“I find geophysics absolutely fascinating...”

An interview with John Bancroft

Coordinated by: Satinder Chopra | Photos courtesy: Penny Colton
John Bancroft

John Bancroft is a senior research geoscientist with the University of Calgary with CREWES and a member of the adjunct faculty. Recipient of the 2005 CSEG Honourary Membership Award, John has written and spoken extensively on various topics in seismic data processing and analysis. He is better known for his extensive work on seismic migration, which he has compiled in the form of two volumes of course notes on the subject. John conducts these courses on the SEG and CSEG platforms as well as other inhouse courses. He has won several best paper and honourable-mention awards for his presentations at the CSEG, SEG and Australian SEG Conventions.

It was a pleasure getting to interview John and he was quite forthcoming with his answers, conveying his points of view very frankly for all the questions asked. Following are excerpts from the interview.

John, tell us about your educational background and your work experience.

My academic studies began at the University of Western Australia, but in the middle of a semester, my family decided to move to Canada. I then completed my B.Sc. and M.Sc. in Calgary and my PhD at BYU in Provo Utah with all degrees in Electrical Engineering.

My Master’s research was high frequency time domain reflectometry that was used to identify system parameters by analyzing input, reflected, and transmitted impulse signals. My PhD research was on electronic speech recognition, specifically being able to identify the phonemes of both male and female speech.

What kind of work experience did you gain?

My first job was in Rochester New York, at the National Technical Institute for the Deaf/Rochester Institute of Technology. I taught classes, but my main work was research in identifying and measuring the speech components of deaf students. The main idea was to objectively measure and evaluate the improvement of “deaf speech” before and after therapy. I also worked on a team that was developing a cochlear implant, or bionic ear, my responsibilities being estimating the electrical field strength from the electrodes to the nerve cells.

We loved the Rochester area of New York, but after five years decided to return to Calgary. Job-hunting was tough because there was no interest in a Speech Scientist. Fortunately, a computer salesman who worked with me in Rochester had moved to Calgary, and was able to make the connection with my headhunter that speech signals were very similar to seismic signals. When we left to come to Calgary for a vacation, I had one possible interview, but when we arrived, my whole vacation time was spent interviewing.

In the oil industry where did you land your first job?

Don Chamberlain convinced me to join Geo-X and my mentor was John Hodgkinson. I really enjoyed my time and work at Geo-X, and still appreciate all the people I worked with there. After eight years, I went to work with Veritas and was there a few years meeting more fine people. We parted company after about 2 years but they hired me back as a consultant for a considerable time, and that got me into the consulting business. After four years of consulting and teaching industry courses, I was invited to join CREWES and have been there since.

How come this switch from electrical engineering to geophysics, was it partly the job situation that made you switch?

The processing of seismic data is signal processing, i.e. transforms, filtering, deconvolution, etc., all that I used in speech analysis. The only thing I really had to learn was seismic migration, and that was quite different. Migration was a new “toy” in the industry at that time so I developed a set of course notes to help clients.

John C. Bancroft

Migration is one of the subjects that is considered very difficult. I have attended your course and I know first hand that you try and put it in a very simple way. Why is that phobia around migration? Why is it not usually explained in a simple way so that people don’t run away from it, but actually try and learn about it?

When I was a graduate student I had a Corvair and the automatic transmission blew out. I took it to an automatic transmission repair shop, they repaired it, but I declined the warranty. Two months later it quit again so I decided to try and fix it myself. I figured it would take a week, so I cleaned out Dad’s garage, and took it apart, piece-by-piece and spread it around the garage. I found parts that were worn and not replaced by the transmission dealer and to my surprise it was quite simple to put it back together using the reverse order. The whole repair only took one day and I was pleasantly surprised when it worked. This experience showed me that with a little bit of mechanical ability and guts, I could tackle what would appear to be very difficult tasks. I make the analogy that the mystique about migration is similar to the mystique about an automatic transmission. The basic concepts are very simple. I also like to understand why things work and get a physical intuition for it. That is when I know I can write a program to accomplish the task. I don’t like to transfer equations from a paper without understand the physical principles. Understanding the physical principles then leads into visualization of the problem. I enjoy creating figures or images for my own understanding, that way I feel I can write programs more efficiently.

I had a mathematical geophysicist tell me that he refused to use figures, as they will always put a limitation on the application of an equation. I believe the opposite is true, in that figures help describe what the equation is telling us. Attaching a physical interpretation also helps me remember the equation and really helps when trying to debug software. The math of many processes can be very intimidating and often hide or obscure the simple processes.

Apart from seismic migration, what other areas of geophysics fascinate you in particular and why?

I have enjoyed most aspects of seismic processing. I have really enjoyed velocity analysis, and my work in this area was beneficial when I recently helped a graduate student who was using a high resolution Radon transform to remove multiples. I have also done a lot of work with surface consistent statics and have enjoyed the fun and problems you can run into when you don’t pay particular attention to the separation of source receiver components.

In one paper I did with Chuck Ursenbach, we estimated trim statics for high fold Saudi data, but we used Alberta data for the correlation model. After stacking, we ended up with good-looking Alberta data. So there are all sorts of weird things that can happen with statics, so you have to be very careful. I find that fascinating.

Deconvolution, what an interesting topic. We try to get information in a linear sense, trying to maximize the signal to noise ratio in the frequency domain. I also like the concepts of sparse-spike where you can use non-linear methods to extend the bandwidth beyond what is given to you. I find that very interesting and probably a fruitful area in the future.

If the input data doesn’t have that bandwidth, where does the excess bandwidth come from?

That’s a very similar question to interpolation; given some data with a known sampling rate, you fill in between those points assuming certain criteria. And yes you can fill in an infinite number of points between each pair of samples. In a true sparse-spike situation, where you know the wavelet, you can estimate the position and amplitude of one wavelet by cross correlation. You then subtract it out and place a spike on the reflectivity “trace” with that amplitude and time. That reflectivity spike now has “infinite” bandwidth. Note that the cross-correlation will place the spike at a time resolution much greater than the sampling of the input trace. We then repeat the process for all other wavelets. The result is a high-resolution reflectivity spike sequence with “infinite” bandwidth. This is a non-linear process that allows us to recover the amplitudes that are not present in the original trace. However, in our real data the wavelets overlap and we might not know the wavelet, and this forms the starting point of a number of advanced deconvolution methods.

I get that question asked a lot.

Certainly, and I think you know more than me on this topic. There are other possibilities too. I will use an example from my speech processing days. On the telephone, a voice might be low cut filtered so that the lowest frequency is 300 Hz. How could we possibly get pitch information of a male voice that is below 80 Hz? Well, you can square the waveform, (a nonlinear technique,) and get harmonics that extend beyond the original bandwidth to get the pitch.

Many of the inversion techniques also use these concepts with well-log information that uses a synthetic trace from a sonic log to guide the stretching of the high-resolution log relative to the seismic traces. Boy, I need that pad of paper, or better yet talk to Brian Russell.

I also believe that the phase of a trace contains much more information than we currently use and that there is a lot more to learn there. We are only limited to our current bandwidth by our linear thinking.

Satinder, John, Yong, Chuck
Satinder, John, Yong, Chuck

You have done some good work on equivalent offset migration and you also received the CSEG Best Paper Award in 1995 for this work. Tell us something about it, briefly, and also why it has not taken off just like some of the common techniques, Kirchhoff, or finite-difference or F-K or something like that?

Wow, I need a pad to do some sketching, but I’ll give it a shot. I had spent a lot of time trying to make pre-stack data appear to be simpler than having a CMP location and source receiver offset. Then one day the light bulb flashed in my mind to take the source and receiver and co-locate them, forcing the doublesquare-root equation into a hyperbolic form. This trace, along will all input traces, can be summed into a prestack migration gather with no moveout correction. Data that is naturally hyperbolic with the RMS velocity assumption will now lie on a hyperbolic path in the gather. The extreme fold of this gather is well suited for high-resolution velocity analysis. With these high- resolution velocities, normal moveout and stacking completes the prestack migration for that trace. Even when the input data is non-hyperbolic, we can still force the high order DSR equation into a hyperbolic form. It also works very well with converted waves and data with a rugged topography.

Now the second part of the question?

Companies are using it for conventional and converted wave data. One company uses it for onboard processing, and I hear it is used a lot in Houston. Companies in town use it for velocity analysis then use their own algorithms for the final prestack migration. Rob Vestrum uses it with his prestack depth migrations and credits it with significantly improving that process.

I think the biggest reason we don’t hear much about it in Calgary is that the original paper in Geophysics contained a citation at the end indicating the process was patented and that may have scared off a lot of users. It has also come to my attention that this process, or one similar to it, was used in the internal processing of a major oil company, and was developed well before my initial discoveries. People that do use it love it.

To use it, companies have to become members of CREWES to get the software, or because it is patented?

This is a difficult thing to answer. One of the original purposes of the patent was to provide a free licence to those companies who belonged to the CREWES consortium. It would also have been nice to receive some form of a royalty, however we have never received a cent, and the University of Calgary now controls the patent. I wish there was no patent and people would use it openly. Again, one of the original ideas for the patenting was that companies could use it royalty free if they were associated with CREWES.

Do the universities normally patent their techniques?

Yes, the university community does encourage the commercialization of products. There are departments in the university that do a tremendous amount of patenting, like geomatics for GPS work or the pharmaceutical groups.

However, the geophysics community seems to avoid royalties for products that are patented. I am only aware of the Coherency Cube that was patented and developed into a commercial product with tight licensing.

Well, that came from an oil company – Amoco had the original patent.

I think a lot of the patenting that’s done in the oil patch companies today is just for protection that no one can claim that they invented it before them. Not that they want to prevent other people from using it, but they do want to protect themselves to be able to use it.

In your list of publications, I see there are more non-refereed publications than refereed publications. Is there any reason for this? You don’t get the time follow up?

That relates to my job at the University. I do not hold a full faculty position as I am employed by CREWES as a researcher. I do hold adjunct faculty status and funding from CREWES allows me to supervise graduate students. Consequently I have a very large number of students, right now I think I only have eight, but I have had been up to fifteen at one time. A typical faculty member would only be able to support two or four graduate students. The graduate students allow me to leverage the amount of research that I could not possibly do by myself as they do a tremendous amount of work. So, I haven’t had the necessity of “publish or perish” to produce peer review papers as much as producing papers for CREWES meetings or industry sponsored conventions.

How effectively are we able to account for anisotropy in processing our seismic data from a practical standpoint? What is the status?

Well, we have been able to handle anisotropy by totally ignoring it for many, many years. The effects of anisotropy have been included in our stacking velocities, and that is one reason stacking velocities do not tie with velocities derived from well logs. Consequently it has not been that important to us. Also, if the anisotropy is elliptical, then the moveout is still hyperbolic and there is negligible evidence of its presence apart from the well logs. Now as we get into more complex media and as we are getting into more accurate imaging, we do need to take anisotropy into account. We can and should use anisotropy in our depth migrations, however, our problems are the same as in all imaging problems; what are the values of the input parameters or models that we should be using. Rather than have one structured velocity model we now need three additional ones for the anisotropy parameters of epsilon, delta, and the dip. We can get the dip information from iterative processing and then use field measurements of epsilon and delta to build their models. This requires interpretation of the seismic data to define the formations, and then fit the field derived anisotropy parameters into the model.

I have one student who is trying to measure anisotropy parameters from seismic data. Results are reasonable when we have well control, but when there is no well control we use the non-hyperbolic nature of moveout. In this case, the anisotropy can not be elliptical and we may be including the non-hyperbolic effects of vertically increasing velocities. So yes, we can estimate the parameters and with well logs do a lot better job.

That’s the bottleneck.

Yes, but I think when we become familiar with an area and we know the formation parameters then we can do a reasonable job.

John C. Bancroft

Then you start using those parameters for your methods.

Yes, they become part of the model.

That’s fair enough.

I’ll add another thing: isotropic media with velocities that vary with depth are non-hyperbolic and appear to be anisotropic. So, by having high order move-out equations we can still compensate for these effects with our imaging and focusing. However, in these cases we don’t get the information that may help resolve rock properties. I will also caution that using large offset data may be detrimental to stacking as the NMO stretch becomes quite large and could harm the resolution of the data.

I talked about migration being difficult and how you make it simple and all that; tell us something about how you got involved with migration and how you decided to start offering a course on migration to the industry?

Well, in the early 80s, migration was becoming economically viable to process data and we were selling the process and charging more for it beyond the normal processing. We needed to explain to the clients why and what we were doing and the benefits or improvements that would be achieved. That was the beginning for me of finding ways to explain why we should be using it and what is wrong with not using it. I just collected materials, collected ways of explaining it, and it evolved into course notes and now there are over 950 pages.

Interesting. You are a senior research geophysicist with U of C with CREWES as you mentioned, and a member of the adjunct faculty there. What responsibilities does your present job entail apart from supervising that you have mentioned? Do you also have to teach and get funding in for projects, etc. etc.?

My duties at CREWES are specifically research. I can do more research by supervising graduate student, by providing little guidance and let them do the work. That’s where I am able to accomplish a lot of the work that appears on that publication list.

I do teach undergraduate and graduate courses, with my migration courses as a graduate course. It has labs associated with it that are not in the industry presentations. Right now I am co-teaching a Natural Disasters Course with Rob Stewart and I really do enjoy that.

Yes, so do you also need to apply for funding for different projects that you have in mind?

I am fully supported through CREWES and I was able to get one NSERC grant for myself many years ago, but when I applied again they told me they would not fund me because of my adjunct status, especially when I am supported by CREWES.

Okay, what are your aspirations for the future? Do you aspire to be a member of the regular Faculty?

Yes, I would like that to happen some day. The department is expanding and it may happen.

John, how about your own ambition?

I really enjoy what I am doing now. I love the research and teaching. I can’t wait to get to work. I am really fortunate to have a job that I can enjoy so much.

That more than compensates for what you are looking for in your job?

Oh, I am approached or offered jobs for more money, but the work at the University is so rewarding, I don’t think I could find a better job because of that.

Okay, you have an impressive list of publications that I mentioned before also. Your research weighs heavily towards migration and anisotropy and there are a few entries on research into techniques that characterize reservoirs, like attributes or AVO or rock physics or something like that which directly have an impact on the reservoir properties. Would you like to comment on that?

The objective of seismic and the other geophysical tools such as VSPs are to build depth models of rock and fluid properties. We have come a remarkable way to achieving those goals, especially in areas where we can record quality seismic data. However, in areas with rugged topography, or high velocities at the surface, we do not do as well. For example in Class 2 AVO, where there is a polarity change in reflection amplitude with offset, the stacked amplitude tends to cancel, making velocity analysis and imaging difficult. By using the matched filter concept, we can search for and identify them quite readily. This sounds great, and it is, but we still have to contend with artifacts that think they are Class 2. There is still so much to do in just this area.

For example, the hyperbolic moveout with constant amplitude will focus to a small area on a high-resolution semblance plot or a Radon transform. However, a Class 2 event may appear with only a small amplitude that is smeared. Well that smear has very useful properties and how you handle it is very significant. Using a matched filter would focus the Class 2 energy, and let the “normal” data smear.

What is your impression about the important developments that people could expect in geophysics in the near future, anything path breaking or revolutionary that we can expect in the near future? Just like you know we had the 2D method in the late 70s and then came the 3D revolution. Do you see anything like that in the near future? Some people I interviewed earlier had mentioned 4D seismic; apart from that is there anything else that you think might want to mention?

Well, the recording and processing of converted waves is having a huge impact on the industry, especially when we talk about rock properties. But aside from that aspect, my interest is the inversion aspect. Modeling seismic data from rock properties is the forward process and relatively easy to do. Estimating the rock properties from seismic data is the inverse process and is extremely difficult. For example, seismic migration is a very simple form of the inversion process. When relating inversion with linear algebra it becomes a simple transpose process that ignores the covariance matrix. We are seeing imaging techniques that are trying to alter that concept and get closer to estimating the rock properties.

Today we are acquiring and processing 3D ground penetrating radar, just like we acquired seismic data in the 1980’s, with large amounts of missing data. With the 3D least-squares approach (closer to true inversion) we have the possibility of doing a much better job. These data sets are much smaller and can afford expensive algorithms. However, success in that area will eventually benefit the seismic industry, especially when computing speed continues to increase.

Being on the Publications Committee of the SEG, I notice that your book on migration is essentially course notes that have been published; you’ve got a lot of examples, you have lots of literature, why didn’t you ever think of putting that in an actual book, that could benefit the geophysical community?

I tend to think of my course notes as a book. When I have specific topics to teach, say the Fourier transform, or decon, etc, I write my own set of course notes. I probably have a hundred topics of course notes that I can pull out at any time and teach. Putting them all together would be thousands of pages.

John, the reason I ask the question is when you have piles of literature in the course notes, you have files or folders. A lot of the teaching that you do you is actually doing verbal communication and then for the examples you probably use projections or whatever. Now when you sit down and write all that for people, of course it is a lot of hard work, but then more people can look through it, not only the students. That is the reason I ask this question.

Yes, I do have many notes on specific topics and I call them monograms. Just basically one topic and written like a chapter. I would like to pull all these things together in a uniform manner and make them available, first to the CEWES sponsors, and then to the geophysical community. My trouble is making them uniform; maybe this is enough of a kick in the butt to do something with them.

John C. Bancroft

The idea is multiple people will benefit from the work that you are doing there.

What is your impression about the current state of the Canadian Universities, in general and then with respect to Geophysics in particular? How do they compare with other North American and European Universities? You may also want to talk about problem-oriented research, the funding, dearth or abundance of students or whatever?

I would like to think that we have one of the best departments in the world. We have good funding, especially with the consortia. We also have incredible expertise in the department, and great support from the industry. We get many applications from people all over the world wanting to come as graduate students, and that should be a good indicator. Many of us in the department are carrying a very large number of graduate students. This is only possible by the funding we get through our consortia.

The next one may be along similar lines – How about the University of Calgary, what sets it apart from other Universities, what is it that would attract students to come and work here, would you like to comment on that?

We have many, many fine Departments in the university. On general terms, I do know that we have one of the top Geomatics departments in the world, and there are other top departments that are in space research, reservoir and chemical engineering etc. All of this is aided by our proximity to the oil patch.

Maybe the yardstick that MacLean Magazine adopts in evaluating different Universities needs to have another look at?

Oh yes, and they know that and they themselves admit the bias. Oops, we had better not get too political here.

Would you like to comment on the quantity and quality of students coming to your Department for research?

We have a number of very good Canadian students who apply but nowhere near the number of applicants we get from the foreign countries. One of my foreign students, Xiang Du, who has just completed his PhD, was simply brilliant and probably accomplished three times the amount of work required for the degree. Another example is John Millar who came to me with an idea from another discipline, and has done some incredible work applying the multigrid technique to geophysical applications. I have many other students who are also or have done similar high quality research.

One difficulty we do have with our graduate students is keeping them here at the university to complete their degrees. As soon as they get close to graduation, they have jobs offers that pull them into the industry, making it difficult to complete their degrees. We have a similar problem attracting our own undergraduate students to continue on to get graduate degrees. This is great from the student’s perspective, but tough from the faculty’s point of view.

John, what are your other interests apart from Geophysics?

Well, as humans we live in a 2D world on the surface of the earth, but I really enjoy escaping into a 3D world with three degrees of freedom. Consequently I love scuba diving and flying. I have a private pilot’s licence and like to get in the air as often as possible.

So do you go out occasionally?

I go out as often as I can.

And that would be how many times?

Oh, normally I go practise flying between once a week to once a month, but for a while my medical was cancelled because of my pacemaker. That is all corrected now and I am happy to be back “up” again.

My family, of course is also centered in my life and I have five sons, one daughter and along with their spouses have seven grand kids. I naturally became heavily involved with Scouts Canada and really enjoyed the activities with my boys that I may not have accomplished without that program. I was a scout leader for, I believe, sixteen years. To keep things somewhat balanced, my wife was involved with Girl Guides and she also enjoyed her time with our daughter. Central to our family development was our activity in church programs.

You mentioned about your pacemaker. I understand you had quite a year with heath problems. How did you deal with that?

Hmm, let me back up for a minute. I was fortunate to have a great lunch group while I work at Veritas. However, we all seemed to have very negative attitudes, and as a group we decided to be more positive, that is to try and find the good in a “negative” situation. For example, I was buying a computer and went with a computer salesman to visit an on site computer. When we got there, the machine was apart being repaired, (they were big in those days), and the salesman quickly exclaimed, “look at our expertise and how quickly we respond to problems”. He knew how to make lemonade from lemons. Our lunch group tried to be more positive, and I remember seeing a STARS air ambulance fly overhead. Our initial reaction was “ wow, isn’t that sad, someone is in serious trouble”, but with a little effort we all agreed that “wow, isn’t it great that we have this service to help someone in need”. Better feelings with the more positive attitude. Ever since then I have tried to follow that philosophy, and that was a great help in my “medical year”.

What was that?

Over a period that was about a year in duration, I was diagnosed with a tumor behind my left eye (somehow it went away by itself), received my pacemaker, the “growths” on my lungs turned out to be scaring, (but it was a long wait to find out), some problem tumors, one as big as the palm of my hand, were removed that turned out to be benign, and then the passing of a kidney stone. Throughout this period I would say to myself, “boy, I am glad I went flying when I did”, “I am glad we know what these problem are”, or “I am glad we found this out when we can do something about it”. I know I was fortunate to survive that year and think the positive attitude almost made it a great time.

I have one last question. What would be your message for young geophysicists entering our profession?

I find geophysics absolutely fascinating. There are so many interesting aspects to it that they have made a great decision and should look forward to a very interesting career. Whatever they do, they should do it with the idea of having fun, and enjoying the thing you are doing. For undergrads, concentrate more on finding interesting things with the course material that will keep you motivated. It takes work, but is worth it. Graduate student have an advantage in taking courses of their choosing, but when things do get tough, remember to search for the positive.

I remember taking courses when the professor would say there was no application, and you must learn the basics before you could apply them to something useful. That is rubbish, laziness, and incompetence. I believe that an instructor is responsible to help motivate student to maximize information transfer. Oops, I am getting preachy.

We learned it the same way. We learned all Math the same way, integration, differentiation, the differential equations, the integral equations, and all that without application. At that time they looked so boring. Except that you learn how to do it, or how a particular type has to be solved, you do not appreciate it. But if all that is supplemented with real applications, it will make it so much more interesting.

And that is where experience is so valuable. Hopefully, the 14 years I worked in the oil patch helps me to include examples in my teaching.

For example, I have three boys in University of Calgary right now, and one is taking Linear Algebra. I was able to say, “Oh, this is so cool” then showed him some examples of solutions to surface consistent statics. That really changed his attitude towards the course. I remember the same course, many years ago, and how awful it was. Now I find that one of the most interesting subjects to study.

Another interesting memory is when another one of my sons came to my office and asked “Dad, do you ever use calculus or differentiation”...

On the other hand one son does not like to ask me questions because he “doesn’t have enough time”. I do tend to lecture.

Any last words?

In the mid 80s some geophysicists thought we had invented everything we needed, up to and including pre-stack depth migration. There was no need for people to learn exploration geophysics any more, so why go to University and study geophysics. Well, it is 20 years later and we are nowhere near the tip of the iceberg. There is so much to learn and so much to be done. Our ride continues to be challenging and exciting and I am sure it will always continue to be so.

I would like to thank you for giving us the time to sit and chat with you. Thank you.


Share This Interview