The Habitable Planet: A Systems Approach to Environmental Science
Oceans Interview with Mark Cane
Interviewer: You told me earlier that you are a mathematician. How did you get involved with climate?
MARK: I was working for a while at the Goddard Institute for Space Studies in New York as a mathematical analyst and programmer during the Vietnam War. They started on a project to look toward the impacts of satellite temperature soundings on weather forecasting. We were doing these simulations, and it was pretty clear to me after a short time, that the people I was working for really didn’t understand much about the atmosphere. So they brought in this consultant from MIT, a man named Jule Charney, and I was amazed because he really understood how the atmosphere worked. I thought this was one of those unknown mysteries and the fact that this guy really had some sense of the mechanism of this thing was remarkable to me. I later learned how important a figure he actually was in this field.
Later on I decided to go back to graduate school, and I thought about going back in something that would now be called computer science, which if I’d done, I’d probably be a lot richer than I am today! But that’s another story. Instead I went to MIT and became Charney’s student. He wanted me to work on some problem that had been around a long time. It was a very mathematical problem. There wasn’t much physics to it and I wasn’t that interested. But I didn’t want to just tell him no. I felt I should offer to do something else. There was someone else who was supposed to do a model of this phenomenon in the ocean called the equatorial undercurrent. But he didn’t want to learn how to program and I already knew how to program, so I said, “I’ll do that problem.” So I started on that. The ocean is very interesting in its own right, but I also had all this training in atmospheric science.
When you have an El Niño event, there are things that happen around the world. They don’t always happen, but they usually happen and they have some bad consequences for people. There is typically a poor monsoon in India, which if it’s poor enough in old times leads to famine, but even now is a huge hit for the Indian economy. There are droughts in Southern Africa. There may be flooding in East Africa. There can be droughts in Ethiopia, drought in northeast Brazil, which is the poorest part of Brazil. When there are droughts people leave and they go to the, to add to the slums of Rio and San Paolo where they go into the Amazon to find work. In Peru and Ecuador you have catastrophic flooding in many places that washes out infrastructure; stops their oil wells from producing. In the highlands of Peru and Bolivia you have drought and people are in trouble again. So these El Niño events cause difficulties in a lot of the world and to some extent if we could forecast them and get this information out there, there’s the possibility of taking actions that will mitigate these impacts.
So I got interested in El Niño, which clearly involved the ocean, involved the atmosphere, and was in the tropics. I related to a whole lot of things that I knew about, or at least should have known about, or might have known about, or hoped to know about. And so I started working on that and tried to do this with a very, very, very simple model. I worked at that for a while and I had some conversations in particular with an oceanographer named Varkey who had done a tremendous job of putting tide gauges out in the Pacific to get a picture of how the ocean changed.
He steered me to the work of Jacob Bjerknes who had a hypothesis about how El Niño worked. It wasn’t complete, but the basic idea that still is there for us today was that this was an interaction between the ocean and the atmosphere in the tropical Pacific. He noticed that the state that we consider normal is very odd, because the western Pacific is warm, like 30 degrees Celsius almost, high to mid-80’s Fahrenheit. The kind of water even my wife would swim in, it’s so warm. Bathtub temperature. The other end, the eastern end of the Pacific, has temperatures like 20, 22 degrees Celsius, around 70 Fahrenheit, which some of us swim in but a lot of us don’t. Why this enormous temperature contrast? Both ends of the ocean get about the same amount of sunlight. We’re talking on the equator here so you expect them to be heated up a lot. And the odd thing is that you have this normal state which is so skewed. The reason for it is the way that the ocean responds to the winds.
This is the region of trade winds, so winds from the east tend to blow the water toward the west. But the surface waters are governed by something called Ekman Transport, which means because the earth is turning, the currents in the northern hemisphere are deflected to the right of the wind, and the ones in the southern hemisphere are deflected to the left of the wind. So on the equator then, because there are easterlies, the waters are tending to go away from the equator in both hemispheres. You have to replace that water and that water is replaced with water that comes up from below, which is colder. And in fact, at the equator it can be considerably colder than what is at the surface. The ocean has the structure where there is a layer of warm water, which at the equator is 100 to 200 meters deep. Then there is a very sharp thermocline, a region of very sharp temperature change. In another maybe 50 meters or so, the water temperature will change by 10 degrees Celsius. Then there is another 10-degree change in the next 3,000 meters to the bottom, 4,000 meters to the bottom. You pull up this thermocline in the east because the winds are pushing all the warm waters at the surface towards the west, pulling it up at the east; and so the water that’s coming up can be really cold. That’s the principle reason why there is this temperature change.
What Bjerknes did that was so important was notice that if you had cold waters on the eastern end of the ocean and warm on the western end, then that was going to set up a pressure difference in the atmosphere. So you get higher pressures at the eastern end, lower pressures at the western end, and then the air would flow down, the pressure gradient would flow downhill from east to west. In other words this temperature contrast between east to west, which was generated by the winds, was also helping to generate the winds — a positive feedback between the ocean and the atmosphere. Bjerknes, writing in the 60s, called it a chain reaction; that’s when nuclear energy was good and nobody worried too much about it. Bjerknes noticed that El Niño, which had been known to people on the coasts as a coastal warming, actually extended out into the central Pacific, all the way from the coast line to the date line. The Pacific is big. That’s a quarter of the circumference of the earth. That’s a lot of area to change the water in and that’s a big disruption in the temperature, the thermal boundary condition that the atmosphere sees. That was enough to be really important and perhaps to account for all of the global impacts that Bjerknes knew were associated with El Niño.
Now he knew about that in good measure because of the earlier work of a man name Gilbert Walker who was actually a mathematician and not a meteorologist. Walker took over the observatory in Poona, India, in the earliest years of the 20th century. This was right after a severe famine associated with the El Niño events of 1899. In fact, this observatory had been founded in the aftermath of the 1877 El Niño, which was probably the most destructive one in history. There were very severe famines in India and China and Ethiopia and Northeast Brazil associated with that event. The British set up this observatory to try to understand and predict the monsoon since it was the failure of the monsoon rains that had caused this famine. Walker focused on the Southern Oscillation, which his friend Norman Locklear, who had been the editor of Nature Magazine, knew about and had told him about. The Southern Oscillation is a change in the sea level air pressure usually measured at Tahiti in the Eastern Central Pacific and Darwin on the northern coast of Australia. It turns out that if you look at the fluctuations when the pressure is higher than normal in Darwin, it tends to be lower than normal in Tahiti and vice versa. Sea level pressure is a measure of the mass of the atmosphere above you, so you’re getting a seesawing of mass between the eastern Pacific and the far western Pacific. Walker had the idea that whatever was going on with the Indian monsoon was not caused locally but was related to these more global scale things so he started looking for connections to this Southern Oscillation. And he found them in the Indian monsoon and all sorts of other things, like the wintertime temperatures in Calgary. But he didn’t pick up on the connection between the Southern Oscillation and the El Niño phenomenon, which initially was a coastal warming off the coast of Peru and Ecuador named because it came after Christmas, after the time of the birth of the Christ child, El Niño. And it was an annual thing that was relatively benign but once in a while it got out of hand and interfered with the livelihood of fisherman and so on.
Bjerknes put together Walker’s Southern Oscillation and this El Niño index because he saw this connection, this feedback between the ocean atmosphere and saw that you could think of the El Niño state as just the opposite of the normal state I described earlier. So imagine that for some reason you get a pulse of water coming through the ocean and pushing down its thermocline, this region of sharp temperature gradient in the east. The water that is coming up is not going to be as cold as it was before because you’re now moving the top of the thermocline down, and so warmer waters are coming up to the surface now. The temperature gradient is less, so the pressure gradient will be less and the winds will be less. And that will reinforce things because now you won’t have as much upwelling, you won’t have as much current driving the waters across the Pacific and so the thermocline will relax more. The temperatures will decrease more. The winds will weaken more and so on. Again, it’s a positive feedback now, in the opposite sense. Bjerknes saw this way of thinking about the El Niño phenomenon as the same sort of physics as involved in the normal state; the same positive feedback, but going in the opposite sense. He knew that the El Niños came regularly, but roughly every four years. But he didn’t in this picture he put together understand why you got the turnaround from one state to another. The positive feedback story would seem to lock you in one way or the other. Either you would stick in the normal or stick in the El Niño state. The answer to that came a little bit later because that’s involved in ocean dynamics. And really, it was, more than anything, Klaus Zwerky’s work that pointed people in the right direction because he showed essentially that there was a big movement of mass in the ocean when an El Niño event happened, that the warm layer of water in the west moved to the east. It wasn’t just something about surface temperature. It was this whole dynamical change that involved at least the upper layer of the ocean over much of the tropical Pacific. We set out to make a model, a numerical model, that would capture what we thought following Bjerknes and Zwerky would be the essential physics, and that physics involved Bjerknes’ feedback, but it also involved some equatorial ocean dynamics, which had actually been the thing I’d worked on most from the time of my PhD thesis on. And so we made a model, a relatively simple model, that was designed just to capture those things we thought were important. And to cut what was actually a long story a few years short, we got it to work in the sense that it would set it going and it would produce oscillations about every four years, but regularly with the general special characteristics of an El Niño event.
Interviewer: Could you describe that more? What data did you use to create the model? How far did you go to get information to put into this model, and then when you ran it, what did it look like?
MARK: A computer based model is a collection of lines of code for a problem like this or equations describing the governing physics about how the water moves, about how heat is exchanged between the ocean and atmosphere, about how the atmosphere moves in response to this heating. And there’s a lot of other stuff about getting information in and out. One of the things people often ask is what data did you use and how did you use it? Most of the data that we used was something we absorbed in our head first, that gave us a sense of what the structure of this model should look like. So it wasn’t this very precise notion of okay, we made a model of exactly the governing equations, which we took out of a textbook and we figured out what data to use. In this case, and this was 20 years ago when computers weren’t all that powerful, we had to simplify the equation so we could run this thing long enough to get decades of behavior out of it. But we did do something that with hindsight seems to me now one of the cleverer things that we did. It’s proven very difficult in models that are much more complicated, much more complete than ours to get the mean climate, the average climate, about right.
You make a model of the atmosphere and you tune it up putting in observed sea surface temperatures. And you get it so that it does a pretty respectable job of simulating the atmosphere. Then you have a model of the ocean, which again is pretty complete. As far as we understand the physics, it has to parameterize, it has to represent certain physics more crudely because we can’t represent every parcel of water in the ocean. But you get this model so that when you drive it with observed winds and observed heating, you get a pretty good representation of the ocean. You couple these two models together and instead of returning the climate that’s out there in the world, they drift off into a climate of their own. It’s very hard to stop this from happening. What we did was to make a model that only tried to get at the anomalies; the departures from mean climatology, and we specified the climatology from data. Now that actually isn’t as easy. It wasn’t then, and actually still isn’t, because there weren’t any climatologies of ocean currents. We had to do various tricks in a sense. So, for example, we got the climatological surface currents by using an ocean model driven by observed winds. But there weren’t satellite observations of winds then, so they depended on observations from ships, and there are a lot of issues with those. But anyway, despite all the reasons why this shouldn’t work, it worked. The model was able to simulate this phenomenon. And the important data we could use was to see that what we’re getting is something like what we’re supposed to get. We could also see that in certain ways it wasn’t like what we were supposed to get, which you know, always meant that there was something else to do.
Interviewer: What were the limitations of your raw data?
MARK: One of the things that makes this whole working on climate difficult is that the data that we have is so limited in time. I was trained studying modern meteorology, modern oceanography, and one of the things we work with are maps, global maps or maps of areas. You want a spatial picture of data. And if you do what I do, which involves ocean atmosphere interactions, then you need data over the ocean. There aren’t people with weather stations sitting around. The data mostly comes from ships. And there aren’t that many research ships, so most of the information, especially if we want to go back in time, has to come from merchant ships who routinely or are supposed to routinely, wherever they are in the world, take observations, at zero Greenwich time and 12 Greenwich time, of temperature, wind and surface pressure, air temperature, sea temperature and so on. That’s the main data we have. Now if you were working as we were in the early late 70s, early 80s on this, there weren’t any really solid compilations of this. There was one attempt to take all the data there ever had been or that was accessible, I should say, and make out of it the kind of average year, which is always dangerous because, as we all know, one year is not that much like the next and you don’t know whether you have enough of a sample to be representative. But that’s what we had. This was done by two people named Rasmussen and Carpenter. It was a very coarse resolution data set because they averaged data over big areas to be able to get something that would be at least somewhat reliable as an area average and so on. It was pretty insufficient.
Later on, this wonderful big project was completed by NOAA to make this thing called COADS, Comprehensive Ocean Atmosphere Data Set. What they did was to take all of the merchant ship observations available and clean them up and try to make them into something that was useable. That data started interestingly in 1854 when there was an international convention of people led by an American who was then the oceanographer of the Navy, Matthew Morey, who had one of the early theories of the Gulf Stream, by the way. It was agreed that everyone would collect this data, certain in this prescribed way and then it would be archived. But to do what the COADS people did, you have to take all these ship logs and paper archives and put them on computers and clean up all of the errors. And you record the data and they record the location where they made the observation, and you discover it’s somewhere over land because somebody made a mistake when they wrote it down. Then you have to try to figure it out and fix that. So it was a big project. So with that we were able to get data sets that went year by year, but when we started, in particular when we started trying to make forecasts, we needed wind data because we didn’t have any ocean data to speak of and we were going to reconstruct that using an ocean model and the winds. We needed wind data and essentially nobody believed that the wind data for the tropical Pacific and many other places in the world where there aren’t that many ships going, nobody believed that what we had before or around 1970 was adequate.
So we could only start doing this around 1970, which meant through 1985, we had 15 years. We were trying to learn about a phenomenon that came every 4 years, so we had 3 or 4 of them as examples. It would be as if you trying to learn to predict the weather and the weather changes every 4 or 5 days, so you would have something like 3 weeks of information and that would be it. You have to try to figure out how the weather worked from that. What happened over time was that we got more and more confident, you might say maybe over confident, that we could reconstruct these fields of past climate by using models and some sophisticated statistical techniques to fill in the big holes that were left by just the raw observations themselves. If you looked at where ships go in the tropical Pacific or the South Pacific you’d see there is almost no coverage. That’s a vast ocean. There are a couple of shipping lines that ships travel on. The rest of it is practically a blank. Be we developed ways of filling in these areas. And there were problems with it because the coverage is not very good. But the idea was to combine all the data we had about the atmosphere and the surface temperature of the ocean and a model that brought in the kind of dynamics that connects different parts of the world and using those dynamics as a sort of interpolator to tell you what was going on in the places we didn’t actually have data, and then getting a more complete picture. This has been a godsend for climate research. This is a product that is something like 10 years old and there must be thousands and thousands of paper using it and literally hundreds of discoveries that resulted from this.
Then we tried to go back a little further and that’s where we say all right, we have temperature observations very scanty from before 1950. And what can we say about that period from 1854 to 1950 and onward. That’s work, some of that I’ve done myself with colleagues here, in trying to fill in the sea surface temperatures fields from the very few observations that worked before. We did this using basically statistical techniques. People are now hoping to try to fill in even what the atmosphere was doing from the very limited surface observations. We would had almost nothing in the way of upper air observations before 1950. So to reconstruct the whole column of the atmosphere above is a real trick. And it will work to some extent. Then there’s the issue of how good it is and what you can use it for. But anyway, we went ahead and reconstructed this sea surface temperature patterns all the way back to the 1850s. This is work that’s principally associated with my colleague Alexi Kaplan. The neat thing is we’ve been able now to use those sea surface temperature reconstructions to do hindcasts or retrospective forecasts of El Niño going all the way back to about 1870. It turns out knowing what we know today, we could have predicted the terribly destructive 1877 El Niño up to two years in advance. It’s kind of a marvel that there is enough information and very few ship observations to be able to do that. And we’ve done something else. My colleague Richard Segar across the hall has led this effort. Using those same temperatures and then using them as the boundary condition for an atmospheric model, that model has been able to reproduce the major droughts over the last 150 years of American history, in particular the dustbowl drought of the 1930s as well as droughts during he Civil War and the 1870s and the 1890s and the 1950s.
All of these seem to be related to temperature anomalies, in the tropical Pacific particularly and to a lesser extent, other parts of the globe. These departures from normal are only a few tenths of a degree Celsius, less than half a degree Fahrenheit. It’s a marvel, really, that the techniques we used were able to get apparently correctly enough of these temperature patterns from the very limited amount of data. The ‘30s are a particular challenge because there was less shipping than before or after because it was the depression and there was less worldwide trade. By the way, the worst year in the record for data coverage for a long way back is 1918, because of the influenza pandemic. That’s one level of reconstruction.
Why do we do this? Well, we’re trying to learn how to predict El Niño better and understand it more. With only a few cases it’s very hard to distinguish what’s essential from what is incidental and just happens to have come along at that same time by coincidence. Also we now know that things vary a lot from decade to decade, so that the 1980s, late 70s and through the 1990s were a time of a lot of strong El Niño events. So were the late decades of the 19th century. The middle decades of the 20th century were periods with very few El Niño events. This is not just, we think, a matter of random luck. There were things about the conditions in those different times that made some more favorable, some less favorable. We would like to understand that better. It’s particularly pressing in a way, to understand that as we know we’re going into a period of global warming because of human activity.
We’d like to understand these temperature patterns, which had such an impact on a drought in North America and drought, in fact, in many, many parts of the world. You may remember the last one because we invaded Afghanistan at the time, but there was a big drought in Central Asia at that period, from 2001 into about 2004. That was related to these temperature patterns in the Pacific and so were a whole series of droughts at similar latitudes around the world. We’d like to understand what’s going to happen with this in the future as we change the climate. And we’d like to understand the past and so even going back to 1850 only gives us then a few examples of these kind of decadal long changes. We’d like to be able to go back even further. Pretty soon we’re pretty much out of luck with instrumental data. People simply didn’t make temperature measurements very much. There is a project in Europe to try and recover what can be recovered for the period from about 1750-1850, but it’s spotty and there wasn’t yet this international agreement on exactly what you should measure and when you should measure it and how you should measure it. So that’s going to be tricky too. But really, if we want to go back earlier than that, we must rely on paleoclimate indicators, proxies for temperature, for rainfall or moisture in the ground.
Interviewer: What are some of the proxies that you can use to reconstruct climate before there were any accurate historical measurements?
MARK: Probably the best known proxies are tree rings. I think it’s widely known that you can get some indication of temperature or of water availability by looking at the width of tree rings. The other thing that is important is the density of the wood in those tree rings. You really have to know what you’re doing to interpret this because there are other reasons that the widths might change—perhaps your tree was under a big tree and the big tree fell down and so all of a sudden it’s a lot sunnier for your tree. That’s not a climate change. There are changes in growth over time and so on. And so it takes some real understanding and experience to do this well. But we’ve gotten a lot of information out of those tree rings. Tree rings, of course, do not measure the key indicators of El Niño, which are temperatures in the tropical Pacific. They can be indirect proxies for that because they do measure in some places, like the southwestern United States and in other places in South America, they do measure rainfall or really more correctly, soil moisture, which would affect the tree growth. Soil moisture in those places is very strongly related to El Niño. But this is now a proxy for the climate in a place, which is related not perfectly to the tropical Pacific thing. We would like to understand more fully the relationship between these droughts and the El Niño phenomenon. To use the land based proxy for that is to assume that we know that relationship already and that means there is one less thing we can get out of that information than we would hope to.
What’s nice now is that in the last few decades another really powerful proxy for conditions in the tropical Pacific has come along, mainly corals. Corals also have annual sort of rings—not always as easy to read as tree rings, but often clean enough, and often you can find coral heads that go back quite a long ways, for hundreds of years. These are not analyzed by looking at width or density or anything quite as simple as that. You have to analyze the geochemistry, the chemistry of these different bits of coral, and in particular, the ratio of the heavy isotope of oxygen, oxygen 18, to the more usual oxygen 16, which is a measure of both temperature and of salinity and therefore of rainfall. That can be a problem in the sense that you can’t always tell whether it’s due to a temperature change or to a rainfall change. But fortunately for us for the El Niño story, they go in the correct sense. That is, warmer temperatures go with more rainfall both in the El Niño phenomenon and in the oxygen isotopes signal. So the oxygen isotope is a very good indicator of El Niño even if you might have more questions about whether it’s telling you temperature or rainfall. There are other kinds of geochemical measures that get a temperature independent of rainfall, in particular, the ratio of magnesium to calcium in these corals. All of those proxies come with baggage. It’s not like you had a thermometer, and even when you had a thermometer, there were problems. For example, ships measure temperature with thermometers, but the temperature might be the temperature of the seawater. People measure the temperature of the seawater with thermometers, but the thermometers might be where they take in the cool water to cool the engine so it’s already been heated up by the ship’s hull. Or they might have before that taken the temperature measure by throwing a bucket over the side and dropping a thermometer in it and waiting for it to come to the water temperature and then reading it off. And that sounds great. But you’ve taken this bucket, which for a long time was made of canvas and you put it up on the deck and the wind is blowing past it. So the water on the wet canvas is evaporating; it’s cooling the bucket. Again you might get the incorrect measurement. So even the instrumental data has its problems, but the proxy data is more difficult still. One can overdo the difficulties in speaking about these things.
Interviewer: Back to when you were in the midst of developing this model. First of all, on a personal level, what did you hope to accomplish? Were you thinking, “I want to help prevent drought,” or whatever?
MARK: All of the problems I work on may be interesting to me and a few other people. It would be hard to explain to most of the world, but we are really trying to understand something about the system because the climate system is important. And even when I do something small. I’d like to think that it’s going to contribute to something big. So here is this El Niño phenomenon, which I knew about because it had caused these anomalies in wintertime circulation over North American in ’77, ’78. I was living in Boston at the time and it dumped this incredible amount of snow on us. We were really pretty conscious that somehow these tropical Pacific patterns had been connected to this locking pattern, which locked us into something and we kept getting hit by snowstorms. And then we finally got hit by the big one. And it was a really interesting problem because then I learned more about it and I realized from reading that it was connected to all of these; the Indian monsoon to the climate in Northeast Brazil to the Southwestern US and so on. So this was something that had some global importance, but I was interested in understanding this really neat thing, which was this coupled phenomenon that involved changes in the ocean, changes in the atmosphere. There wasn’t really much we had about such things at the time. so I thought at that point, because of its impacts and because of the kind of physics involved, that this was the second best problem in my field. I thought the best problem was why are there ice ages, but I didn’t know how to solve that one. This one I felt I should have the tools. So that was the goal. And the thing was, can we understand how this works.
How do you know if you understand? That’s tricky. I remember at the time I was talking to my son who was then 11 and he said to me, “What you do is really boring. You work on the same problems for years and the best you can do is an approximate answer, and then you’re not even sure it’s right.” A colleague of mine (I put this in the front of some article I wrote) called up and said that was the best thing he’d heard coming out of any of us in a long time! Because it captures what we do so well.
The way I think about is if I understand something, then I should be able to turn it into a set of equations that I can put on a computer, and they’ll simulate the thing I understand. So I’ll be able to, in that sense, make a model of it. I’ll be able to take what I think and reproduce a version of that phenomenon. But my son is also right; even then it’s only approximate. And then when you explained what happened, maybe your explanation is wrong, because maybe you didn’t really understand what was going on in the model, or there’s still always the possibility you’re getting the right answer for the wrong reason. It’s always hard to rule that out 100%. That’s something people should understand about science.
The next thing that happened then was we finally got this model to the point where it would reproduce what seemed like the salient features of this phenomenon. It would get these warmings in the tropical Pacific, or in the model’s tropical Pacific about every four years. They would be centered on the equator and in the east in the proper way. The winds would weaken the way they were supposed to, or not exactly the way they were supposed to be honest about it, just as the temperature patterns weren’t exactly right. Although even now it would be hard to tell you what ‘exactly right’ means because every one of these events is different. But they were very close to the right thing. It would be hard to say whether they should have been even closer or not, although there are certain things we knew even at the time were not right. I’ve been doing this kind of science for 30 years and that’s still the highpoint. Not just because we solved something important, which you also don’t get to do very often, but also because there was a clear time when you could say this really worked. This model simulated the phenomenon and most people looking at would say yeah, that’s it. You got it. That doesn’t happen a lot. It’s more like what my son said. Then something interesting happened.
I was in a meeting in the fall of 1982 and it involved most of the people who were considered the experts on this subject at the time, and the prevailing sense of things was there is no El Niño going on and there won’t be an El Niño and so on. Now this was say, something like September, if I remember right. And we were already well into what would be the largest El Niño in 100 years. And I thought, what a great prediction problem to go after because if you’re going to do something. you want to say can I do better than the prevailing level? And if the prevailing level was we’re well into it and we don’t even know it, then we ought to be able to do better than that. I thought this would be a good problem to work on. And I did feel I had some understanding of the phenomenon. The other thing, which I think is important for me personally in this, was my advisor, who was one of the handful if not one of the two leading atmosphere scientists ever, had worked on weather prediction. A lot of my colleagues would feel that somehow a practical problem like prediction was a little bit beneath what we should be doing. We should be striving for enlightenment, understanding. But I had if you will, a kind of license to go ahead and do this. Anyway, I had it somewhere in my head that prediction would be a good thing to do. One day I was reading a paper describing a statistical idea of making predictions of El Niño. I didn’t really understand the method terribly well, but it did rely on the notion that you could predict what was going to happen by following the – – not the money, but the upper level warm water pool in the Pacific. And I thought huh. That idea, that’s good because that fits with our theory. That fits with our understanding, that these El Niño events are a kind of follow through, that the water is sloshing around in a very big basin governed by rules that are slightly different than the bathtub you hold in your hand, but nonetheless, sloshing around. And that fit with our ideas of how El Niño worked, so if we started with a state that gave a good description of where this layer of warm water was, where it was higher than normal, where it was lower than normal, then if we could just initialize a model with those conditions, start it off in that state and let it go forward simulating the world, then it might well work to simulate what as going to happen next in the real world. And that, in essence, is a prediction. That’s what weather forecasting models do. They start with a description of the state of the atmosphere now and use a model which has a very sophisticated version of equations of motion and so on for the atmosphere. And they go forward in time to predict the weather. We were looking to do the same thing for this coupled ocean atmosphere phenomenon where instead of looking to go a few days ahead like weather, we would be trying to go many months ahead. That would be more useful for climate and more appropriate.
The first problem was we didn’t actually have any data about what the state of the Pacific warm layer looked like. But we did have some wind data going back to about 1970 that we thought might be okay. And so we took the ocean part of the coupled model that we’d made to do the El Niño Southern Oscillation combination and instead of using the winds that came out of the atmosphere part of the model, we used the winds that came from this data to drive the ocean model. That would recreate the state of the upper layers of the Pacific. And so we did that from 1970 on. Then we said okay, using that as an initial state, we will try now to go forward in time. What we thought to do was to start in June of 1982 because we knew if we could forecast 3 months ahead or 4 months ahead, we would have succeeded in forecasting the state that existed that wasn’t even called correctly. We tried to do this and we did. We started from June and we went ahead and it worked. First thing I ever did that worked. Then we went back and we tried it from April. And it worked again. And then we noticed that there was a bug in the code, and we fixed it and it still worked; we were a little scared by that! And then we went back. We kept going back 3-4 months in time. We did January ’82 and October ’81 and all the way back to, actually I think, to October 1980. And we kept doing it. If I remember right, for April of ’81, all of those forecasts actually gave something like a very large El Niño event. In that sense they were correct. Now obviously they didn’t all predict correctly, every detail of the timing or everything else. Maybe it isn’t obvious, but it’s true anyway. But they did, almost all of them, give us this event. Then we went back; before we would want to forecast the future in any way, we’d want to get some better measure of whether this procedure we had evolved had any skill or not by going back over as much time as we could to kind of have some way of testing it. We did what was, if you want, a kind of simulated forecast or retrospective forecast, they’re called now, which is a funny sort of oxymoron. But it just means that you go back in time to where you know what the outcome is, but you still try to act as though as you didn’t know and make a real forecast. And of course it’s a model. It’s completely objective. It’s a completely totally written out set of rules that are in fact written out in a computer code.
Interviewer: How did you apply this model to do real-world forecasts?
MARK: So we did all the hind casts we could afford because in those days, even though it was a very simple model, that we can now easily run hundreds of thousands of years, this is 20 years ago and it would take us many hours to simulate a year. But we did a whole set of these forecasts, and by and large it worked, but it didn’t always work. There were years that are tough. Then we started forecasting ahead because hey, who could not want to know what’s going to happen next, right? So we started sometime, this would have been by the fall of 1985 we would have be doing that. And we went back and we kept getting these forecasts that were for an El Niño event at the end of 1986. As we got more information, as we went forward, we would get more wind information, which we could then use to move our initial condition for the state of the Pacific forward another month. We’d do it for another month and we would get another forecast, and the forecasts were very consistent. They were all saying there should be an El Niño coming up. We wrote a paper reporting the forecasting that we’d done and making a prediction. We sent it to Science and it was rejected. The reviewer said I don’t think this is right and if it is right, it’s too important to be published in this short form without a whole lot of scrutiny. Then we actually did a little more work, bolstered it up a little bit, and then we got the data from January 1986. Typically, in the past, we had a feeling, although we didn’t have enough cases to be statistically solid, that forecast using January data tended to be the better ones among the ones we could make for the following December or so. That forecast also was solidly El Niño. Then we thought okay, we’re forecasting an El Niño. Now what do we do? Do we sit on the information on the grounds that well, you don’t want to scare people? Remember this is four years after the 1982 El Niño had rearranged the shoreline in California, had washed out a lot of the transportation infrastructure in Peru and Ecuador and had done a tremendous amount of damage. There was a huge forest fire in Kalimantan (Borneo) that actually never stopped burning. There are coal seams in there, and it’s just kept going all this time. There was a more monstrous one in the 1997 El Niño. You couldn’t talk about El Niño in some part of the world without getting people really frightened. On the other hand, there are things you can do to prepare for it if you know it’s coming. So what did we do? We took this to someone who worked on earthquakes and earthquake prediction. He had been so taken with the idea that anybody in the earth sciences could predict anything that he really went to work. Anyway he brought in some people who looked over what we did and reviewed it and the conclusion was well, this is good science. So let’s do it.
With the institutional support we went out and made a public forecast. We also submitted a new paper to Nature — this time one that had more information than the old paper in fairness to Science. And we made a public forecast, which a lot of my colleagues were really unhappy about. A lot of them thought this was grandstanding, I guess. Basically they thought that scientists do science in their quiet corner, which would be nice if it were that way, actually. You know, you don’t go out there with this kind of thing.
But we’d done it and actually as an aside here, we learned something interesting from that exercise. We wrote, with some help from a professional, a press release. It turned out that if you do that, then the story gets reported much more accurately than if you don’t do that. Scientists often complain about the way the media covers them. One thing we can do is write press releases.
We made this forecast public in March and there actually had been a little bit of a warming from the early part of the year. So we thought okay good, it’s going away. Then it stopped, and turned around. And it started getting cold again in the Pacific, and this is, say, around June. We were not happy campers. We were pretty far out on a limb and it looked like we were going to have that limb sawed off by Mother Nature, because Mother Nature always has the last word in this game. And whatever you think you know, well it’s going to turn out you didn’t know as much as you thought you knew. So that was kind of a low point. I was pretty disheartened over the summer I have to say, and then sometime in August a good colleague and friend, Jim O’Brien, said don’t give up yet. Wait until you see the August numbers. And sure enough things started really turning around in a noticeable way and the world started warming up. And pretty soon it had caught up to our forecast. And indeed an El Niño event did develop in 1986, and as in our forecast, it was a moderate El Niño event. It wasn’t nearly as scary as the 1982 event, but it was an El Niño. It did have consequences. And we had captured it if not exactly correctly, we had captured the main idea of it.
Now the interesting thing is my personal lesson in how new ideas are adopted or go over. We made this forecast and that was successful, but that didn’t mean that the model or the ideas behind it were broadly accepted because well, they weren’t. We’re all trained as scientists to be highly skeptical, and my colleagues certainly lived up to that training as a rule. Some people were very supportive, very enthusiastic. Some people were just agnostic, you know, not hostile. When we made the forecast in the first place, one of my colleagues said, “Well this is great. Either you’ll be right and we’ll all look good, or you’ll be wrong and we’ll all have a good laugh at your expense.” So that was one way of looking at it.
But in 1990, early in 1990, some of the kind of signs or telltales that people thought of as precursors to an El Niño showed themselves and so a lot of people were calling for an El Niño at the end of 1990. But our model said no. It actually said that there will be an event, but it will come at the end of 1991, a year later than these people were thinking. And that’s what happened. We got a lot more credibility within this community that time out because people had followed this all from inception because it was the second time out, I guess, and because there were so many people paying close attention that time. But it was actually the forecast of the non-event that made this more widely accepted. The only other thing I really wanted to add is, you know, I was also terribly naïve. We thought that we’ll make these forecasts and they are things that impact people so people will pick them and use them. Not our responsibility. Not our interest. And that didn’t happen. And I got more interested in two things. First of all, this issue of how do you convey this information so people will actually make use of it, because I felt we’d been handed a kind of gift in the sense that I did something which was very abstract. And we did good work but it might have turned out that El Niño was not predictable because the nature of the system might have turned out to be that it simply was the kind of phenomenon that is chaotic or random or something like you know, like flipping a coin. And you can’t tell. You can’t predict it. You can understand the dynamics of how a coin goes through the air, but you still can’t predict whether it’s going to come up heads or tails. You can finally understand why you can’t predict whether it will come up head or tails. But you still can’t predict it and so, in that sense, it’s less useful than if you can predict it. So that was fortunate. And the fact that this had an impact made the work socially useful, that was fortunate as well.
So one issue was how do you get people to use this? What are the implications for agriculture, for water management, for tourism, for ecosystems. And, if you know those implications, what should somebody do about it? It’s not always obvious. The second thing was we were two academics in a small time operation making these predictions. To do it right you should have an operation more like a weather center, and so the notion began to form among some of us that there should be a kind of center that would put more resources into making the forecast, and then also put more effort into studying the issues that arose and then how to use these forecasts. And that institution actually now exists. It’s in the building next door, which you can maybe get some pictures of if it stops raining.
Interviewer: Tell us more about what the institution is.
MARK: That institution now exists and that institution is called the International Research Institute for Climate and Society [at Columbia University]. It has about 80 people and they work on predictions; they work on using all the information there is to make better predictions. They work on how to connect this information to agriculture and, in particular, how to make it useful to decision makers like small farmers and others wherever they may be. But as it happens, the the strongest influences of El Niño are in the tropics and that means primarily in developing countries. So most of their work is focused on developing countries, the poor countries of the world.
It does feel great. It’s one of the totally unexpected kinds of bonuses. I could add to this that in the 60’s and so on, I was a political activist. I was in the Civil Rights Movement, and when I got into oceanography I felt one of the nice things about this is that it’s very academic, totally useless, and gets me away from all of that contentious stuff. That turned out not to be true. First it turned out not to be true in a kind of milder sense that the El Niño work and what came out of it turns out to have some political, some social utility and that gets you into considering political and social systems again. And then of course, I studied climate. There is no doubt that humans have altered the climate and that we face, if we don’t act soon, we face some very severe consequences. Maybe I won’t, because I’m not that young, but my children will and my grandchildren surely will unless we do something quickly. It certainly wasn’t something I had any foreknowledge or any foresight about, but we’ve landed back into a situation where, alas, we have to do something to move the political system or the planet is in serious trouble.
Interviewer: As we end the interview, can you talk a bit about climate change and human impact on climate?
MARK: I am now concerned about the future; the future of El Niño is one of the concerns, but more generally humans have done things by burning so much fossil fuel to fill the atmosphere with CO2 and it’s inescapable. That alters the climate. There is no way around it. You can argue about exactly how much the climate will warm up given this, that, and that other thing, but the physics of this are very straightforward.
We’ve changed the climate. If we warm up the planet enough, and it doesn’t have to be much more if you look at the paleoclimate record, then we’re going to melt enough ice so that the coastal areas are going to be in trouble. And a good fraction of the world’s population lives in coastal areas. Hundred of millions of people in India, Bangladesh, China are going to be in trouble. The people like me who live on the east coast of the United States. The people on the west coast are going to be in trouble. Hurricane Katrina will seem like a small disaster compared to what we are letting ourselves in for if we don’t do anything, and do it soon. We have to do it soon because there is warming in the pipeline. There is a lot of gas in the air that isn’t coming out anytime soon. Even if we stopped adding to it now, there is more warming that will continue, more ice will melt, sea level will rise more. I used to think we had more time to get at this problem than I think now, and that concerns me a lot. I don’t know of too many examples in human history where people take a pretty substantial serious action decades in advance of when the problem will be. We just more typically wait until the problem is upon us before we take action. That could very well be much too late in this case.
The other thing I know is people don’t like to hear this message because it’s not a happy message. I don’t like to hear this message. Nobody likes to hear bad news. I mean this goes back to the Old Testament prophets. And this sounds a lot like the Old Testament prophets: “You people all did bad, you acted in a bad way toward your Mother Earth. You burned all that fossil fuel and you now are going to have to pay the price.” It’s not a message anybody wants to hear. I know that. So we practice denial: it will be okay. Technology will figure someway out of it. Why we think technology will come along and enable us to solve this problem, but we don’t think that technology will keep us from perhaps losing a little bit of our wealth while we give up our SUV’s and our other energy intensive or energy wasteful activities. Why we don’t think there won’t be a technological solution that will leave us better off, I don’t know. Why we think the technology is only able to solve the harder problem I don’t really understand except it’s not a rational thought.
3.1 Oceans Video
Ocean systems operate on a range of scales, from massive systems such as El Niño that affects weather across the globe to tiny photosynthetic organisms near the ocean surface that take in large amounts of carbon dioxide. This program looks at how ocean systems regulate themselves and thus help maintain the planet's habitability.
Unit 1 Many Planets, One Earth
Astronomers have discovered dozens of planets orbiting other stars, and space probes have explored many parts of our solar system, but so far scientists have only discovered one place in the universe where conditions are suitable for complex life forms: Earth. In this unit, examine the unique characteristics that make our planet habitable and learn how these conditions were created.
unit 2 Atmosphere
The atmosphere is what makes the Earth habitable. Heat-trapping gases allow ecosystems to flourish. While the NOAA Global Monitoring Project documents the fluctuations in greenhouse gases worldwide, MIT's Kerry Emanuel looks at the role of hurricanes in regulating global climate.
Unit 3 Oceans
Oceans cover three-quarters of the Earth's surface, but many parts of the deep oceans have yet to be explored. Learn about the large-scale ocean circulation patterns that help to regulate temperatures and weather patterns on land, and the microscopic marine organisms that form the base of marine food webs.
Unit 4 Ecosystems
Why are there so many living organisms on Earth, and so many different species? How do the characteristics of the nonliving environment, such as soil quality and water salinity, help determine which organisms thrive in particular areas? These questions are central to the study of ecosystems—communities of living organisms in particular places and the chemical and physical factors that influence them. Learn how scientists study ecosystems to predict how they may change over time and respond to human impacts.
Unit 5 Human Population Dynamics
What factors influence human population growth trends most strongly, and how does population growth or decline impact the environment? Does urbanization threaten our quality of life or offer a pathway to better living conditions? What are the social implications of an aging world population? Discover how demographers approach these questions through the study of human population dynamics.
Unit 6 Risk, Exposure, and Health
We are exposed to numerous chemicals every day from environmental sources such as air and water pollution, pesticides, cleaning products, and food additives. Some of these chemicals are threats to human health, but tracing exposures and determining what levels of risk they pose is a painstaking process. How do harmful substances enter the body, and how do they damage cells? Learn how dangers are assessed, what kind of regulations we use to reduce exposures, and how we manage associated human health risks.
Unit 7 Agriculture
Demographers project that Earth's population will peak during the 21st century at approximately ten billion people. But the amount of new cultivable land that can be brought under production is limited. In many nations, the need to feed a growing population is spurring an intensification of agriculture—finding ways to grow higher yields of food, fuel, and fiber from a given amount of land, water, and labor. This unit describes the physical and environmental factors that limit crop growth and discusses ways of minimizing agriculture's extensive environmental impacts.
unit 8 Water Resources
Earth's water resources, including rivers, lakes, oceans, and underground aquifers, are under stress in many regions. Humans need water for drinking, sanitation, agriculture, and industry; and contaminated water can spread illnesses and disease vectors, so clean water is both an environmental and a public health issue. In this unit, learn how water is distributed around the globe; how it cycles among the oceans, atmosphere, and land; and how human activities are affecting our finite supply of usable water.
unit 9 Biodiversity Decline
Living species on Earth may number anywhere from 5 million to 50 million or more. Although we have yet to identify and describe most of these life forms, we know that many are endangered today by development, pollution, over-harvesting, and other threats. Earth has experienced mass extinctions in the past due to natural causes, but the factors reducing biodiversity today increasingly stem from human activities. In this unit we see how scientists measure biodiversity, how it benefits our species, and what trends might cause Earth's next mass extinction.
unit 10 Energy Challenges
Global energy use increases by the day. Polluting the atmosphere with ever more carbon dioxide is not a viable solution for our future energy needs. Can new technologies such as carbon sequestration and ethanol production help provide the energy we need without pushing the concentrations of CO2 to dangerous levels?
Unit 11 Atmospheric Pollution
Many forms of atmospheric pollution affect human health and the environment at levels from local to global. These contaminants are emitted from diverse sources, and some of them react together to form new compounds in the air. Industrialized nations have made important progress toward controlling some pollutants in recent decades, but air quality is much worse in many developing countries, and global circulation patterns can transport some types of pollution rapidly around the world. In this unit, discover the basic chemistry of atmospheric pollution and learn which human activities have the greatest impacts on air quality.
Unit 12 Earth’s Changing Climate
Earth's climate is a sensitive system that is subject to dramatic shifts over varying time scales. Today human activities are altering the climate system by increasing concentrations of heat-trapping greenhouse gases in the atmosphere, which raises global temperatures. In this unit, examine the science behind global climate change and explore its potential impacts on natural ecosystems and human societies.
Unit 13 Looking Forward: Our Global Experiment
Emerging technologies offer potential solutions to environmental problems. Over the long-term, human ingenuity may ensure the survival not only of our own species but of the complex ecosystems that enhance the quality of human life. In this unit, examine the wide range of efforts now underway to mitigate the worst effects of man-made environmental change, looking toward those that will have a positive impact on the future of our habitable planet.