent seismic data. How different was that experience when dealing with both PP and PS data?

One big difference, compared to my experiences up until then, is how much bigger the thought space needs to be when discussing the shear components. Conventional PP processing can be thought of as a stand-alone project, whereas PS processing relies on multiple contributions from the PP processing. Moreover, some basic assumptions that are acceptable for PP seismic processing need to be reconsidered in order to handle the converted wave data. These additional complications reinforced the geophysics I had learned and motivated me to keep learning more. Another notable difference is how much more client communication and involvement is needed to meet the goals of a PP/PS project.

How did you decide to move over to Key Seismic Solutions Ltd. in 2011?

The opportunity to join Key Seismic Solutions Ltd. (Key Seismic) came about at the end of summer 2011. I had a few exploratory meetings with Key Seismic managers in which my potential role with the company was discussed. They were looking for someone with multicomponent experience to help guide the final developments of their PS product line, and after, to lead the group that would process the PS data. Through these early exchanges, I discovered that the company was highly technical and had a clear focus for its research and development investments. After 5 years in the industry, this was an incredibly exciting break for my career advancement. An additional incentive was that at first, I would be working closely with Dr. Richard Bale, with whom I had enjoyed working previously at CGGVeritas (now CGG). I had an upcoming September vacation planned where I made the decision to seize this opportunity. When I returned to Calgary, I handed in my resignation and started my new adventure with Key Seismic at the beginning of October.

You have been engaged in processing of regular seismic data first and then multicomponent seismic data, both at CGG and Key Seismic. Why the preference to process seismic data and not get into its interpretation or reservoir characterization?

There is no real preference per se. It just so happens that my gateway into geophysics was as a processor at a time when I was unsuccessful at landing a job teaching calculus. Through training, both “in house” and through CSEG & SEG, I quickly developed an appreciation for geophysics as a real-world application of theoretical physics and math that had been my academic path. Through dedication, perseverance, and a bit of luck, I have been able to take on more challenging positions and progress my career to where I am today. Although my exposure to other geophysical disciplines has been limited, I find them interesting and would have been equally dedicated to them if I had started out differently.

BikeDay: Discovering Mount Rigaud bike trails, Rigaud, Quebec.

You are now the VP, Technology at Key Seismic. What kind of a role does this position entail, and what challenges do you come across in discharging your duties?

As a technical company, Key Seismic strives to stay current on important industry developments. As VP of Technology, this means being involved in investigating, planning, and deploying our research and development activities. This also means I work with our IT and software teams to ensure optimal utilization of our computer facilities. Another important aspect of my role is to maintain procedures of best practice in order to continually deliver our best products. Furthermore, time permitting, I involve myself in data processing.

In the big picture, the most challenging responsibility of my position is to understand the intricate interdependencies of a processing system and to prioritize actions in order to maximize the positive impact while minimizing any possible negative effects.

I have been VP Technology for just over two years and my enjoyment of the role has only increased over time.

What in your opinion is the promise of multicomponent seismic data, i.e., why should oil companies spend more on the acquisition and processing of multicomponent seismic data? How much more value can an interpreter expect to get from multicomponent seismic data than just the vertical component?

A question I get asked often…

The technical arguments for using PS data more regularly are many. Firstly, knowing that the AVO curves of PP and PS are different and mutually independent is an important starting point. Secondly, understanding that due to the asymmetric ray paths of PS waves, for the same surface positions of source and receiver, the incident angles sampled by the PS are different, notably larger, than those of PP waves. Thirdly, so much of the inversion work that is carried out on processed seismic data is statistical in nature. Together, these three statements imply that from a technical point of view, recording, processing and analyzing PS data along with PP data is a means of doubling the usable data pool for reservoir characterization. Consequently, joint PP-PS inversions typically yield more accurate density inversions and better sensitivity to fluid effects. Lastly, for thermal plays, time lapse monitoring of converted wave data can yield information regarding the heat front generated by steam injection and can help delineate zones of mobilized bitumen. As bonus, a by-product of 3D PS processing is azimuthal anisotropy information about local stress and/or fractures that result from shear wave splitting analysis.

I am aware of cases in which the benefits of using PS data are appreciable. However, the valuation of the uplift provided by incorporating multicomponent data is ultimately dependent on our clients’ particular principles and is not universal to all scenarios.

Multicomponent seismic data has been around for the last three decades now, and still not many oil companies go in for it. Even if they acquire multicomponent seismic data, they just process the vertical component data. Why do you think the acceptance of multicomponent seismic data has been a gradual and slow process?

I think that many companies and geophysicists are interested in using multicomponent data more often than they are able to pursue it for various reasons. A non-exhaustive list of reasons would include time constraints, restrictive budgets, availability of 3C equipment and trained acquisition crews, unfamiliarity with multicomponent workflows, and availability of project resources to make use of the processed data once delivered. Over time, the demand for multicomponent seismic work has strengthened and weakened in intervals, likely linked to the economic health of the industry. The last time that multicomponent processing was on the rise was back in 2014; I doubt this is coincidental. I am optimistic, however, that with the strengthening of commodity prices and the increased emphasis on investment returns, there will be improved interest in multicomponent processing in the future.

DogSled: Training Maple, our snow dog, to pull the home-made dog kick sled.

The wavefield recorded by the vertical geophones planted on the surface of the Earth record both the P-P data as well as SV – P data, the former referring to a downgoing P-wave reflected as P-wave, and the latter to a downgoing SV wave reflected as a P-wave. What this also implies is that the seismic data that we have traditionally been acquiring, has both the P-P and SV – P data. But we have always processed it for P-P data. Bob Hardage had spearheaded the separation and development of the SV¬–P data from the conventional P-P data recorded in our legacy surveys and the creation of SV– P images for our target reservoir intervals. He demonstrated his work at the 2017 SEG conference and some of his articles. What is your experience in trying to retrieve the SV–P data from the conventional P-P data?

First, some technical points. It is a convenient and mostly accurate assumption that the vertical component measures the reflections of compressive waves and that the horizontal components measure shear waves. However, it is important to recognize that in general, the data recorded on any receiver component has mixed modes of seismic data. I have seen datasets in which the horizontal components were inundated with compressive information to the point of being able to produce a respectable PP section. I have also seen datasets in which the vertical component is riddled with PS reflections masquerading as PP multiples.

Now to answer the question. I have no experience with SV-P separation from a vertical component dataset. However, I have had the opportunity to experience mode separation on data recorded through permafrost. In this environment, the returning waves at the surface are not propagating near the vertical axis so that both propagating modes are mixed across all components of a three-component receiver. The technology I was using was actively being developed by Dr. Richard Bale (US patent 20150338536A1). Witnessing the invention and refinement of the method was my first real exposure to research and development in a professional environment. My participation in the project gave me a better appreciation for the results our industry consistently extracts from surface seismic. It also made me aware of how potential information is left unexploited in seismic data.

From a scientific viewpoint, I wish there were more opportunities for service companies to explore and develop new ideas. Similar concepts that have piqued my curiosity include the imaging of multiples, PS refraction analysis and combining PP & PS to simulate SS data. I think these and many other concepts have merit, they simply need time and resources to make them viable.

FamilySki: Family Day weekend skiing at Lake Louise.

What are some of the trendy aspects of seismic data processing these days and what value-addition do they bring?

Acquisition planning along with its impact on data processing and data quality are some of today’s popular topics. This includes but is not limited to compressed sensing survey design, slip-sweep Vibroseis acquisition, simultaneous acquisition and the miniaturization of receivers and sources. These pursuits have an immediate benefit by reducing the disturbance to the natural environment in which seismic surveys are undertaken. In some instances, exploration and production companies are targeting the same quality of seismic information as from traditional acquisition techniques for less operational costs. In other instances, the goal is to attain improved results for relatively the same cost as from traditional acquisition techniques. This latter case is very interesting from a data science point of view as we are paying closer attention to subtle details in the data that have previously been undersampled.

Another prevalent trend is cross discipline collaboration. Joint efforts across the disciplines of survey planning, acquisition, processing, interpretation, inversion, and engineering are more frequent in the mutual pursuit of improving our industry throughout.

We have seen the adoption of FWI in the processing of seismic data over the last five years or so. How is it benefitting the processing of seismic data?

I have no experience with FWI to draw upon to answer this question. Regardless, I know that for any emerging technology to be successful, the basis on which it relies must be robust. I expect that as FWI becomes a more common pursuit in geophysics, the accuracy of the geometry and the preprocessing applied to the data will prove to be critical. The benefit for seismic processing is continued refinement of our techniques through a new lens. Notably, this includes revisiting established algorithms and methodologies to improve upon any compromises that were necessary at the time of their developments.

We are witnessing more and more application of machine learning techniques for mundane tasks in seismic data processing such as editing of traces, velocity analysis, etc. Can you elaborate on how it is proving to be useful?

One usage of machine learning in processing is for data driven analysis, which is very appealing when dealing with large data volumes. A typical processing flow may include many of these tasks, such as first break picking, velocity analysis, multiple identification, etc. Once reliable training data is in place, these tasks can be handled consistently and relatively quickly with machine learning compared to doing them manually. This is advantageous for both client and service company.

Anecdotally, my first assignment as a front-end processor was manually adjusting first break picks on a large grid of 2D lines. The area was covered with small frozen ponds that caused abrupt shifts in the first breaks that traditional auto-pickers weren’t able to accommodate. The work was carried out by a team of front-end processors and took an entire month to complete. This is a good example of a large commitment of people resources that machine learning can be used to alleviate. Although there was value in being exposed to this problematic data, the knowledge gained was covered in the first few days and the rest of the month was pure repetition. With machine learning, this type of work would be completed quicker and with more consistency than having many people doing the work manually.

Seismic data processing comes with many challenges by way of data calibration, regularization, noise, and multiple attenuation, as well as signal enhancement. Added to these is the challenge of handling large data volumes being generated these days. How can these challenges be addressed?

These challenges can be overcome by leveraging the diverse knowledge and experiences of colleagues and industry professionals. By incorporating multiple points of view from multiple disciplines in brainstorming and problem solving, there is a higher likelihood that a solution will be found quickly and will be more universal in its applicability. I believe this is a fundamental reason that cross-discipline collaboration is more frequent as our industry takes on greater challenges.

Handling the challenge of large datasets is another opportunity to involve disciplines outside of geophysics. I am fortunate to work alongside computer scientists and network specialists that regularly share new strategies from their disciplines. Solutions to data inflation include denser storage media, faster interfaces, increased network bandwidth and modern algorithms from machine learning. Additionally, highly customizable cloud compute services exist that are suitable for scenarios in which onsite compute capacity expansion is not realistic.

What is the most important thing that you learnt in your educational years which influenced you later in your professional life?

Throughout all my years studying, I always took every opportunity to tutor or to be a teaching assistant. These experiences have underlined the fact that “You only understand a concept as well as you can teach it to others.” Keeping this in mind, I try to actively engage my peers in technical discussions whenever I undertake learning anything new. This approach creates a positive learning feedback loop that can persist beyond the learning stages.

TopDown: Warm summer evening cruising in our 1974 MGB.

What do you attribute the biggest successes in your life to? How about your largest failure, if any?

The optimism and determination that were instilled in me growing up have served very well in my adult life. I have consistently met life challenges openly and with a positive mindset.

I would never claim to have escaped failure. My biggest failures – optimistically referred to as “unrealized opportunities” – have been the result of breakdowns in communication and errors in judgement. I believe it is important to scrutinize failures and to be open to constructive feedback in order to learn the most from these events.

Let me ask you a philosophical question: it is usually said that the ‘result achieved is usually in direct proportion to the effort applied’; do you think this is true? If not, how should it read?

I consider this statement to be correct though it can be misleading. With my mathematician’s hat on, this statement can be viewed through the law of averages: To meet or exceed expectations requires an investment of time and energy.

I prefer Pareto’s principle: “roughly 80% of consequences come from 20% of causes”. I like this principle because it more accurately describes the humanistic aspect of typical effort given. Many daily tasks come as second nature to our peers. It is only upon being challenged that people’s strengths are put on display. I think that overall results are most often a reflection of how difficulties are handled along the path to obtaining the results. From a managerial point of view, looking at results through this metric also makes it clear to whom praise is owed and when it is deserved.

What are some of your other interests besides the work you do for a living?

I grew up in a very small town with virtually no light pollution and have always been an avid sky watcher. I like photography too and so I have naturally tried my hand at astrophotography, but I still have a lot of learning and experimenting to do before I can claim that I am any good at it. I am also an avid tinkerer whether it is mountain bike maintenance, computer upgrades or woodworking; I always have a project on the go. My most recent completed project is a home-made dog sled for our Bernese Mountain dog.

Above all else though, my favourite activity is spending time with family and friends. At home, we play a lot of social games, but we also try to get outside often for walks, bike rides, swimming, days at the ski hill and exploring local attractions.

What would be your message for young geoscientists joining our industry?

My best advice is not industry specific: “Learn to identify and accept change early”. It is impossible to predict the future but being able to take the measure of a situation and get to action quickly when the unexpected happens is indispensable.

We thank you for sharing your experience and sparing time for this interview. We wish you good luck with your work.

I enjoyed it. It is an honour to be interviewed for the CSEG Recorder and I would like to thank the CSEG for all the resources it provides to the geophysical community.