May 27, 2022
When people in California talk about "The Big One," referring to the next big earthquake, Dave Jackson wants to reframe the article in the sentence and the underlying assumption. In his five decades of work exploring ideas of cycles and the ability to predict earthquakes, his overall conclusion is that, to the best we can tell, earthquakes cannot be predicted with any degree of useful accuracy, and they do not occur according to a cycle that would tell us that so many years have passed that we are due for "The Big One." Instead, the science tells us that "a big one" will inevitably arrive that may, or may not, be of greater magnitude than the last large earthquake to shake California. The best mitigation strategies, therefore, are focused on earthquake early warning—alerts in those precious few moments before the shaking begins—and advances in earthquake engineering that make buildings more robust and minimize loss of life and property damage.
Jackson enjoyed a unique relationship with Frank Press, one of the giants in the field, who would go on to be science adviser to Jimmy Carter and to lead the National Academy of Sciences. As an undergraduate at Caltech, Jackson conducted original research with Press's encouragement, and when it was time for graduate school, Jackson followed Press to MIT. Jackson subsequently joined the faculty at UCLA, and in the past fifty years, he has been centrally involved in advances in earthquake science in California and worldwide. As one of the founding members of SCEC, the Southern California Earthquake Center, Jackson has made key contributions so that the region is best prepared for a big one, if not the big one.
DAVID ZIERLER: This is David Zierler, Director of the Caltech Heritage Project. It's Friday, May 27th, 2022. I'm delighted to be here with Professor David Jackson. Dave, it's great to be with you. Thank you for joining me today.
DAVID JACKSON: So good to be here and join this discussion.
ZIERLER: Dave, to start, would you tell me please your title and institutional affiliation?
JACKSON: Right now, I am a retired, let's call it Emeritus, professor of Geophysics at UCLA. I retired about ten years ago. So, I've got a long history at UCLA and, of course, at Caltech and MIT, as well.
ZIERLER: Dave, are you enjoying a full retirement? Are you active at all? Are you staying on top of the literature?
JACKSON: You know what retirement means.
JACKSON: It means you change the direction, the flow of money from a salary to some other kind of support. [laugh] No, I'm actively involved in research, as well, and paying attention to students who are at UCLA trying to understand what they're doing and interact with them, as well.
ZIERLER: Given that you're still active what are you currently working on?
JACKSON: I'm currently working on understanding some geological data, that have been compiled by other people, related to understanding the future of earthquakes that may happen . And I'm doing statistical analysis on those data. I'm also working on a global model of earthquake occurrence of magnitude 6 and larger earthquakes and so that's also more of a statistical thing than straight seismological.
ZIERLER: Dave, thinking about future earthquakes, does that mean that this work might get us somewhat closer to earthquake prediction?
JACKSON: That's what I think a lot of people, almost all seismologists, hope for. I'm someone who doubts that predicting the times of earthquakes is going to be very successful. I think the places of earthquakes and the kind of earthquakes that may happen and what's going to happen in the event of an earthquake already occurring, those things I think we can deal with. But trying to predict the times is, I think, a long way in the future and maybe never going to happen.
ZIERLER: Now, if it never happens, to what extent is that a statement about our limitations with sensing and technology and to what extent is it simply a fact that perhaps the earth itself doesn't know when an earthquake is going to happen?
JACKSON: A lot of both but the fact that earthquakes occur below the surface of the earth maybe five miles or deeper, and in many global earthquakes a lot deeper, those we just can't observe directly. The other thing is that you need a long time-record. We didn't observe many earthquakes before, say, 1900 in any kind of detail. We can't turn back the clock and understand earthquakes that have happened a long time before. [laugh] That's one of the reasons I'm pretty pessimistic about this. I might mention that at the Seismo Lab there have been lots of discussions about the predictability of earthquakes, so there are some optimists and some pessimists. I shouldn't call myself a pessimist but let's just say I'm much more inclined to think the problem may be unfathomably complicated.
ZIERLER: Dave, over the course of your career have you been more on the theoretical side or more on the observational experimentation side?
JACKSON: A mix of both. More on the observational and more trying to understand the implications of earthquakes that have happened without having to do a lot of mathematical things. But my statistical research is definitely on the mathematical side.
ZIERLER: How big a role has modeling played in your career; specifically once computers started to get really powerful?
JACKSON: It's, again, kind of an intermediate story. There are people who do much more detailed modeling than what I do. But I do modeling, and I use existing models putting in data from recent earthquakes or in some case past earthquakes and trying to understand how those data fit into the models. But I'm not creating anything that's a radically new model.
The Origins of SCEC
ZIERLER: Dave, a few administrative questions. Have you been part of the Southern California Earthquake Center or SCEC from its inception?
JACKSON: Yes. I was one of the people who helped to write the original proposal that went to the National Science Foundation for SCEC.
ZIERLER: Oh, wow. Who were the others?
JACKSON: Prof Keiti Aki at USC first proposed the idea. He then formed a team with Leon Knopoff at UCLA, Lynn Sykes at the Lamont-Doherty Lab at Columbia University, and Robert Wesson at USGS. Tom Henyey at USC, and I at UCLA, partnered in writing much of the original proposal to the National Science Foundation
ZIERLER: What was the need for it? What did you express in the proposal?
JACKSON: The need for it was a way of integrating what we knew about earthquakes with what could be done to deal with the risks that are posed by earthquakes. There were a lot of approaches to that, a lot of measurements of earthquake data, seismological data, and geological data, and a lot of modeling going on with that. But before SCEC it was mostly done by individuals all competing with one another. That is a very healthy process. But Aki realized that if people could work together some of the observational data could be collected in a more organized way. Then people could get together regularly and discuss their results, still competing but also sharing. It was, I think, a very good and important perception.
And by bringing together scientists, we started with a small group of universities and now it's grown to 50-some different universities. Aki retired from SCEC, I think in about 1997. Now, sadly, he's deceased.
ZIERLER: What were the initial institutions involved in the creation of SCEC? Was it UCLA and USC initially?
JACKSON: It was UCLA, USC, Caltech, Columbia University, U.C. Berkeley, Scripps Institution at U.C. San Diego, and a few others. The United States Geological Survey was also closely involved.
ZIERLER: But Caltech and Seismo Lab were not?
JACKSON: They were involved, but a little reluctant to get heavily involved. [laugh] And I think that's an interesting story about seismological history in California. The US Geological Survey focused mostly on central and northern California, and U.C. Berkeley maintained a seismic network in northern California. Caltech was the place in southern California. There was some competition going on. Part of Aki's idea was to sort of bridge those gaps, the divides that occurred, but Caltech was pretty happy in its own territory, and they had a lot to do already. So they were a little bit reluctant at first, but now they're very strongly participating and there's no question that it's pretty much a unified group of researchers.
ZIERLER: Dave, Caltech's reluctance to join SCEC initially, does that play at all into the story about Caltech being slow to the plate tectonic revolution also? Are those interconnected stories in your view?
JACKSON: I think not. I think the plate tectonic story was a global story and involved a lot of geologists and seismologists well beyond Caltech. Others included MIT and the Scripps Institution on, Lamont-Doherty at Columbia, Woods Hole Oceanographic Institution, Cambridge University, and universities throughout the world. It took a long time for the plate-tectonics hypothesis to develop and pass some tests. I wouldn't say that Caltech was behind in that. I think we were all behind.
ZIERLER: [laugh] When finally did Caltech and Seismo Lab get involved in SCEC?
JACKSON: It was partially involved at the very beginning, and it grew, but I think it was largely because Caltech had so much going on and they sort of regarded SCEC as being a way for the other universities to catch up. I couldn't put a date on it, but SCEC started in about 1990, a little bit before that in terms of that proposal that was written and so forth. So, there wasn't really a date when Caltech said, oh, we'd better get involved, because there were several people from Caltech who were involved in SCEC from the beginning.
ZIERLER: Dave, have most of your collaborations regionally been through SCEC or have you pursued independent collaborations at other institutions including the USGS?
JACKSON: I have been involved strongly in SCEC, serving as its Science Director for a few years after Aki retired. I've also collaborated with other universities including some global partners. SCEC is of course focused on Southern California, but important earthquakes also occur elsewhere. [laugh]. Many of us were also involved in global studies and my statistical studies aren't necessarily focused on SCEC.
ZIERLER: A few questions administratively about geophysics and seismology at UCLA. At UCLA is there an analog to the Seismo Lab? Is there a program that's sort of separate from the rest of the institution?
JACKSON: There was a strong geophysical group at UCLA before SCEC started and the primary leader of that was Leon Knopoff. He had been at Caltech, had a PhD from Caltech, so he was very strongly connected with Frank Press and with the other leaders of the Seismo Lab. Also, he was very strongly connected with Kei Aki. When SCEC started, Kei Aki invited Leon and some people other people as I've mentioned before. But at UCLA we had nothing comparable to SCEC, especially regarding such an organizational structure.
ZIERLER: In your career how active have you been in the AGU?
JACKSON: I think I got a pin of some kind a few years ago as a 50-year veteran of the AGU. [laugh] I went to my first AGU meeting in about 1966 or 1967 and have been attending AGU meetings pretty much all the time. There were one or two that I missed.
ZIERLER: For geophysics and seismology for you is that the most important scientific society?
JACKSON: Oh, it certainly was. Now the Seismological Society of America is very much a strong participant and there are global organizations like the International Union of Geodesy and Geophysics which also are very strongly involved. Again, I wouldn't say that there's a single most dominant society but certainly the AGU is very strongly in a position of leadership.
ZIERLER: Dave, just a headline news question. Only a few days ago there was an earthquake something like 4.3 in the Los Angeles area. Are you set up where you're ready to look at earthquakes as they happen?
JACKSON: Let's say I'm set up such that I could. I generally focus on earthquakes bigger than magnitude 6 in California, so I don't pay a lot of attention to all magnitude 4 and 5 earthquakes. [laugh] I notice them, but sometimes I find out about them because a reporter calls me up and says, what was that about?
JACKSON: The first thing I do is I say, well, tell me what you know about it. [laugh]
ZIERLER: [laugh] Dave, I've learned that there's a few schools of thought here in California, of course. One is that we're waiting for the big one, however long that might take and however big it might be. The other is there's really nothing to wait for because there's not necessarily a cycle to these things. Where do you fall in on those debates?
JACKSON: No cycle. I resist the temptation to talk about "the" big one because I think the article "The" is wrong. A big earthquake may happen at any time, and we know that several have happened. I wouldn't describe it as being a unique thing to look for, so I don't like talking about "the" big one. Some big ones are going happen. [laugh]
ZIERLER: How big theoretically could an earthquake get in California?
JACKSON: I resist trying to put a number on it. We don't know. The biggest ones that we know about having happened already in California are about magnitude 8. There was the 1906 earthquake, sometimes called the San Francisco Earthquake, and there was an 1857 earthquake in Southern California. We didn't have good seismological measurements at that time, so we don't really know how big those were. A lot of people expect that those are the primary types of earthquakes that may happen, even in the same place. I think we don't know that. We can have big earthquakes in lots of other places. The biggest earthquake recorded by a network in Southern California was in 1952. It was about magnitude 7.5, and it was not on the San Andreas. It has not been followed by any comparable earthquakes. I think surprises are coming; that's my prediction.
Earthquake Preparation in California
ZIERLER: [laugh] Whether it's a big one or the big one, in Southern California how well prepared do you think we are right now between advances in earthquake engineering and also advances in earthquake early warning?
JACKSON: Another difficult question. It sort of requires first for us to talk about acceptable risk. We know some of the risks that happen with earthquakes. We have a pretty good idea of how frequently damaging earthquakes may occur at a certain level of damage and we know the places where earthquakes could be most damaging if an earthquake were to happen. Los Angeles is one of those places, and I include Pasadena in that description, but there are a lot of other problems that people address. Those are also underserved in terms of our facility for dealing with them. For example, COVID is something that—should we be diverting more attention to COVID than to earthquakes? I'd have to say yes. COVID is killing lots and lots of people. The number of people killed by earthquakes in California since 1850 is several hundred.
So, I would say we are appropriately prepared for earthquakes throughout California. We know what's happening. We've done engineering studies that help us understand how best we can make new buildings safer. Dealing with older buildings is a lot more problematical because first of all you can't just make them new and there are other issues including people who already live there, and do you move them out and do some serious retrofitting and it takes years for them to come back? So, I would say, dealing with all the other risks we face, we must stay focused on quakes, but we shouldn't be embarrassed about the level of preparation that we have. Could we think of all kinds of ways to improve our safety? Yes. But we have to balance that with other needs.
ZIERLER: Where do you put the level of sensors that we currently have in Southern California? Are you satisfied that we have enough for all that we need them for?
JACKSON: Again, what's a need? We have very good seismic arrays. Other countries, particularly Japan, have much more coverage with seismic instrumentation and they have led us in California in sort of showing how to use networks and closely spaced stations so that we can compare the results of an earthquake at one place with that of another place nearby. That's important in improving our knowledge and testing whether our understanding of earthquakes is valid.
So, I would say it would be scientifically productive to increase our coverage and put in new kinds of instruments. We're making some progress but there's plenty to do there and I would say that would be very fruitful to increase our expenditures and put in more instrumentation.
ZIERLER: Dave, we find ourselves in the midst of a quantum science revolution and there's a lot of excitement about what quantum sensors can do. Even though earthquakes are a classical system do you see a future role for quantum sensors in earthquake prediction and early warning?
JACKSON: For basic scientific studies, and for earthquake early warning, probably yes. Regarding prediction, I don't believe advanced computing can overcome the deficiencies in observation, I don't fully understand what can be done with quantum science and quantum computing. I've read some descriptions of how grand it will be, but I'm not prepared to say what's the likely use of the quantum computing in seismology. I go back to my story that there will be surprises.
ZIERLER: Let's go all the way back now to your time at Caltech. First, just to set the stage, as a teenager in high school were you interested in geology and geophysics and things like that?
JACKSON: I was a curious kid, and I was definitely interested in those things. I went to a summer program at the Colorado School of Mines after my junior year in high school, in 11th grade. I was very much excited by that. I got started in geophysics from a high school advisor who urged me to apply for every scholarship that did not have my name on it as an exclusion. [laugh] She said if you see an announcement for a scholarship in biology and you would have fun in biology apply for it. I applied for more scholarships—I don't think I set a world record, but I sent a lot. When I got a scholarship in exploration geophysics I had to go to the dictionary and look it up and find out what it was. But I got the scholarship and I loved it. The more I found out about geophysics the more I liked it.
When I got to Caltech, they didn't have exactly the right geophysics program to match my scholarship, which was more about oil exploration. I went to the Seismo Lab and talked with Frank Press who was the leader at that time and told him my situation. He said, "You should go into the physics department at Caltech. This is the path that most seismologists take." And he said, "Your scholarship group, they'll be happy to have you do that." And that was true. I talked to them, and they said, "Oh, we'd be delighted to have you be in the physics department as long as you go talk to Frank Press and other geophysicists occasionally." And that was great advice.
ZIERLER: Dave, was it easy to get an appointment with Frank Press? Was he imposing? Did you feel a little nervous going to talk to him?
JACKSON: He was not imposing. He was very, very friendly. I'm always a little reluctant to knock on somebody's door not knowing what kind of reception I would get, but I was not shy in that way. Frank was really very receptive. He was delighted to have me there. I don't think he had known that I had applied to be there, however, but when I told him my story he said, "Well, definitely we'll find a place for you at the Seismo Lab and get you a summer job," and things like that. So, no reluctance whatsoever on his part or on mine.
ZIERLER: Just so I can visualize the process, so this was happening on campus, and they sent you up to the Seismo Lab to meet Frank Press?
JACKSON: I don't remember whether my first meeting was up there on San Rafael Avenue or whether it was on campus, because he had an office on campus, as well. I don't remember where that meeting was, but in any case, I was at San Rafael within a week and meeting with other people there.
ZIERLER: What was it like at the old Seismo Lab?
JACKSON: Well, it was exciting. People were always ready to go measure something if an earthquake happened, and people were working on building new instruments and finding places to install them. In those days they didn't have the sort of telecommunications that we have now. We had to find places—I say "we" because I became involved in these activities as an undergraduate—find places where we could install seismograph stations on good solid ground, which is better for recording earthquake shaking. We needed sites close enough and accessible enough that we could come collect our data on a regular basis and collect it on an irregular basis if an earthquake happened, so I'd say exciting and dynamic. [laugh]
ZIERLER: Dave, what were the big debates happening at the Seismo Lab in the early and mid-1960s?
JACKSON: There were lots of debates. Maybe I could start with a little bit of—my view of the scientific method and how I've seen it play out at Caltech and MIT and so forth. I'm sure you probably know kind of the mantra about you make observations and you get curious and so you ask some questions and then you develop a theory and then you come back and make some new observations to sort of validate or invalidate your theory.
The first step to make these observations sometimes is made just by guesswork. You notice that some buildings collapse or that they're shaking in the ground and there may be some seismic stations around and you observe those. Then you make some guesses about what caused the shaking or what you observed. Those two steps can be quite different. The observation can be you just walking around and looking at things or looking at some data that somebody had shown you. Then when you try and make the explanation it involves more theory, sometimes some mathematics, sometimes some kind of new observations. Then you get back to something that is a little bit less fixed. You're trying to envision a theory and then you come back, and you test that with some new observations which, again, put you on the ground and you take some seismometers with you and put them in the right place at the right time.
Those four steps—I used to think when I was in high school that one person did all that. You see some geraniums and you plant some seeds, and you water them and see what happens and how much watering you do. But in real life most of the scientific steps take a long time. Just to categorize the observations that get you started in the scientific method involves a lot of work. That's what was happening at the Seismo Lab at San Rafael. A lot of it was just putting out seismographs in places where, if there was shaking, we could measure it and get the data back.
Then there's the theoretical step of kind of coming up with some explanations. When you talk about possible debates, it's not just a matter of debates, it's a matter of different styles of research, when people see what's important. There are people who are terribly important in building instruments and making the seismographs and making sure that when the ground shakes that you get some data that tells you what that shaking really was rather than just some sort of random yes or no or something like that. Then when you get into the third step, the step of making a kind of a prediction of what might happen, that can involve a lot of analysis, but it can also involve a lot of interpretation of what you've already seen.
Then there's the testing step which is how do you make it an experiment that will look at things that haven't happened yet or that you haven't measured yet and compare that with your theories? And a lot of the prizes come at the AGU from that third step where you're taking a hypothesis and saying here's what I think happened and describing it in a way that it could be testable. Then at the very end there's the step that may involve some statistics and that's where I think I fit in mostly to this process.
But what happened at Caltech was a natural evolution of the importance of those steps because there hadn't been a lot of carefully observed earthquakes in California in 1961 when I came into Caltech. What seismologists talked about were the global earthquakes, the big ones, the 1960 and 1964 earthquakes. The 1906 earthquake in San Francisco, people talked a lot about that but there wasn't a lot of data on that, and so we were in the situation of kind of collecting data and trying to make a story out of what we could see.
But in the ‘60s computers became usable. We had more data produced by Gutenberg and Richter and people like that in California and in the Berkeley Lab in northern California. The shift to a more analytical type of operation was taking place. Frank Press coming into the Seismo Lab was very much a part of that. He was an analyst and he had written a book with some colleagues about the theory of earthquake wave propagation. This is something that hadn't been that important in the early days because we didn't have the kind of data that would be testable by these kinds of theories. But when Frank Press came in this was a paradigm shift. Maybe paradigm is not quite the right word because it wasn't a change in the ideas of what may be going on, it's a change in the importance of getting observational data and then putting it into a kind of an analytical theory.
I came in, I didn't know where I would fit in this sort of balance, and it's only after kind of seeing different universities in this kind of change in what's important that I could get a better picture of the scientific method. I thought I was going to come in and make some observations and plant some geraniums on a fault and see what happened, but it didn't work that way. [laugh]
ZIERLER: Dave, a more prosaic question, could you take courses at the Seismo Lab or that was strictly laboratory and observational work?
JACKSON: That's a good question because the answer depends on the difference between undergraduate students, graduate students, postdoctoral people, professors, and all those who were, of course, involved in the research. I took classes full time. I was a full-time Caltech student, and I was taking five classes at one time for one semester—one quarter. We didn't have semesters. For me I had to live on campus, and I had to thumb a ride up to San Rafael to visit people at the Seismo Lab.
ZIERLER: There was no shuttle? You had to get up there on your own?
JACKSON: I did, yeah, and eventually got a car. There were graduate students who maybe were taking one or two advanced classes and they could reside at the Seismo Lab up on San Rafael or at least spend a lot more time there than I did. Then, of course, the postdoctoral people, some of them spent most of their time on campus and some of them were spending their time up at the lab. Most of my time was spent on campus and I would see Frank Press when I could, sometimes on campus. I loved going to the lab so I did that as much as I could.
ZIERLER: Dave, you already suggested as much, but as an undergraduate if you showed the inclination and the initiative could you be involved in research?
JACKSON: Yes. Yes.
ZIERLER: What kinds of things did you do as an undergraduate?
JACKSON: One of the things I did was look at data provided by other institutions, and I actually was a coauthor on a paper with Frank Press in, I think it was 1965 or '66, on the Alaskan earthquake in 1964. Data came in from the US Geological Survey and others and we had an analyst named Bea somebody, I forget her last name, who was also harvesting a lot of data. I was sitting at my desk and using a very elementary computer at the time to do some calculations about the aftershocks of that earthquake.
I was also involved in some field studies. When someone was going to a new place to install a seismometer or something like that I would tag along and occasionally I was asked to go to one of those locations and just look to see if the instrument was in place and if there'd been any new construction or anything like that at the site that would affect the observations. There were a lot of things that were easy for me to just take as a suggestion of something I might do. Then there was this project with the aftershocks of the Alaskan earthquake that I had the data and I needed to sit down and do the computing. I knew what to do but I had to find the time to do it.
ZIERLER: Did you interact at all with graduate students and postdocs on this research or they were more geared toward things that an undergraduate could do?
JACKSON: I would say in my case it was more for undergraduate things just because my time was programmed by the institute itself and courses that I had to take. I did know a number of the grad students and would chat with them, and we would have weekly get-togethers and just talk about what was new, and those were on campus. It was pretty easy to interact with them. But the atmosphere was such that anybody who showed an interest would be invited. The weekly get-togethers were important in that role. Grad students would have their own ideas and so we definitely were brought into some investigations by the grad students as well as by the faculty.
Early Computers at the Seismo Lab
ZIERLER: Dave, at the Seismo Lab were computers or at least the very earliest versions of computers already in use when you were an undergrad?
JACKSON: You bet. [laugh] In fact, the computers were pretty elementary. They were giant. They were refrigerator-sized computers, and they couldn't do what you could do now on your cell phone. One of my jobs was to do some editing of paper tapes. We didn't even have IBM cards in use at that time. We had these paper tapes coded by holes punched in them. Sometimes the tapes would break, or somebody would, in their programming, instruct the computer to punch a hole in the wrong place. You had to take the tape, cut out the offending piece—it was almost like a kind of surgery—and then put in some blank tape that could now be punched and read again. [laugh] And at the end you got this long paper tape with a lot of holes in it. Part of my job was to look at those holes, translate them into the computer instructions, and see if the instructions were right, and then have to correct them, which was most of the time.
ZIERLER: Dave, besides Frank Press, who were some of the other faculty who were influential for you as an undergrad?
JACKSON: Don Anderson was definitely influential, and he was at the Seismo Lab. Again, he had an office at Caltech. I started on a job with David Harkrider. David was getting a project started and Frank just thought, oh, there's an idea. I did work with David Harkrider for a while, but I was soon diverted to punching the paper tapes and going out in the field and trying to examine some of the seismic installations.
ZIERLER: What areas were people like Press and Harkrider and Anderson collaborating on and where were they off pursuing their own independent research?
JACKSON: That I don't have a very good picture of. I knew that on these weekly meetings people would get together and chat and it all seemed to be very much collaborative chat. Don Anderson was more interested in the properties of the materials through which seismic waves passed so he was part of this progression from open observations to testing the theory and building some tests. He was in kind of another wing of that.
Frank Press was definitely in the analytical wing where you got some observations and you're trying to explain them mathematically. But Frank also was very good at instrumental design, so he developed some new seismographs that then became standards throughout the US and Japan and some other places. To my knowledge, for example, Don Anderson didn't do that part of the job. But I don't know, there was a lot of collaboration but definitely in different directions.
ZIERLER: Dave, what about the founding generation of the Seismo Lab? Did you ever get to meet any of those people? Would they come around?
JACKSON: I met Richter. I think I met Benioff early in the day, but I don't remember very much what we discussed. But I do know that there was some resistance when Frank Press was appointed there because of this evolution of the importance from straight observational data to the analytical part. Some people thought maybe that Benioff or Richter should've been appointed because they had already a history at Caltech and Frank Press was coming in from the outside. I'm not sure who made those kinds of decisions but obviously they recognized that having computational capabilities and analytical capabilities that Press had that he would be an addition to the armory at Caltech. I was 18 at the time so I didn't really get to know all these people.
ZIERLER: [laugh] Right.
JACKSON: And remember I was still looking up in the dictionary to find what geophysics was.
ZIERLER: Right. [laugh] Dave, did you spend any time at JPL and more generally were there any institutional collaborations between the Seismo Lab and JPL?
JACKSON: There were. I didn't recognize them as important parts of the development of the Seismo Lab, but I did meet people from JPL, and I lived very near to the Arroyo Seco so I would go hiking at JPL. I know that as time went on JPL at Caltech got much more involved in seismology but that was more after I left Caltech and went to MIT.
ZIERLER: Dave, what about beyond Caltech, do you have a memory of the Seismo Lab being an intellectual magnet, a place where people in the field would come and visit, share papers, learn what was going on, that kind of thing?
JACKSON: Oh, you bet. You bet. When I was at MIT—I followed Frank Press to MIT. He went on to be the chair of the department at MIT. He suggested I should apply, which I would have anyway But he made a very strong case and it sure made it easier for me to get into MIT because Frank had recommended me. But my family home was in Altadena and so I would come to my family home for summertime and come to the Seismo Lab and discuss with people. When I went to the American Geophysical Union meetings, I would meet my buddies from Caltech and go to hear their presentations and read the papers that they were writing. There's no question that the Seismo Lab was a huge strong magnet bringing people in. At MIT many of the people that I met had been at Caltech and had come to MIT as professors. Nafi Toksöz was one of those and Shawn Biehler. I don't know if you've encountered him.
JACKSON: So, there was a very strong connection between MIT and Caltech. Same for Scripps and the Lamont Geological Observatory. We were all keeping track of each other's progress and paying attention.
ZIERLER: Dave, what about the culture of data sharing? In other words, before the internet and networking that allowed data to flow freely from lab to lab from scholar to scholar? Was your sense that the Seismo Lab had data that was in some sense proprietary, that you had to come to the Seismo Lab in order to access it?
JACKSON: Yes, but it wasn't necessarily that people resisted sharing data. There was some resistance, of course, because people had worked hard to put those seismometers in the ground and record the data over a period of years, and they had a plan for testing a hypothesis. That was going to be their mark of success. So, each researcher had a kind of personal trove of information. That's part of what SCEC was all about. SCEC recognized that to make scientific progress data sharing would be important. You still need to restrict privacy and credit the work done by somebody collecting those data. The idea behind SCEC was to share these data and share the credit explicitly.
ZIERLER: Dave, was there a senior thesis or a capstone project for you at the Seismo Lab?
JACKSON: It was really this story about the Alaskan earthquake and my part in this publication but otherwise I did not have what you would call a capstone project.
ZIERLER: Now, as you indicated you went with Frank Press off to MIT. Was the timing perfect? Was he going just at the time when you were graduating?
JACKSON: He really left just before I was graduating and yeah, the timing was perfect. Had I been two years later or two years earlier, might've been a different story.
ZIERLER: Not that he would've confided in a 20-year-old, but do you have any sense of his motivations, what was attractive to him about going to MIT?
JACKSON: I knew that he also felt that the tide was turning toward analytical work. He was very good in mathematics. I think MIT was attracted to him in the same way that Caltech was. MIT was more geological than Caltech had been. The geology was a much more important part of earthquake studies. I think MIT felt they needed him, or they needed somebody like him, and what discussions they had in their boardrooms I don't know. What role money had I don't know either. I don't think that's a very important reason for changing fields for somebody like Frank Press who would be rewarded in any case. He was hugely successful. But I think he saw scientific advantages in using what MIT had available and they definitely thought they had advantages in using his talents.
ZIERLER: What was available at MIT? Where were they ahead of the curve relative to Caltech at that point?
JACKSON: Good question. I think in some regional geology. Caltech geology was really focused on California and at MIT they were more involved in global geology, at least continental geology, and they had oceanography as part of their program. They had the Woods Hole Oceanographic Institute that was pretty closely related to their department. When plate tectonics began to be debated that was a good place to be because oceanographic data were coming in, the kind of shipboard magnetometer data that were a big role in sort of convincing people about plate tectonics. MIT had those kinds of data, and facilities that Caltech didn't have.
ZIERLER: Did you apply elsewhere for graduate school or with Frank Press that was a simple decision just to go there?
JACKSON: I told you about my high school advisor. [laugh] I remembered her advice, so I applied to quite a few places. I think I applied to six or eight places and that included MIT, included Harvard, included Caltech again. I did apply to several different places but I was hoping for MIT and so it was an easy choice.
ZIERLER: Would Press be your thesis advisor at MIT?
JACKSON: He was kind of a co-advisor. He was getting his feet on the ground at MIT and so he was not quite so much involved in advising individual students, but his office was always open. I talked with him maybe not every week but fairly frequently. My advisor was Gene Simmons who was more a geologist and kind of jack of all trades. He was a very good advisor for me for lot of reasons. For one, he told me that I was going to be responsible for my graduate education and my thesis. He wanted me to tell him about it, but he wasn't going to tell me what to do. It was a very active relationship, with much discussion and lots of ideas. That was perfect for me.
ZIERLER: The hands-off approach worked for you?
JACKSON: Oh, yeah. Yeah, it was great. Then, Frank Press was quite like that as well. He would give me great idea. When I met him at 18, I needed some more direct suggestions. But by the time I went to graduate school I had a long list of suggestions that I was already working on.
ZIERLER: Dave, it's impossible to answer this, it's a counterfactual, but in going to MIT whereas you emphasized there was a greater emphasis on the analytical approach, do you think that made your thesis more analytical than it otherwise would've been had you stayed at Caltech?
JACKSON: I don't think so. My thesis was some experimental observations and a lot of just putting together things that I had read and make them into a story. I wasn't doing the kind of analysis that Frank Press was doing at MIT, so I don't think that would've made much difference.
ZIERLER: How much fieldwork did you do for your thesis?
JACKSON: Not a lot. At MIT you don't have the quick trip to the San Andreas fault, or you don't have many neighborhood earthquakes to deal with, so my work was more indoors. But we did have yearly field trips and things like that that were not particularly aimed at my kind of thesis work. In fact. I was doing an experiment on some rocks and ceramics to try to measure what caused energy to be absorbed within the material and taken out of the seismic waves that might go through them. That was quite different from what I had been doing at Caltech, but it followed on one of the major questions that people were discussing at Caltech. The question is, what causes the amplitude attenuation of seismic waves, and why is the shaking is stronger in some places than in others. Why do the waves get weaker as you go further away from an earthquake; how much that depends on the size of the earthquake? I followed those questions into a physical experiment. I went back to Caltech on my summer vacations, and I talked with Don Anderson who was involved in some work that I did. It was definitely a follow-on to my Caltech work but quite different from what I had done on campus.
ZIERLER: Dave, even from earlier in the decade were there computational advances that you were able to take advantage of for your PhD?
JACKSON: Oh, yeah. Yeah. Some of them being books that people had written and some of it being the speed with which you could analyze data. In my classes and in my work analyzing the experiment that I'd been doing I was using the computer quite often. Then when I went to UCLA as a faculty member same story; the computer usage was growing. I think once you recognize the need for analysis, you're gonna be using a computer a lot.
ZIERLER: Dave, what would you say were your principal conclusions or contributions with your thesis research?
JACKSON: I wrote some papers about what had caused the seismic attenuation to be important and the effects of temperature and pressure on those. I was doing laboratory work to generate hypotheses that you use in that third step of the scientific method. Later that was followed by other people also doing more laboratory work. There's a fellow in Australia also named Jackson [laugh] who's doing a similar kind of experiment to what I had done. His followed mine by quite some time, and his was much more elaborate and much more controlled than mine.
It has been a longstanding question about what happens within the rocks when a seismic wave passes and deforms the rocks and then absorbs some energy in that passage. I contributed a pretty detailed hypothesis that could then be tested against laboratory rocks and against the seismic waves themselves and I regard that as my contribution.
ZIERLER: What aspects of the thesis have held up over time and what has changed as a result of new theory, new observations?
JACKSON: I think the general ideas have held up over time. People have come through and looked at a lot of different kinds of rocks and one of the questions that arose from my thesis is what do the little grains in rock do? Because they shift around like that when a seismic wave is coming along. There's an interaction between what's in the volume of the grain and what's at the surface of a grain. One of the things that's held up on my work is that that interaction is very important, that you can't isolate one from the other.
You get a pure rock with giant grains and its behavior will be a lot different from the same material but broken up into little grains and then welded together by temperature and pressure. That behavior would be quite different. That's something that wasn't appreciated much in the seismological context because we didn't have that kind of data. An earthquake would happen, and we'd measure the amplitude of the waves at some distance in lots of different places and for us it was all homogeneous material. I think I managed to change that story and that's held up very well.
Frank Press and the 1960s
ZIERLER: Dave, unrelated question. On the social side being in Cambridge, Massachusetts in the late 1960s were you political at all?
JACKSON: Yes, I was. I had several conversations with Frank Press. That's one of the ways where he was my kind of associate advisor and I discussed with him because I knew him better. I knew the importance of his influence seismologically and I knew that also he was politically involved and involved in trying to do something to help out people, to make the work applicable. I was at the time possibly threatened with going to Vietnam as part of the Vietnam War. I was a conscientious objector and I talked to him about that. He was very helpful, and he would tell me, "I really understand your depth of concern about what's going on in the country and what people need." Then he said, "But play to your strengths. Take the things that you know how to do and figure out ways that you can make them help fulfill some of your interests and your needs in politics and in making the world better." And so we would talk about that on occasion, and it was very helpful to me. I didn't drop out of school.
JACKSON: I think he was afraid I might be ready to do that, and I might have if the Vietnam War had gone in a different direction or if I'd been called to serve.
ZIERLER: Dave, in the way that you interacted with Frank on these issues, did he already seem to be on a trajectory that would put him in Jimmy Carter's White House and leading the National Academy?
JACKSON: I would say it wasn't a surprise at all. He was very much involved in the space program. In fact, one of the instruments that he developed was then modified and taken to the moon. He definitely was much broader than earthquake seismology and he was certainly deeply involved in advancing plate tectonic theory. Now through planetary geophysics and building instruments he knew a lot about putting pieces together and putting teams together. He was really good at building teams, and I think had a very broad sense and so his role in the National Academy of Sciences and as the science advisor I think was not obvious but certainly unsurprising for a lot of people.
ZIERLER: Dave, after you defended, what opportunities were available to you? Where were you looking to get a job or a postdoc?
JACKSON: I knew that I wanted to be still involved in science in geophysics and so again I searched broadly. I settled on UCLA as a place that was interested in me and where I could do a lot of productive work. I also applied to Stanford, MIT, Caltech again, Scripps Institute, so I had a lot of possibilities, but UCLA worked out best for me. When I first got there, I followed my thesis work. We're talking about attenuation of seismic waves more from the seismological perspective rather than the laboratory instrumental perspective. But then, in talking with Leon Knopoff, who was also just a great intellectual leader and advisor for me, I got more involved in California earthquakes. There had been a magnitude 6 earthquake in Parkfield in 1966. It was after that that I went to UCLA but that was a subject of big interest in the California seismological community. So, I went more into earthquake seismology rather than I had been at MIT.
ZIERLER: Now, was it initially a postdoc at UCLA or you joined the faculty straightaway?
JACKSON: It was sort of middle ground. I had an "in-residence professorship" which was an academic position that didn't necessarily involve any teaching. It was kind of one step up from a postdoc. Then it was after I think about two years that I got put into a regular faculty position.
ZIERLER: Now, how new was UCLA to you being at Caltech, growing up there?
JACKSON: Obviously a different kind of institution but Caltech is very small. MIT is much larger, kind of middle ground; and UCLA is a whole city. It was a different environment, but I didn't find the change to be radical in any sense. I had academic heroes in all those places and so I didn't find it that much of a change. It didn't take long before I was involved in teaching classes and that was a change, but I also loved teaching anyway so again I really welcomed getting into the classroom.
ZIERLER: How soon after you joined the faculty were the origins of SCEC already up and running?
JACKSON: Oh, that was quite a while. I joined the faculty in 1969, and SCEC really started in 1989, with those discussions with Kei Aki and Leon Knopoff and other people at USGS. Those discussions resulted in the proposal that I was part of writing in about early 1990.
ZIERLER: Being back in Southern California being closer to the field work, how did that change your research agenda?
JACKSON: Oh, quite a bit. I spent a fair amount of time planning experiments. I wasn't an instrument builder as Frank Press had been, but I was very much involved in deciding where seismographic instruments ought to go to deal with future earthquakes. Part of that is putting together hypotheses about where earthquakes are going to happen and trying to anticipate what's going to happen on a particular fault. That was new to me, that kind of planning, and teamwork [laugh] that involved directing a bunch of associates to come out and do some field work. That was fun. I enjoyed that. But that was a new step for me.
ZIERLER: Dave, some technical questions. What work did you do early in your career on Rayleigh waves?
JACKSON: Not a lot really. Frank Press and Leon Knopoff both were involved in modeling wave propagation. I was involved in taking observed data and looking at amplitudes of those Rayleigh waves and saying okay, what kind of rock attenuation is involved in stealing energy out of those waves? But I was not doing any theoretical work and not making new observations of Rayleigh waves.
ZIERLER: Dave, what's the application of linear inverse theory to your work?
JACKSON: Oh, it's everything. [laugh] It's basically the technical approach to solving a lot of coupled equations. When you have two different equations about, seismic attenuation at two different frequencies but the same host material, you have coupled equations. They have a different effect on what you observe. Linear inversion theory is taking a set of linear equations and lining them up in such a way that you can harvest the information you have about the material that the waves have gone through. In many scientific studies, including seismological, hypotheses are described in terms of "parameters" used to calculate theoretical values to compare with observed ones. Earthquake rates at magnitude 4 in different regions, say northern and southern California, are parameters. Those parameters can then be used, with other ones, to calculate rates of larger earthquakes. Those can then be compared with observed rates.
Linear means that these equations are ones in which if you increase the size of one parameter, you also increase the size of the output by a proportional amount. It's much more complicated when the equations are nonlinear, but in some cases small changes result in effectively linear effects. So you start with the linear theory, then make small adjustments to correct for nonlinear effect... Like others, I did that often. Many people stop with that when the nonlinear theory is difficult to handle, but it is possible, and sometimes necessary, to use other methods. A lot of problems are not really linear, but for small parameter changes they appear linear, then that's a useful theory. There are many books written about it. These methods are useful not just in seismology and geophysics but biology, economics, etc. The whole world, I think, depends on inverse theory.
When there are many parameters, as is often the case, it is helpful to think of "parameter space," with the number of dimensions equal to the number of parameters. With just two parameters, the space can be plotted as a map, with the elevation representing the sum of squared differences between the theoretical and observed data. Then the pair of parameter values with the lowest elevation shows the best fit to the observations. For linear problems, the elevation contours would be ellipses. For more parameters, you would need a multi-dimensional space.
ZIERLER: A new term to me, you wrote about the edgehog method. What is that?
JACKSON: A Russian colleague of Press, Knopoff, myself, and many others devised a technique for describing the parameter space using what he called the "hedgehog" method. The idea is that if you can find one set of parameters that provides a good fit to the observations, then you can imagine a rodent eating its way through the low-hanging fruit in the parameter space. This method can be very effective for non-linear problems. I realized that for linear problem, the shape of the hedgehogs' consumption has multidimensional elliptical contours. I devised a much simpler method which I called the "edgehog" method, because it described the boundaries rather than the volume of the parameter space
ZIERLER: Dave, in the context of how you use the term, what does a priori data mean? In other words, isn't all data a priori?
JACKSON: Not really. Consider a planet, assume it is spherical, and we have one measurement that describe it total mass. We would like to determine two parameters: its radius and density. Even if we assume the density is uniform, our one mass observation can't resolve its two parameters. But if we have some example planet, like those in our solar system, we can estimate an average density, with big uncertainties of course, for our subject planet. That approximate density is "a-priori" date. Now we can add another equation to solve for the radius of our new planet, again with large uncertainties. It's a very simple example, but its relevant for much larger problems, in which the equations derived to explain our observations are inadequate to resolve our parameters.
ZIERLER: I see.
JACKSON: If you get new direct observations, say of the radius, then you may not need the "a-priori" density information, but you would still want to compare the inferred density with you're a-prior estimate.
ZIERLER: Dave, tell me about the development of the magnetometer array in Southern California and what that meant that for your research.
JACKSON: There is a phenomenon that we thought may have some relation to future earthquakes. The idea was that the stress that's building up might also have a piezomagnetic effect. It's essentially what happens in a phonograph needle. When you put a record on, and that little needle moves back and forth, it causes a stress in the phonograph needle which affects its electromagnetic properties. You can record that, and that's where a lot of our music comes from, actually.
JACKSON: Nowadays we don't use phonograph needles but in the Beatles' time it was all phonograph needles. The idea was that maybe we can use these magnetic changes to understand stress changes in the earth's crust. That would be very helpful in places where we can't use geodesy, or geodesy isn't accurate enough. However, there are other things that cause magnetic changes and so we had to recognize the difference between the seismological and the non-seismological causes of magnetic field changes. The earth is bombarded with electrons and protons from the sun and those cause magnetic changes in daily variations and long-term changes, and they're higher in some places than in others. Those have nothing to do with earthquakes. We use linear inverse theory to try to decouple those effects from space and from earthquake stress building up. We tried to do that decoupling, but we could not do it successfully. The stress effects were too small, and too like the solar effects. That led in part to my feeling that earthquake predictability is not really on the table [laugh]. The magnetic signals come from close-by things in the crust of the earth and the earthquakes themselves are resulting from five miles below the surface. The observable magnetic signals are not very strongly related to the earthquake stresses or the causes of future earthquakes. That was a good, fun experiment but we did not get the desired conclusion. That's the scientific method in action.
JACKSON: We had these observations, we had some explanations for them, and then we tried to put them to the test and make some new observations, and what happened was we had to put thumbs down on some of the theories.
ZIERLER: Dave, was there a period in your career when you were most intensively focused on crustal deformation?
JACKSON: Yes. When I got started in the earthquake work at UCLA then one of the basic problems is how can we understand the stress that's accumulating from past earthquakes and possibly building up for future earthquakes? It's certainly building up, but where and when and to what extent? And what's the effect of that stress on future earthquakes? Some of the energy goes into things that you can measure, not with seismic waves, but with geodetic surveys, or "geodesy" for short. We can measure elevation changes with leveling, distance changes by running a laser beam between two mountaintops. Those changes are what we call "deformation.", and it results in stress changes. Some of those changes result from fault motion due to plate tectonics. Geodesy has become very important in understanding the past and the future of earthquakes. So, I got involved in trying to deal with some very simple mathematical models for explaining what happens when a fault is deformed or when an earthquake happens and trying to use that to look at the future.
ZIERLER: Dave, in the last decade, your increased interest in looking at earthquakes at the global scale, global earthquake activity rates, were there advances in technology that allowed you to have literally a more global perspective of seismology?
JACKSON: There have been all along. Global earthquake networks have improved dramatically, so we can now determine locations, magnitudes, fault orientations, and other features much more accurately than before. Geodetic studies of deformation both on land, at plate boundaries, an even at some ocean-bottom sites, give us a much better picture of strain and stress accumulation at and between those sites. These improved data have now been used in global forecasts that incorporate both earthquake and geodetic strain data. One such forecast is the Global Earthquake Activity Rate (GEAR) mode specified at 0.1 degrees spacing over the whole earth. It is a testable model, and it shows very good agreement with locations of earthquakes that occurred after its publication in 2014. I was involved in that study, along with my UCLA colleagues Yan Kagan and Peter Bird, as well as other colleagues.
ZIERLER: Dave, in 1991 along with Yan Kagan you wrote a retrospective look at the seismic gap hypothesis. What is that hypothesis and how did it age over that decade?
JACKSON: That hypothesis starts by presuming that plate boundaries, and some very long faults, are divided into independent "segments", lined up end-to -end. Each segment is then assumed to be controlled by a "characteristic earthquake," big enough to release the accumulated stress on that segment. Then there is an assumed characteristic "recurrence time" needed for enough stress to recover for another characteristic earthquake on that segment. The idea is more than a century old, but it has been in the form of a "gut feeling" until the 1970's. Then seismologists looked at the history of big earthquakes around the Pacific rim ("the ring of fire"), selected some really big ones, and associated those with what seemed to be segments. In some places there had been two or more large events that seemed to have been on the same segment. Then the average times between those quakes was used to estimate the characteristic recurrence time. Other segments had just one quake big enough to call characteristic, so the recurrence time was assumed longer than the elapsed time. The fact that some sequences of apparently characteristic earthquakes occurred on individual segments got the seismological community very excited, and there were published tables listing the approximate magnitudes and recurrence intervals of over 100 segments. Then approximate probabilities were assessed for the occurrence of the next characteristic earthquake on each segment within the next 30 years. When the elapsed time was long compared with the estimated recurrence time, that segment was called a "seismic gap."
The theory was beautiful for several reasons. First, it was wonderfully intuitive. We know stress accumulates, is released by big earthquakes, and plate tectonics drives the stress accumulation. Also, very important, it is, within limits, a testable hypothesis. If big earthquakes occurred where the calculated probabilities were high, and many did, that lends support. It set out target dates for future earthquakes with some probability, so some statistics were involved. But given that there were 120-some locations around the Pacific Rim where there was good data and you could use this hypothesis to at least estimate when an earthquake of a given size was due, that was an intriguing hypothesis. I bought into it at first. Kagan and lots of other people did as well. Frank Press bought into it big-time [laugh], published some papers about earthquake prediction that were based in part on that hypothesis.
JACKSON: Also, the seismic gap theory is a dangerous idea. Okay, you had a big earthquake. Now can you relax? Can you rebuild your cities after a San Francisco earthquake and not worry about future earthquakes for another hundred years? Not a good idea.
ZIERLER: The RIP in your paper with Kagan and Geller, Characteristic Earthquake Model 1884 to 2011, is that a "rest in peace" RIP?
ZIERLER: What were you making into an obituary about the characteristic earthquake model?
JACKSON: Ok, so why did we declare the seismic gap hypothesis dead? The main reason is that one needs to look at the overall performance of the model, using a prospective approach: count successes and failures using quakes that occurred after the probabilities were published. The proponents were looking only at the successes, and retrospectively. We noted that here were many low-probability segments on which large earthquakes occurred, and high-probability ones where none occurred. We compared the seismic gap theory results with a simpler "null hypothesis," assuming that the rate of big earthquakes on each segment would be equal to its rate before the probabilities were calculated. Guess what? The null hypothesis performed better. Even more damning, many huge earthquakes occurred where the characteristic magnitudes would not have allowed them. The 2011 Tohoku earthquake, magnitude larger than 9, was an example, and we thought that failure was so egregious that we had to call the gap theory a failure. The RIP was saying, okay, that hypothesis is not working, especially the characteristic magnitude part. Let's abandon that model and move on to something that is consistent with the data, at least.
ZIERLER: Meaning that it requires a new hypothesis or requires abandoning the gap hypothesis?
JACKSON: Some of both. It certainly requires abandoning that hypothesis as it was formulated because that hypothesis was specific, and the tests were conclusive. Also, the seismic gap theory is a dangerous idea. Okay, you had a big earthquake. Now can you relax, as the model suggests? Can you rebuild your cities after a San Francisco earthquake and not worry about future earthquakes for another hundred years? Not a good idea.
ZIERLER: Right. But could the hypothesis be fixed?
JACKSON: Some would like to patch up the seismic gap theory. Perhaps it could be done, but there are some major flaws in the way the theory was formulated. First, the idea of independent segments does not describe the way quakes happen. As soon as you draw boundaries, some quakes will cross them and screw up the scheme for counting characteristic events. Also, the earthquakes labelled as characteristic and used to compute recurrence times occurred long ago, when seismic data were not so precise, so it is hard to define and count characteristic events precisely. And the idea that characteristic earthquakes empty the stress bucket on a segment, or anywhere for that matter, can't be validated in any meaningful way. And we know that large earthquakes are clustered, so the concept of a recurrence time for a fixed segment is completely ambiguous. So, if you would like to patch up the theory, bring a lot of plaster.
In Defense of Huge Earthquakes
ZIERLER: Dave, a short article you wrote in 1996, The Case for Huge Earthquakes. I can't help but smile. It sounds like you're writing in defense of huge earthquakes! [laugh]
JACKSON: Yeah, I am. [laugh] I don't recommend having giant earthquakes, but rather, recognizing that they can and will occur. It's in part related to this idea about seismic gaps, because one of the parts of the seismic gap hypothesis is that when the stress and/or deformation reaches a level released by a previous record-breaking earthquake then it is about time for another one. The problem is that we can't assume that our records include the largest possible event in that place. Records are made to be broken, and recent earthquakes have done that. Well, when data show that earthquakes aren't obeying those times, magnitude limits, or probabilities, the gap hypothesis must be re-evaluated. Perhaps there's something wrong with the data or it could be that maybe the earth could build up stress for a longer time and have a bigger earthquake.
Part of the preexisting story was that when you had a San Francisco earthquake that's the upper magnitude limit and so you look for an earthquake of that size. When the deformation has been big enough that you ought to have such an earthquake then you should be concerned about that. Well, you should always be concerned, but not just because of this limit. It could very well be that the San Francisco earthquake was not any kind of an upper limit to the ones in our future. If you buy the seismic gap theory, you're also buying an estimate of the limiting magnitude being the largest known earthquake in recent history on your "segment." But "known" means observed within a limited time, and bigger events likely happened before that. If bigger ones occurred earlier, we don't know when, so we can't estimate a recurrence probability. So, we don't like huge earthquakes, but we can't neglect them.
JACKSON: You must understand that they do.
ZIERLER: Like you mentioned with COVID, we want it to be over, but nature may or may not have other plans.
JACKSON: That's right. That's right.
ZIERLER: Dave, another nomenclature question. Long-term earthquake clustering—first, what is earthquake clustering and is there a short-term earthquake clustering?
JACKSON: Good questions. Yes, long- and short-term clustering both occur. Clustering refers to an increase in the probability that an earthquake will happen given that another one has just happened. A traditional idea is that stress builds up at some location, a big earthquake happens, it releases the stress, then it takes a long time for the stress to recover so that you can have another earthquake like the San Francisco earthquake. Clustering is the opposite of that. Clustering is what's most observed. Aftershock occurrence is a form of clustering. Sometimes a big earthquake may be followed by another big earthquake or even a bigger earthquake. We don't know the limit of the size of earthquakes. So, you may have a magnitude 8 earthquake in San Francisco, but that doesn't prevent a magnitude 8.5 in waiting someplace. The stress is already there. The San Francisco magnitude 8 hasn't relieved all the stress.
The stress doesn't occur just in one place or in one large region, the stress is redistributed. There are some places near a fault where the stress becomes much higher, and that's what aftershocks are caused by, we think. Aftershocks are observed in almost all California type earthquakes. Only for very, very deep earthquakes, hundreds of kilometers below the surface, do we not really get aftershocks. Aftershocks are one form of clustering, but the fact that big earthquakes might follow each other is also a form of clustering. It does occur, and it is contrary to this theory that the earth must rest after a San Francisco scale earthquake.
It's true that some large earthquakes might be followed by a century or more till a similar one. But short intervals, too short for the stress to recover since the last event, are commonly observed around the globe. The earth doesn't have to wait. It's ready to go just about any time. That's another reason why I'm not expecting earthquake prediction to succeed in the foreseeable future.
ZIERLER: [laugh] Dave, you mentioned geodesy earlier. Is there a portion in your career where you remember that becoming particularly relevant?
JACKSON: Yes. That has to do with this crustal deformation, and geodesy is a way of measuring positions of points on the earth and those respond to stresses. To the extent that we can measure that response, we can then infer the rate of stress. That in turn can tell us something about earthquake potential, whether it increases or decreases. We use geodesy to identify and quantity the deformation. We generally find that the places that are deforming are the places that have more frequent earthquakes. We can tell something about stress accumulation, but we can't tell the details and particular times. We can say that the rate of earthquakes is increasing or decreasing, but we can't put a date on any future earthquake. We do find that places having lots of aftershocks also tend to have more deformation. It's definitely an important part of understanding what's happening in the earthquake process.
When a significant earthquake occurs, we try to measure the immediate deformation, and the following deformation, as quickly as possible. I was involved in planning some experiments using the global positioning system (GPS) in the very early days back in the late 1980s and 1990s. I got involved in some heavy-duty field work on the 1989 Loma Prieta, the 1992 Landers, and the 1994 Northridge earthquakes. There have been some geodetic controversies. There was an earlier debate about the effect of deformation coming from leveling data. There was this idea that the crust had been uplifting in an area along the San Andreas fault near Palmdale. The story about the Palmdale Bulge—you probably read something about that.
JACKSON: I can tell you've read a lot of my papers.
ZIERLER: Yeah. I do my homework.
JACKSON: Yeah. [laugh] And so there was this apparent uplift of about a foot that was reported to be a result of stress increase in the crust and maybe due to an upcoming earthquake. Again Frank Press and I differed a lot on the interpretation of that. The reports that led to this theory were based on leveling data. Leveling surveys are used for many other purposes besides earthquakes. They are used to measure landslides, subsidence due to water withdrawal, earthquake uplift, etc.. Leveling is used around virtually the whole United States and particularly along the coastlines because we're interested in sea level changes. Those can be caused by plate tectonics, as well as river flow and a lot of other things, so it's useful to measure the uplift or apparent uplift of what we call benchmarks, survey markers along the coast. Elevation on land is compared to tide gauge records just offshore.
Apparent uplift at the coast might be caused by land uplift, or by a temporal decrease in water level. And of course, apparent down drop could be cause by land subsidence or sea level rise. That's a source of geological and especially geodetic uncertainty. How do you understand the playoff between ocean change and solid earth change at this boundary?
Levelling can be very precise, even more so than GPS elevation measurements. But levelling can be subject to systematic errors, such as errors in the leveling rods (10-foot-long vertical rulers) and atmospheric refraction caused by temperature variation along the line of sight. After the "Palmdale bulge" reports, our team at UCLA examined the levelling records and determined that systematic errors were a much better explanation than the hypothesized tectonic uplift. Prof. Ross Stein from Stanford and USGS, then carried out precise levelling experiments, controlling the systematic errors, and measuring their effects. They agreed with us. Now the "Bulge" has been completely forgotten, although not officially dismissed. One more reason I'm skeptical about earthquake prediction!
ZIERLER: Given what you've already said about earthquake prediction, and your paper in 1996, Hypothesis Testing in Earthquake Prediction, does that mean that hypothesis testing is a way to demonstrate the futility of earthquake prediction?
JACKSON: It's broader than that. Testing hypotheses is a fundamental part of science. In the scientific method, you get ideas, you make some hypotheses, you test them, and you reject or revise as appropriate. Your hypotheses rest on assumptions as well as observations. To predict earthquakes, you need to validate your assumptions. When your new data don't fit what you learned from the old data you've got to change your story. It may be that you change your definition of what you're looking at, the size of earthquakes maybe, or it may be that there's something wrong in this story and you have to modify it. The stories about earthquake prediction all involve many assumptions. But for prediction that justifies action, you must make all the piece fit together. That means the hypotheses need to be specific. If you say, there is some evidence that earthquakes are more or less periodic, that is not sufficient to call it prediction. You must specify where, how big, and what the period is, and how uncertain all these parameters are, before it is useable. If you've had an earthquake in 1906, and it was the start of a hundred-year or so period? Then the next ought to be 2006 or so. If true, we are overdue and subject to exceptional risk. But the periodic assumption can't be validated, and the info is not useable in a scientific prediction. At least you need to show that your assumptions can be validated in many other places with equivalent situations.
That's where the testing comes in. You can apply that theory for San Francisco but must also apply it to a lot of other places. Does it agree in those other places? And if it does, if you can document it in a lot of other places then it really is something you need to be especially prepared about. That would make it a useful prediction. If it just occurs at random, some places it does seem to agree and some others it doesn't, then it's not really a predication nor valid science. You say, okay, look out for an earthquake. We already know that, but we needn't believe that a recent earthquake obviates another one soon, and an old earthquake is more dangerous than others. That's where the hypothesis begins to fall apart. I am not trying to discredit earthquake prediction per sec, but rather to evaluate the many assumptions that go into seismology and earthquake prediction.
Science Communication and Managing Expectations
ZIERLER: Dave, a more general public policy and science communication type of question. For when scientists like you or Bob Geller tell the world unfortunately earthquakes cannot be predicted, that's not what the public wants to hear, right? They want to hear that earthquakes can be predicted. What have you learned about communicating the scientific process, doing the best job you can, what have you learned to be effective in communicating how science works to the public?
JACKSON: That is definitely a tough one and you probably know that Geller and I and Kagan and Mulargia, this Italian coauthor of ours, we all caught hell because of the publication.
JACKSON: Including from Frank Press. Frank Press was quite mad at me for that publication.
JACKSON: A large part of it was that we used a specific definition of earthquake prediction, but most readers ignored that definition. We didn't say that earthquake research was not valuable, but that if you expect a warning immediately before an earthquake happens you are dreaming. We have a lot of information about where and how often earthquakes are likely, and about preparations that can improve safety. So, you should prepare now. You should always do the obvious things knowing that earthquakes are going to happen sometime, rather than waiting for a telephone call or email that says, earthquake on the way, get out of town. That was our message. People didn't read that definition of earthquake prediction, they just read the title. We focused a lot on the reasons why prediction is likely to fail. That's important, because the underlying assumptions are important, and there has been a lot of wishful thinking rather than good science in many prediction efforts. But we needed to spend more time emphasizing the main message. I learned a lot about the communication, and about what we can and cannot say about earthquakes. I also learned that we need good definitions of prediction, forecasting, conditional probabilities, and warnings, which must be tailored to specific audiences like seismologists, officials, engineers, teachers, and the public.
JACKSON: We had a lot of discussion, particularly between me and Bob Geller, about that title [laugh] Did we say that earthquakes could never be predicted under any guise? Did we say we can never predict the location of any earthquakes? No, we weren't saying that. Geller insisted that we needed a punchy title. [laugh] It got too punchy.
ZIERLER: Mission accomplished. [laugh]
JACKSON: Yeah, we got punched.
ZIERLER: So what exactly was Frank Press's concern, that you seemed to have foreclosed on the possibility of future advances?
JACKSON: Yes. Yes. He was concerned, first of all, because he'd written papers that said earthquakes could be predicted and he was being contradicted. [laugh] He didn't like that. But I think many people, including some at the National Science Foundation who were funding our research, thought, "You're going to cause a reduction in funding of earthquake science because of this paper." Well, that's a testable hypothesis. We have looked at the funding. The funding has grown, quite slowly, as it had in the past. [laugh] This research is under-funded, but our paper had no discernable effect on that. In fact, for the most part, it was ignored after the punches that we got for having published. [laugh] So those who suffered were the authors, not so much as the readers of that paper.
ZIERLER: [laugh] Dave, just so I understand the nuances, so what about earthquake forecast models? When you were looking at short term and time independent earthquake forecast models what does a forecast model mean in a scenario where earthquakes can't be predicted? What's being forecast?
JACKSON: We decided to avoid the term "prediction" because it is such a loaded word and subject to snap judgments. So many of us used the term "forecast" to mean something we can accomplish without having to say, "time to duck." A forecast provides general information about earthquake probabilities, useful for planning, rather than a short-term emergency. It took some time for the seismological community to adopt the different terminology. The forecast terminology has now been widely adopted, partly because of the reaction to our paper. Now we can write constructively using the distinction between prediction and forecasting. We can now provide useful information, but it isn't going be the exact time , place, or magnitude. We said, "suppose you take a time interval of 10 years or whatever, we think there are some well-defined places where you'd expect an earthquakes over magnitude 7 at a high probability within that time period." That's a forecast, not a prediction. That's not saying the time and it isn't saying that it must occur at that place. We're saying it's much more likely.
Now we can discuss, test, and debate the scientific information without the loaded nomenclature. In fact, the seismic gap theory was used as a forecast, not a prediction, and we surely debated it. Now the community has adopted this forecast terminology, and that is an improvement for which we take some credit.
ZIERLER: So, even more nuances in the terminology. What about earthquake likelihood model testing? Is that not a forecast?
JACKSON: That is forecast.
ZIERLER: That is forecast.
JACKSON: That is forecast.
ZIERLER: Why not just call it forecast then?
JACKSON: That wasn't my choice. [laugh] That title comes largely from the US Geological Survey who published regional forecasts over California, Utah, Washington, the western continental U.S, etc. They constructed models of individual faults and listed probabilities of earthquakes of specified sizes on those faults. They used the term "likelihood," rather than probability, because in some places the information was more uncertain than they would like. It's a good idea, and in science there are lots of times when you use different words to varying grades of certainty or importance. I think it's been a very constructive enterprise to do these likelihood models. Now my job as a statistician is to formulate tests in which you could say this model is working or not. If not, let's try to modify the model, change it, and test the new version. That's a work in progress.
ZIERLER: The work on next-day earthquake forecast, next day, does that get us closer to earthquake early warning or is early warning only minutes or seconds?
JACKSON: So far, I think the real experimental work is on immediate occurrence. And now, we are using the term "warning" instead for prediction or forecasting. The early warning means the earthquake has already happened and it's been measured just now but someplace else, and the seismic waves are on their way to you. The warning is to say those seismic waves are coming, not that the earthquake's about to happen. It's already happened. There's a middle ground to that because of clustering. Clustering occurs when an earthquake just happened and some aftershocks are also happening. Some of those aftershocks could be bigger than the first one that you knew about, so there's an enhanced probability of an earthquake happening at the place very near to where the first observed one is. You can send out a warning on a magnitude 6.5 earthquake and that warning is also useful if a magnitude 7 happens at that same place within minutes.
So people would get the warning, they'd duck and cover, then move to a place that's isolated from strong shaking, and then, as long as they're patient enough to allow time for aftershocks, they're much safer than they would've been. That's a predictive kind of operation but it's not predicting the earthquake, it's predicting the waves that have already started and they're on your way.
ZIERLER: Dave, your contributions to the big survey in 2009, California Earthquakes 1800 to 2007, we established right at the beginning of our conversation that you don't see any cycles in earthquake history, right? So what then about the survey is valuable beyond that criticism of stamp collecting, that we're just collecting this data? What's the purpose of a big survey if you know already that there aren't cycles to be deduced from it?
JACKSON: That paper was not about cycles, but rather about reviewing and improving the information about the locations, times, and magnitudes of past earthquakes. Those data are important whether they show cycles. There have been many different publications with varying definitions of magnitude, different time standards etc. We wanted to standardize the data, eliminate duplicates, and delete information that had already been corrected after some early publications. Anyone could use our publication to look for cycles or other effects if they wish. That would be fine, but of course it would be based on selected data, and the selection criteria should be carefully specified. In our own work, we found no evidence of periodic behavior, beyond the random occurrences you commonly find in large data sets. We find lots of clustering in those data, well beyond random.
ZIERLER: So, what are the takeaways then absent the cyclic predictions of engaging in such a big survey like that? What do you learn from that exercise?
JACKSON: I think you must remind people that surprises are coming, and you can't rely on the time forecast. But you'd better rely on the fact that earthquakes are happening. You need not trust too much in apparent patterns, but absorb the basic message that earthquakes can be dangerous, and you need to prepare for them. You need to take appropriate measures, and you need to decide what those appropriate measures are, and so just mentioning the name earthquakes is constructive, and what I tell people who find out I'm a seismologist and they say, "Oh, when's the next one coming?"
ZIERLER: [laugh] Yeah.
JACKSON: I just say, "We don't know but we know earthquakes are going to happen sometime and you need to do some obvious simple preparations in your home. You need to have water in case a water main fails, and you need to have a place to poop when your toilets back up, and you need to know where your children will be if there's an earthquake at school." So things like that. Appropriate measures are used a lot, but some people forget. I say, "When you hear a radio report or you see in the LA Times or the New York Times or whatever an article about an earthquake coming, just take the word earthquake and say, oh, okay, I forgot to do the normal preparation. Make one step in improving your safety. It doesn't have to be everything you know you ought to be able to do but make one step. At least talk to your kids' teachers and ask, "okay, what's likely going to happen if there's an earthquake on campus, how am I going to find my kid? How long should I be expected to wait?" So do those kinds of things. At least that's something that's doable. You don't need to say, "Oh, my God, earthquakes: stop the press!" But I think that's constructive. When you have a survey about earthquake occurrence it may seem to imply "Oh, my God, the big one is coming." Maybe a big one, or "the" big one, isn't coming. Nevertheless, at least hear the word EARTHQUAKE and say, oh, yeah, let's do something constructive about earthquakes in general.
ZIERLER: A little bit of vigilance just to remind ourselves that earthquakes are on the way even though we can't tell when and where.
JACKSON: That's right.
ZIERLER: Dave, what have been some of your most recent interests in paleo earthquakes and events?
JACKSON: That grew out of some SCEC work by Clarence Allen at Caltech, and Kerry Sieh, one of Clarence Allen's students at Caltech who was about my vintage. [laugh] But a lot of other people have been involved in looking at faults, digging trenches across them, and finding that sediments that have been displaced, presumably by earthquakes. These sediments break through some layers below a certain level and others above that level have not been displaced. That's apparently because the earthquake was before those later layers were put on top. So paleo seismology involves finding a time interval after the displaced layers and before the undisplaced ones.
There was a big study putting together a lot of paleoseismic data into a table in one of the SCEC reports. In that table for each of the sites there were 30-some trenches that had been dug to get these dates, there is a date of the most recent displacement at each one of those sites. I just looked at the dates and said on average there are about 20 to 25 years apart for the whole state. But when you add them all up recently there have been no displacements in the last century. What's going on? [laugh] So I just did some simple statistics on these dates and said, okay, could this occur at random? And there are random variations at each site and maybe somehow just by coincidence they all have a delay from the last earthquake until now but the chance of that happening at random is less than 1 percent. We usually kind of think of a 5 percent anomaly as being one we can tolerate. But 1 percent, no.
So I wrote this up and said that there's a discrepancy here and it's quite possible that the data themselves are not relating to earthquakes but some other cause of the displacement. That's still one that people aren't accepting and based on my history in dealing with earthquake prediction I'm not surprised that people don't like that idea. [laugh] But my first study of this that I presented at the Seismological Society was in 2014 and so eight more years have passed and still no displacements at any of these sites. I found that in New Zealand there's a similar story, a bunch of sites on the faults and none of them have had an earthquake-type displacement in the last century, so I strongly believe there's something wrong with the data. But people know that I'm a contrarian so that's where we are on that one.
ZIERLER: [laugh] Dave, just to bring our conversation right up to the present, what's the current state of play with the National Seismic Hazard Model?
JACKSON: That I don't know. Partly I've been unable to keep up with the developments over the last year. You might know—I had a stroke in July of 2021 that put me out of business for a while and so I haven't been able to keep up with the meetings that go on to sort of plan the work that's going on. But I know that a lot of the plans are for similar kinds of studies that have happened in the past and to increase the importance in these studies of some purely theoretical modeling of faults, what they call physics-based earthquake forecasting. I don't know exactly how they're going to do that but I'm curious about what they're planning to do.
ZIERLER: Dave, now that we've worked right up to the present, for the last part of our talk I'd like to ask a few retrospective questions and then we'll end looking to the future.
ZIERLER: So over the course of your career what do you see as your most significant contributions to the field?
JACKSON: I think it's an appreciation of the need for testing hypotheses. That we need to turn the hypotheses into detailed enough stories that a time interval like ten years or so can help to validate or invalidate the hypothesis that we're dealing. More focusing on the scientific method, putting the whole picture together rather than individually trying to work on just an observation or just a hypothesis but taking observations, hypothesis, and logic and using those things. [laugh] So that's, I think, the most important. That's where I see the scientific community doing great work but not putting things together in the way that Kei Aki recommended when he formulated the Southern California Earthquake Center.
ZIERLER: To go all the way back to when you were an undergrad at the Seismo Lab and you were taking in all of the debates of the day, from then to now what debates have been settled and what remain as open as when you first encountered them?
JACKSON: [laugh] Oh, I think none of them have been completely settled.
ZIERLER: That's good. That's good.
JACKSON: Because the debates all have some foundation in correct observations. The difficulty is generalizing those to look at some places where the specific forecast or specific hypothesis is made, and it may be made retrospectively. You say, oh, I see what happens, and maybe you look in one other place and you say, oh, yeah, it happened again. That gives some support to the theory. But what needs to be done is to look at many different places to get a more representative set of data. That's hard to do because sometimes you get funding for one experiment but the other. There is a remedy: justify the cost of a testable hypothesis in your proposal.
So, I don't see any of the things that have been debated really being resolved but there's one case which is this Palmdale Bulge that goes back to the late ‘70s and early ‘80s that that was due to some stress accumulation that may result in an earthquake on the San Andreas. Nobody except a handful of us have come by and said the data are not correct, which is the case, that the leveling data that were used in that observation of the uplift—the uplift itself didn't happen. There is a fellow at USGS named Ross Stein—I don't know if you've encountered his name—but he had been part of the original leveling studies and so forth. He went back and looked at data and he agreed with us that that uplift didn't happen.
And people resisted that contradiction for some time but about ten years later there was a review of US Geological Survey studies in California, and they did not mention the Palmdale Bulge. [laugh] So I regard that as settlement of that debate even though it wasn't a definitive kind of admission that there was a problem, just that—what problem? [laugh]
ZIERLER: Dave, the fact that you say that really none of the science is settled is a perfect segue to my last question looking to the future. To the extent that you have opportunity to interface with students who might be curious about where the field is headed, where they might focus their energy, the fact that so much of the science is still open and ready for additional investigation, what do you see as the most interesting and fruitful areas of focus in the coming decades in seismology and geophysics?
JACKSON: And that's a wide-open question. Clearly better observations are always a part of the focus and you probably heard about some of these studies that involve optical glass fibers being stretched and you can measure the length of that stretching from one point to a tiny, tiny little crack in the optical fiber that tells you displacement has taken place. These optical fibers are everywhere. That's a really big advance, I think, in seismological and geological observations. Using these data and getting much more specific locations of earthquakes and descriptions of earthquakes can be helpful in testing some hypothesis.
And then, there are a lot of hypotheses about how you can derive geodetic information from new kinds of observations and then put those together with the story of stress buildup and deformation and possible interaction with earthquakes and the result of earthquakes. When earthquakes happen, we need to know not just that there was shaking but what it's done to the surface of the earth and to the subsurface part of the earth down to five miles deep and deeper. When we can measure those things, we can test some hypotheses about stress transfer and what is the effect of stress, what causes it and then what can result in it. I think those things are really exciting openings for new science and new ideas.
And then, testing. I think it's interesting to get people who are not seismologists involved in testing hypotheses. How do you take some of these hypotheses and take some data that might be relevant and put them together and now say how long does it take before you would know if a hypothesis is correct? And some of these same people who are working on COVID are also now interested in those kind of tests, what are the results of things that we can observe? So I think there's plenty for young people to do. We could talk a lot about that and, of course, there are proposals being written to involve young people in those kinds of questions.
ZIERLER: Dave, this has been a terrific conversation. I'm so glad we were able to do this. I'd like to thank you so much.
JACKSON: You're very welcome.