Notice: We are in the process of migrating Oral History Interview metadata to this new version of our website.
During this migration, the following fields associated with interviews may be incomplete: Institutions, Additional Persons, and Subjects. Our Browse Subjects feature is also affected by this migration.
Please contact [email protected] with any feedback.
This transcript may not be quoted, reproduced or redistributed in whole or in part by any means except with the written permission of the American Institute of Physics.
This transcript is based on a tape-recorded interview deposited at the Center for History of Physics of the American Institute of Physics. The AIP's interviews have generally been transcribed from tape, edited by the interviewer for clarity, and then further edited by the interviewee. If this interview is important to you, you should consult earlier versions of the transcript or listen to the original tape. For many interviews, the AIP retains substantial files with further information about the interviewee and the interview itself. Please contact us for information about accessing these materials.
Please bear in mind that: 1) This material is a transcript of the spoken word rather than a literary product; 2) An interview must be read with the awareness that different people's memories about an event will often differ, and that memories can change with time for many reasons including subsequent experiences, interactions with others, and one's feelings about an event. Disclaimer: This transcript was scanned from a typescript, introducing occasional spelling errors. The original typescript is available.
In footnotes or endnotes please cite AIP interviews like this:
Interview of Syukuro Manabe by Paul Edwards on 1998 March 14, Niels Bohr Library & Archives, American Institute of Physics, College Park, MD USA, www.aip.org/history-programs/niels-bohr-library/oral-histories/32158-1
For multiple citations, "AIP" is the preferred abbreviation for the location.
In this interview, Syukuro Manabe discusses: University of Tokyo; computer modeling; National Oceanic and Atmospheric Administration (NOAA); greenhouse effect; Geophysical Fluid Dynamics Laboratory (GFDL); Intergovernmental Panel on Climate Change (IPCC). Individuals discussed include: Joe Smagorinsky; Kikuro Miyakoda; Akio Arakawa; Jule Charney; John Von Neumann; L. F. Richardson; Douglas Lilly; Bob Strickler; Leith Holloway; Giichi Yamamoto; Sigmund Fritz; Wally Broecker; Sydney Levitus.
Before we start.
Mainly we are here to do an oral history of your career from the beginning to the end.
Or, I guess it's not over yet, so we can't go to the end. But I actually forgot to bring with me a printed copy of the permissions form. So I could print one later today.
Yeah, yeah. I am not going to say anything controversial anyway, so it doesn't make any difference anyway, so you send to me later to Japan or something.
Then it will be [???].
I'm just going to say that this interview will be transcribed and then we will deposit the transcript at the American Institute of Physics, which keeps an archive of things like this.
And other people will be able to use it. If you want to put any restrictions on how it's used, you can specify that. Otherwise we'll be [???].
No, I don't think so. The — I'm worried sometime you know the TV reporter interviews me and then they tell me that, they ask me the question and then they selectively choose the only small fraction of my answer and then put that along with everybody else and then they use each piece from different person to fit to the story which originally they would like to present.
So if they want to originally paint the Greenhouse Warming is a gloomy thing, then they can edit the thing in such a way to enhance their theme, the emphasis of their conclusion. And that what I don't like. That the only thing. But this is different kind, therefore I have no objection. You can put this in there.
Here you can say everything you want to say.
And, you know, my idea is to go for a couple hours now and then take a break and come back. My experience with doing these is that it's important to stop a little bit before we get too tired to go on, because it's very tiring to talk about this for a really long time. So do maybe four hours today, another four hours or so tomorrow, if we have that much time.
I'm from a very rural area. I was born in the countryside, right end of Shokusmore [?] Island. I was born there and I was the youngest son of medical doctor. My grandfather was a medical doctor also. So when I was young I always want to become a medical doctor. And sort of I thought this is sort of everybody and all my relatives and my brother is going to medical school, everybody going there, so I thought that what it is. But then I get gradually more and more interested in the mathematics or physical science, rather than biological science. And at that time I falsely get the impression that biological science is not the thinking science, okay, which is wrong obviously, and physical science because of mathematical elegance therefore it has more logic to it and more satisfying to do. And so as time goes on, even though I still keep going to the track to go to medical school [???], but I start getting despising biology. And because biology I can prepare day before exam. And so that's what I was doing. And then entrance examination I did very well in all this, in biology I memorized everything, physical science problem solving I practiced well, and fast I got in a 6-year continuum medical school. But then —
Uh-huh. Was this after the equivalent of high school?
High school, yeah. And but then I graduated early [?]. But anyway, the Japanese educational system changed, so following year I re-entered University of Tokyo.
What year is this?
That was 1949. So I went to University of Tokyo. And then, as time goes by, I begin to realize that biology I haven't studied that intensely and that I didn't develop any interest. So gradually after two years of the University of Tokyo I decided to go to Department of Physics. In it there are the three branches, one of theoretical physics and experimental physics and maybe, oh, there are four branches. There is geophysics and astrophysics. And I thought that I wasn't good enough, even I said I liked mathematics, not that great in math that good to become a theoretical physicist. In Japan there was a Nobel Prize winner, first Nobel Prize winners of physics came out at that time, a guy named Ukawa [?]. So many smart guys came to theoretical physics. I don't seem to be able to compete with them very well. And then experimental physics, I am not very good in doing anything using hand. Experiment doesn't seem to be good. And I think I always interested in natural phenomenon, so I want to go to geophysics. And then other people may get the data and make something out of this data, and so forth. But then once I got in the physics, I begin to realize that my original motivation I got in physics, that is mathematical areas, is not really a sense of physics. That is really mathematics is a very valuable tool to understand something, but I have to, in order to succeed in learning physics, I realized I have to understand physical logical reasoning to try to understand why that [???] mathematical result come about. And then I gradually realized oh my God, I despise biology. But biology probably a same thing. One used this mathematics than the others. But the basic thing is the understanding to try to dig in logically consistent manner to try to reach the hypothesis and then finally meet a satisfactory conclusion, try to understand the phenomenon. And that's what so much fun is about. So I begin to realize geophysics is a fascinating choice which I made by method of elimination I got into it, but once I got into it I found this is a fascinating thing, because now you have the phenomenon all around, and then you try to understand why. And I thought it may be even better than theoretical physics, because I can see so many things. Now imperial geophysics may not, but you know the atmosphere and ocean, they are the fascinating phenomena. So I suddenly realized hey, this is a great thing, and at the beginning my physics grade was a sort of miserable figure — straight C or D student — and but then I started seeing [?] you know I shouldn't have fooled with mathematics; I should have tried to understand why that mathematics had come about. And once I started doing it, my grades started inching up. And yeah. And so that's how I — so by the time I specialize finally you know, graduate from geophysics and went to master’s course in geophysics in University of Tokyo, I was getting pretty solid way of thinking about the issue. Don't take anything for granted, you know, digging in and asking dumb question, keep digging in. And this, applying this kind of inquiry to natural phenomena was great.
And then though, then I when I finish an undergraduate, I have no job in Japan at that time. Postwar period still.
Right. Speaking of that, before you go on with this, you were born in 1931.
Yeah, that's right.
So you were still young when the war was happening.
That's right. When I was fourth grade in elementary school war started, and the second year of middle school war ended.
Right. How did it affect you and your family?
It's probably I didn't grow as much as I should have grown, because it's undernourished all the time.
Yeah. And but I was, in Japan it was a sort of very extreme entrance examination system. And I was preparing for entrance examination even at the time of war.
Yeah. So I was studying very hard problems for solving a math problem and all that. I remember airplanes flying over you. I was, everybody in the, there was a foxhole there —
Yeah. Bomb shelters and —
Yeah, bomb shelter. I was studying in there. Fortunately the airplanes just pass over us, because we're in the countryside in middle school.
And so the war didn't bother me at all, and just keep on preparing for entrance examination. And that's why I was so successful. Once I want to go to school, I passed all the exams in any school. And but once I get into University of Tokyo, big adjustment was needed, because I wasn't doing right in studying the science topics, looking at, making it like applied math.
And so that, so anyway, when I graduated from university, there was no job for a meteorologist.
During the war there were a lot of meteorologists in Japan. They hired these people who are on the Air Force weather service, the people in the weather service in Japan, or naval weather casters, army weather cast, and all these people. Japan Meteorological Agency hired back, so there was over flooding of meteorologists in the government service.
Had you already specialized in meteorology when you were at the university?
University it's the study of geophysics. But so the geophysics didn't have much job either. Meteorology is a part of geophysics.
And so at graduate at a Bachelor's, B.A., when I got the B.A. obviously I had no job. When I get the Master's there's obviously no job. And when I got the Ph.D., after I finished Ph.D., there was no job. So there was a job but paid very little. I can't support my family. Arakawa or nobody could support a family. Very difficult. Oh, Arakawa was supporting family I guess, but it's not comfortable. And so that's how I decided to, when Joe, Joe Smagorinsky, asked me whether I was interested. Another guy is Miakoda one year ahead of him. And he and I, he named both of us. Joe was doing numerical weather prediction of rainfall, and we are doing it to rainfall manually. Joe was doing it at Princeton Advanced Study computers. We are doing it manually in Japan about the same time. And I remember I worked about 18 hours a day, several one of us, so we replacing computer by hand. And so Joe looked at that paper and then said you know one or two, one person out of it and either Miakoda or myself to come to the USA. So I took that opportunity. I thought this is great, because also in Japan, you asked a little about this question, in Japan there was a kind of a, there was a numerical weather prediction group, and people like Miakoda or Arakawa and then another friend of mine, Krehara, and all these people are very, very good at numerical modeling. And you can see you heard how good Arakawa was. I mean, doing all these magical things about finite differences. By then I wasn't that good applied mathematician. I was doing other things. And I liked more of the thinking about physical process. And so I'm not one of the best members of numerical weather prediction group in devising numerical system to compute hydrodynamical equation. Which Arakawa did so marvelously. And I started getting the feeling of sort of overwhelmed by the rest of the members. So when I came to U.S. I was sort of, kind of relieved, and but then when I come to the U.S. Joe Smagorinsky himself more is over applied mathematician than physicist. So he was, what he was doing at that time was he was, there was a famous Norman Phillips paper which [???] talked about. That is a real pioneering work. But it used geostrophic [?] approximations.
Okay. And Joe is the one who then tried to create general circulation model by using what they call primitive equation of motion, which don't use geostrophic approximations. By so doing, eventually you can extend your computation to [???] latitude, or you can do more of the heat of condensation [???], heat of condensation released. So that then when heat is getting so much, you can't use the hydrostatic approximations. So Joe was getting into, getting in this primitive equation model, which is sort of originally emphasized by [???] and all these people into GCM [?]. And he was running that, and that is still sort of [???] structure of [???] prescribed, and one and a half level or two levels. Very simple atmosphere structure, two temperature levels, or one temperature level and two wind levels or something like that. Very simple structure but put a primitive equation, which in a sense built the foundation for later putting global models including tropics or the heat of condensations and all these things. So that's what Joe was doing before I come.
Okay. Yeah, I want to hear a lot more about that, but let's go back to Japan.
Tell me something about Professor Seono [?] and his organization of that group. And then I'm also curious about how much you were aware of what was going in the United States, especially around computer modeling.
Very aware. Because this numerical weather prediction group sometime you know talk about this work all the time. And so every detail of it we knew what happening in the U.S.
But you had no computers, so you couldn't do the same thing.
No, there was some computer available.
Yeah. So you know I never forget we had done a spectral model calculation at that time, and a guy named Gambo [?].
Spell that please.
G-a-m-b-o. He came to [???] Chinese visiting scientist thing at Princeton, and he came back to Japan and then tried to sort of develop numerical weather prediction activity in Japan. This is how this group was formed as a matter of fact.
Oh really? He came to the U.S. when?
Oh, 1953 or 4 or something like that, somewhere around that time.
What's his first name?
The first name starts with a K. [???] name probably K. K. Gambo. And so he was the one who formed it. Now Professor Sheono is a very what you call [???] professor. So he leaves you alone. And if you can swim it's fine, if you drown it's fine. Also he never gives you physics topics in any specific detail. And he is also highly applied mathematical. Therefore he is not [???]. So it's very difficult to engage in the logical discussion of what your strategy for Ph.D. thesis are going to be. He just left you alone. So sometimes that —
That sounds like he and Arakawa must have had a lot in common, because that's my impression of Arakawa also, very mathematical and not very verbal.
Yeah, but he has a good, Arakawa has a much better physical insight than Seono. Arakawa is good in applied math, but he is one of these rare persons who has a good balance between physical insight and mathematics. Whereas Seono is very extremely strong in applied math but not much physical insight. That's my guess. I'm not sure, but I think. But because of his [???], he always able to get very good students around him, and we are sort of, because now I told you I was sort of [???] in the, originally in the applied mathematical area [???] fine. But then gradually I was switching to more of the physical understanding of what is important, not applied math. So some of our young generation dithered [?] against this professor. We didn't like the way he discuss the problem, because he started writing equations on the blackboard before he telling me what the hell that is. So some of our younger group — older half of numerical weather prediction group went along very well with Professor Seono, but we are very, very critical of him.
This would be you and who?
A few others. Yanayee at UCLA, you saw Yanayee at the — People like that. And Miakoda must be one of these groups, but he was in a transition. But then our groups are very critical. And that was one of the things that I was chosen, even Miakoda and me. I was sort of less desirable to stay in University of Tokyo probably, because I wasn't helping Professor Seono that much — rather than criticizing him all the time. So when the invitation comes for two of us, he said, "You go." [laughs] Yeah. Yeah. But I am grateful to Seono in the sense that he left me alone, okay? Let me alone, and felt like I could rely on anybody. I had to take care of myself. And so then deciding physics topics or something which was [???] to the Sea of Japan at that time. I brought every one of these new idea of thesis topics to, brought to school, and then among our graduate students we discussed the topics, and then they knock your idea down every time you bring a new one, and they keep doing it, and then occasionally come up with some feasible interesting topics. Then I start pushing. But then if it doesn't work out, then I have to readjust my topics continuously. And so we learned among ourselves much more than professor. And this is somewhat different from those who preceded me. And the more [???] applied math type like he himself is. So their communication was very good. And so that in a sense I'm kind of relieved to get out of Japan, because I'm not most favorite graduate student for him. And then numerical weather prediction group is overwhelming me in all these [???] differences and everything. And then when I — Yes.
This seems to have been a pure research group, right, because you would not have been able to do numerical weather predictions at a speed then that could actually be used in forecasting. Is that correct or not?
Yeah, pure research, yeah, of course. And you have to realize that the way originally Jule Charney [?] and Von Noyman [?], all these people developed this numerical weather prediction model, is not to put everything into the, everything, all the dynamical equations, mainly dynamic in full complexity in three dimension into computer, but instead they had very, very simple model. Originally it's two dimensional, what they call barotropic model.
And then they start from that, and then they went to two level model, but a tropic model doesn't predict storm cyclogenesist. They put in two level models, [???] model, which Norman Phillips developed, eventually become famous than other circulation one. And so they have a two level model, and then three level to monitor [?] [???] model. And gradually they were [???]. Still they have hydrostatic approximation, they have what they call casei [?] geostrophic approximations, and so that [???] simplification [?] of hydrodynamical equation was done. And therefore you could do a barotropic model forecasting even by using graphical handicap calculation method. And so you can do a lot of things. Or we had this barotropic model, two dimensional model spectrally, we are predicting spectrally. Gambo was doing that, and I was his assistant. And so in order to do spectral calculation you have to compute fully analysis, two dimensional field of variables, so [???]. And in order to do, and then predict how each component of double [???] series change with time, that's how you do the forecast. Now at that time the thing available in Japan was Fujitsu's machine. Fujitsu is now, it's an overwhelming along with any see Cray computer. At that time Fujitsu was providing, like American Electric or something, the instrument for telephone switch operating system, relay. And in order to get experience in the computer business, they fast start building relay computer. And what do they call, it's in the morning you go there, you blow it out, and ticka-ticka-ticka-ticka, a big sound, and that's how we first calculated spectral barotropic model calculation. And so forth. But then Japan Meteorological Agency shortly thereafter put the Signal One [?] IBM computer, thanks to [???] by Gambo and all these peoples.
Can you tell me around what year these relay computers were in use and when they were replaced by [???]?
Relay computer was done in 1954 or so, something like that. And wait a while, yeah, that's right. And Signal One computer went into meteorological agencies maybe in 1956 or something like that, '56 or 7, something like there about. And so the numerical weather prediction group's doing a lot of things by, calibrating by hand, which also turned me off. Because it's so labor intensive.
And you hardly have any time to think about any issues, which I was enjoying enormously when I was you know going up in the graduate school and then in this participation, in this thing there is a little bit it's competition of hard work. And the computer was available, but it wasn't that we don't have that good access. And even though we have access, it's very difficult because you try to use that kind of machine with 25 memory or something. Unthinkable, you know, and these machine.
Did you program these machines?
And so you must have been doing this in an assembly language of some sort?
That's right. Assembly language. Yeah. And you have only 25 memories. So how to do double —
Twenty-five registers memory?
Yeah. Yeah. Twenty-five numbers I mean. So you can store 25 numbers and then try to do fully analysis. And then you —
Are you talking about the Fujitsu machine?
The Fujitsu machine. And then you put into tape audio computer. So if you do one dimensional Fuji analysis, you put that into computer tapes result, put back any of that tape back in, okay, and then you calculate spectrum of the you know Y component or something. And so it's a pain in the neck. And so I never enjoyed using computer in Japan. Then when I come to USA '58, still USA also getting out of Signal One barely. And Signal One is fast to store the memory computer, which I think originally innovation of Von Noyman again. And of course it's binary computer, and but still when Joe was doing this before even I come here he was, I came here 1958 at Washington. Now Joe was, had very good programmers, had Watt [?] with him. And then I never forget it's everything in machine language, the program nothing like FORTRAN. You have to do machine language, and each number had to be scaled, scaling. And then —
Because it was a fixed point machine?
Yeah. Fixed point machine. So you had to make sure you know the magnitude of each number which you get. And then you have a map of memory. You put a memory map on the wall so you know exactly which memory you are using for which variable. And so you have a map memory, you have a scaling, and everything. And I think index register just coming at that time so you don't have to count the [???] or anything like that. So it was a very early stage and painstaking effort needed to run and you had to check everything, hand calculate everything, and have a [???] machine. I remember when I came here they have [???] machine and then hand calculate all the output and input and all that kind of thing. And so it was very painful. But that's how Joe was doing to laboratory when I arrived. It was a two level model.
Tell me just a little bit about how you came to be invited here? Because of the [???] paper? Had you met somewhere?
Yeah. He was doing weather prediction of rainfall.
And I told you we are also doing that as our group effort.
Yeah. I see that you have several publications on that.
Yeah. So Tokyo there was a numerical weather prediction group, subgroup of that numerical weather prediction group, University of Tokyo group. What we were doing was numerical weather prediction of rainfall. And so at that time we are using I think three level model, [???] model, and then tried to predict the rainfall. You add the condensation at each finite difference level and add them up [???] and then get the total amount of rainfall. So you move the air three [???], supersaturate, you figure out how much condenses, and then you add the condensation this way or this way or this way, add them up and then you get the rainfall. And you calculate everything they had. And so Joe looked at competitor of the publication I guess. Our paper came out probably Joe got it three months ahead of us, yeah, but then in the same year, and so he look at our publication and then say, "Hey, these guys looks very, do things alright." Why don't he invite, because why he invited [???] because there was a — Joe will talk about. You made that?
Yes, I did. I'm going to see him on Tuesday.
Yeah. That's good. So what it is is this. At that time in Princeton the John Von Noyman still, yeah, he was still alive and active, and he thought that this weather prediction model, the next thing is maybe using it for climate or maybe weather prediction you put more physical process in. Then you use it for extending range of weather forecasting. And so that what — the extending range of weather forecasting by using this faster computer, which probably may be available. And he thought that may be good to do it in a national weather service at that time called U.S. Weather Bureau. So you can get a more accurate account of this anyway from Joe. So, and then so Joe Smagorinsky in agreement with Von Noyman and [???], chief of weather service, and Harvey Wexler [?] and all these people agreed on having a group numerical weather prediction group, and that's how he was doing this two level model calculation before I get here. However, now around the time shortly before I arrived here they had this IBM is going to create super super-duper computer very soon. And then —
Was it the Stretch?
That's right. Stretch. So Joe was sort of dreaming of in a sense a system science dream at that time, latter half of 1950, building next generation model, not his two level model. But comprehensive model of climate and extended weather predictions. So he has that dreams there. And worked hard on it when I arrived in 1958. [???] came aboard around 1963 or so, and it was moved to, so our office moved from Sutton, Maryland to downtown.
Washington. And then IBM, I don't know exactly how they did; they may have sold that building to government. So we went to storefront office in IBM in downtown Washington, I remember 615 Pennsylvania Avenue N.W. we moved there. And then they put this Stretch in [?].
There, in your office?
Yeah. You know usually you put this, the downtown office there. Yeah, it's a very nice glass windows [???] IBM office at that time. And so that is about the time our laboratory changed name to Geophysical Fluid Dynamics Laboratory. Some people thought that is better than original name, which is called the General Circulation Research Section. They would change to Geophysical Fluid Dynamics Laboratory, sounded more academic. Bob White thought so. But so the name change wasn't that important. There was any question about it. But then the fact that around that time we started using Stretch machine.
Yes. I did not know that. I thought the Stretch machine was originally built for Los Alamos. Was it?
Yes. So one copy went there. So usually at that time Joe was so good at getting good machine for us. So we are getting Stretch was number 6 or so, manufactured by IBM.
I'm not sure I knew there had been more than one copy made anyway.
Yeah. And Los Alamos is even younger copies. Okay. And it was a marvelous machine. And then the computer program changed from machine language to symbolic assembly language they call it. They use more of the names in it. You don't have to put every memory location in there, and it's much easier to program, but not yet FORTRAN yet. [???].
Time out here. [laughs]
When you try to do this, our system modeling, which was Joe's dream, he is about 20-30 ahead of time, because our system modeling became fashion more recently again. But at that time Joe had almost in my view same kind of grand vision to build this thing there. Now there the interesting about it is that numerical weather predictions is more of the hydrodynamical system of equation based upon the evolution of weather by using hydrodynamical equation. However when you talk about this more climate models, more climate-like model, or more extended weather, extending weather forecasting, then other processes become very important. And one of the most important things is you are already working on the rainfall, how to put the rainfall. Because rain is evaporation, hydrologic cycle is a most important thing in the climate. Though more importantly I think that in order to build a climate system you have to have, incorporate explicitly, the radiative transfer. That is how solar radiation comes here. And then how that emit long wave into space. So that what we have to do. And so the thing we have to do is instead of until then we are assuming particle structure temperature, upper layer, lower layer, to find a different layer in troposphere, we have to get rid of all of these and let model determine these particle thermal structure of atmosphere by itself as a balance between radiative transfer dynamics and all these things, hydrologic cycle, all these things combined, and then as a result of it the atmosphere had to determine the summer structures. And then we also realize that we have to have the hydrologic cycle right. So in order to do that you have to have a heat balance and water balance around the surface, then the surface process. We have been talking about rainfall forecasting, but at that time rainfall forecasting which Joe was doing model [???] was blowing up, get out of the scale. You put the scaling in there, and so this, get infinities and blow up red lamps [?] coming all the time. So the rainfall prediction model wasn't working right. We had to put the rainfall right, and also when rainfall fell, then how the ground water balance. And then how that water balance is affected by heat balance, because water cannot evaporate unless you have a heat in the ground. So that we have to put, first of all we have to put the explicit radiative transfer, explicit condensation process which Joe already started but not working yet, and then we have to put the land surface process. So that I thought at that time Joe was taking care of all hydrodynamics. So I would like to focus on radiative transfer, I would like to focus on and make sure this atmospheric thermal structure would come out right when it's coupled with all these things. There is no guarantee you can get realistic atmosphere. And there is no guarantee [???] at the right place, does [?] that will form at the right place, tropical rain belt will form at the right place. So that how to put the explicit hydrologic cycle in. And so instead of only one or two levels, we put, Joe decided to put about nine particle levels in the atmosphere so that we can determine stratosphere, tropospheric troposphere, all the similar structure of the atmosphere. Joe created nine levels, but I had to take care of radiative transfer which works well. Because others groups like Arakawa and Mintz are not doing that yet. They are sort of giving the same way as Joe's first model, have two levels of temperatures or something. You give these temperatures. And so that the differences, the kind of physical process we put in which is sort of continuing until now, you know everybody have to do it now. And so that's —
So that's what, yeah, Joe felt like he had to hire somebody in addition to the people he had already assembled. So he just like you know National Hockey League finds some Russian hockey players, [laughs], there was numerical weather prediction conference. Okay. Okay. In the, in Tokyo shortly after I came here, yeah that is not original motivation. Yeah, forget about numerical. So anyway, so that's why Joe decided to hire me.
So he was already working on this 9-level, on his idea of expanding to a 9-level model.
Expanding. Started collaborating with IBM people. And so he had a thermodynamical [???]. But when I arrive here, I notice that his plans still he have to care of what he do radiative transfer, how to solve the problem of the model blowing up because of the hydrologic cycle rainfall computations. And so, and then I found out that the dynamical model itself is some problem. And it wasn't, he has this thing proposed by Von Noyman, nonlinear viscosity, that there was a problem of the when you run the hydrodynamical equation you got this, it's not a ordinary linear computation of instability, but nonlinear instability which was analyzed by Norman Phillips again. He did many outstanding work around that time. And —
Was he affiliated with your group? Did you —?
No, no, no. He was working with Jule Charney in the Von Noyman assembled in the Institute for Advanced Study. The group headed by Jule Charney. And that group is one; in that group Norman Phillips was there. And that was numerical weather prediction group in Princeton, New Jersey, and about the time I came, shortly, a year or so before I arrived at the United States Von Noyman died by I think stomach cancer or something. That I am not sure exactly what the name, the kind. He died with cancer. Okay. And so then that group decided to go to MIT. Yeah.
Oh, okay, alright, I didn't realize this was [???].
Yeah. He went to MIT. And around that time when Joe's group started in Washington, D.C. and Joe [???].
Joe was part of this Princeton group at first and then moved to Washington?
Right. But the rest of them went to MIT.
Yeah. That is correct.
Joe was rather a junior member of Princeton group, but Norman Phillips probably is the key member. They are about same age, but Norman, because he developed 2-level volcanic [?] model. And I'm sure Joe already explained to you about that. And so what happened is around that time even before I come here, Joe started putting his grand vision in developing going from 2-level to 9-level model. And following Joe's, this 2-level model, [???] Mintz want to build another 2-level model in collaboration with Arakawa. So that that 2-level model effort is parallel with our 9-level model effort.
And Joe's 2-level model effort preceded both of these.
I'm not sure I'm right about this, but I think Arakawa did not come to the U.S. until 1961.
So you were here three years before that.
And during that whole period when you were here you were working on this?
That's right, yeah.
Expansion of the number of layers.
Radiative transfer [???] and then the land surface process. I was working at that, I was going to the Library of Congress, and the thing is, the thing however is for example this [???], maybe we talk about this later more, but the way I do is usually I went, go to find out what's available about that. And so I studied, I went to Library of Congress almost every day, you know, every weekend I went there at least, and tried to find out what we know about land surface processing. And so then I put various things in, created the land surface process model, heat conduction through porous material, how water goes through, and I even studied a phenomenon like permafrost [?] when temperature is cold, and started reading and reading and reading, and eventually sort of create if I know everything, if I have an infinitely fast computer how I will do it.
And then I say but we don't know the conductivity of soil. If you look just outside, you know how soil varies from place to place, attachment [?], the different attachment. So between freeze and one grid box there was so many things in there. It's like fractal. I just can't, no matter how complicated model I made, I can't feed the information needed for these models. And so at that time then they say, I say, okay, is this complexity warranted in view of our ignorance. So then I went to second stage, systematically chopping off, chopping off, chopping off, and then finally I came to this thing called a bucket model. Bucket. Now the moisture holding capacity of soil varied from one place to another, and but if I reduce this moisture holding capacity in desert I would not almost desert sandy soil doesn't hold water, so that almost I am creating desert by assuming sandy soil in desert doesn't hold moisture. So I decided to keep moisture holding capacity of the soil constant all over the world and made a very, very simple model, taking some idea from [???] scientists and what we call bucket model which [???] varies heavily now. And but bucket model can be parameterized if they wanted to. And but they criticized my original version, my original choice of parameters and everything. [laughs] So then some target to criticize me in order to try to improve, and I became a target, so I'm kind of happy to be the target. And so that's what it is, and the radiative transfer one is the one I did it for the trying to see whether this kind of model will yield right [???] structure when that is put into three dimensional general circulation model. But to try to test in the framework of three dimensional GCM is too clumsy. With that, at that time you know very slow supercomputer which far slower than this one. Okay? You know? And so you just can't do it. And so that's how I get into fast radiative equilibrium, pure radiative equilibrium, and gradually radiative conductive equilibrium. Until middle 1960 I was doing one thing after another. And what then I realized at that time so this is a way I always construct a model. I go through this thing and then systematically simplify again. Or before putting in the full scale model, I create a separate prototype, much, much simpler model, and test its performance in that simple framework so that I can afford to make many stupid mistakes — which everybody makes. And nowadays modelers don't understand this, particularly sometimes there is a committee decides what you should do in the model, an advisory committee or something.
Or something. [???] was advisory committee on some of this. And so then committees, in order to get everybody to use this model, what [???] do you need, what process you have to [???]. Not, they are not thinking in terms of how to optimize the development models. They are talking about end product. And then tell them this is a product I need, this resolution, the parameterization, this, and this, and this. Without much regard to how actually you try to test the model and make it work. And so I actually, instead of doing that, I don't have advisory committee, Joe and I. So, and Joe was traveling. His traveling schedule got busier and busier, so I have to troubleshoot.
Why? Why did that happen? Why did he start traveling so much? I've read this in other places too.
Because he got involved in the Grover Atmospheric Research Program, he's a member of Joint Organizing Committee, and he is very effective as a committee person. He has excellent vision, an ideal person for committee, and also once he decides some goal he goes towards it rather than going into tangent. So he is an ideal person as a member or chairman of committee of anything. That's why. So I have to troubleshoot. So, Joe was great in grand vision, but grand vision, if you put that model in, in the full complexity of his grand vision, just like L.F. Richardson's model, it fails. And that's why this institute model, they start from simple barotropic model 2-layer, and that's how their numerical weather prediction model succeeded. But L.F. Richardson's model failed because he has also grand vision which is even bigger than Joe's grand vision, which is again 20 years earlier. Okay? So L.F. Richardson's vision, Joe's vision, and then now it's our system science vision. They are essentially the same kind of vision, if you look at L.F. Richardson. But when you create a model which works, you don't do it that way. And so you develop this radiative conductive equilibrium model. And then I gradually realized that what is most important in this process. The most important one to determine the thermal structure of atmosphere is sun. Solar energy. Without solar energy we would be near zero degree [???], right?
And then we wouldn't be here.
Yeah. So, right? But then I quickly realized what the atmosphere is made of. Atmosphere is made of nitrogen, which is about 80 percent or so, 20 percent oxygen, and then there is a minute substance like carbon dioxide, water vapor is [???], fresh table [?] water in there, about 2 cm of fresh table water. Water vapor, ozone, and so forth. And realize that if there were, this atmosphere was composed only of oxygen and nitrogen, the temperature at the ground would be something like 255 degree Kelvin. Okay. But the temperature, Grover Mean Temperature, which we live in, is something like 285 or something like that. So there was a 30 degree difference between the temperatures of surface temperature of this planet without any Greenhouse gases. That is nitrogen and oxygen alone, 255. And then you put this Greenhouse gasses and that was 285. So there is a 30 degree difference between it. And then this is Greenhouse gas, Greenhouse Effect.
Right. Now, let's back up just a little. Tell me something first about the working conditions at what with the U.S. Weather Bureau I guess when you first came here.
Because you got this Stretch computer, but that was 1963, and then before that you were working on a 701. Is that right?
701, 704, very quickly 704, and so forth.
Yes. Did you do a lot of the programming for the [???]?
No, I came in there, and at the beginning Joe gave me one programmer who didn't have a college degree, and Joe hired another very promising young scientist at that time named Douglas Lilly.
Yeah. Lilly. And he is excellent applied mathematician, but he is also excellent physical insight. I mean Joe hired this young man about the same time he hired me. And so he assigned one programmer who can't program to two of us. And so at the beginning there was a struggle, but as Joe got busy I quickly inherited just about the entire his programming staff, which is close to ten people. I suddenly, suddenly in charge. I don't have any title, but all of a sudden I see all these people sitting on my lap, and here is some Japanese, all these scared [?] Japanese guys who don't know how to communicate well with the English or who don't know how to read the people's expression on their face. Suddenly I [???] with — also they are all Joe's group. But because, you know, I got sort of I become a very competent lieutenant to him, right? So I end up you know getting all these group to work with me very soon. And so, and then suddenly you know you have to also realize 1958 is the year when Russian Sputnik went into the air.
And then '59 or so, or '60 or so, John F. Kennedy elected President of the United States.
Yeah. And at that time everybody said there is a missile gap. And so they talk about so how to catch up to Russia, because they have these dogs called the [???] on the — and going around the earth. Poor things, you know, come to think of it, because these dogs they are going to die, but it certainly helps inaugurate sort of the golden age of U.S. science, particularly physical science. Now I guess next century they say it's a century of biological science. But certainly at that time they inaugurated marvelous age of physical science, and we are no exceptions. So we are getting very sufficient funding, including purchase of this Stretch, which is a very expensive machine. And so all of a sudden [???] we hear the Ph.D. students in Japan struggling, looking for his identities, and suffering from inferiority complex from other very good people. Like Arakawa, he's probably one of the best in the world at that time, although he didn't realize that in finite difference things. So I got suffering from inferiority complex, and then I couldn't get any job, to suddenly getting to something, here is a grand vision of Joe's, and then a huge computer which is a supercomputer, and suddenly I headed something like ten people working for me.
Right, right. So around when did this happen when you —?
In early 1960. '60. Okay? So 1960 is still not everybody working for me. Around '65 almost everybody working for me. Well, toward the end of '60 more and more so. So these ten people graduated and sort of become completely Joe's control to slowly shifting under my control. And the middle of '60 Joe really getting busier and busier.
Now, at the, you know, this is the general circulation section of the U.S. Weather Bureau that you are working on [???].
By '63 already we are downtown, the geophysical fluid dynamics laboratory, and so —
And so that by then you are separate, pure research doing this [???].
Yeah. Originally. Then [???] circulation research section was, Joe originally came to numerical weather prediction unit, joined numerical weather prediction unit of the National Weather Service, and then gradually separated it into the [???] circulation research sections. And so that's what, yeah.
The joint numerical weather prediction group was U.S. Weather Bureau, Army, Navy Weather Service and Air Force.
Did the military agencies still play a part in the U.S. Weather Bureau group when you were there? Or did you just not hear about them much?
I am not sure. Joe can answer better about that, because that is before me really. I just joined, when I joined already general circulation section, Joe probably couldn't stand it much longer in that numerical weather prediction unit with his individualism. [laughs] And so, but fortunately, original plan of Joe moving in there was to create this kind of modeling group under the agreement between [???] and John Von Noyman and so forth, which Joe is going to explain more accurately. And so that's, yeah. So, and then general circulation research section changed name to geophysical fluid dynamics laboratory, because I think Bob White, who was by then is an administrator, what at that time was called ESSA. It's almost like a gasoline company but it's not. Environmental Science Service Administration.
And then that ESSA quickly changed into NOAA to combine various agencies and Bob White was a head of that, administrator. And he recommended that name. I remember I registered it personally. I didn't write it myself, the change of name, but Joe went along and changed name there. And but around that time is a time when we really started using Stretch. So in that sense it's a very meaningful time using this Stretch. At the beginning we start putting this 9-level model in, and then I remember model blew up because of the, more of the computation of the hydrodynamical equation. His finite differencing of nonlinear viscosity which was put in order to prevent this nonlinear computational instability, which was originally a suggestion of Von Noyman, which I start talking earlier. It wasn't working properly, so model was blowing up, and so that I don't know, at that time it's $8,000. The other day Joe said it's closer more $8,000 a day machine was sitting idle, and machine isn't working, functioning. And I felt terrible about it. And what I tried to do at that time, okay, let's wait for Joe's grand visions and let's simplify it so that I can manage this model. So that my role become now is to toss off any complexity and make model work at least for the time being and put back in later if we wanted to. And at that time UCLA, the 2-level model started working very nicely. And now maybe they are also developing at that time. But [???] Mintz I remember shortly thereafter came around and say, "Why we need GFDL [?] model? Here is a beautiful thing there." Yeah, I remember that. And so first of course we just improved then finite debalancing of nonlinear viscosity, and made it to work for the time being. So then it started running, but at the beginning, I never forget that it was a short period, a very difficult time period. And so we have to run Joe's original 2-level model back into it, and then running and carrying out various numerical experiments rather than this dream model to put on this Stretch for the time being. And so then we bought time, and then tried to, I tried to simplify. I don't think that Joe knows that very well, because I just single-handedly tossed off some of complexity every time he was traveling.
[laughs] So he would leave, and you would take equations out of his models.
[laughs] Yeah. I was simplifying it. Particularly convection parameterization. I put a convective adjustment in, and as Kasahara or somebody was explaining at Arakawa's symposium, the convective adjustment is one of the very effective methods of eliminating this convective instability of first kind. And so that was [???] its causeways or [???]. I was doing it by convective adjustment. And I think it's a very, very successful model, the parameterization, although many people don't like it. I can explain the reason for it. For certain usage it's not appropriate; for other usage it's a marvelous model — which people don't realize that. So I put that in, that was on Joe's more complicated version, and then make it work. And so that's taken care of. And then radiative transfer, I tested [???] by using one dimensional radiative convective model.
Right. Let's talk about that for a little bit.
Because around 1961 or so you started working at Fritz Mueller [?], right?
How did that come to be? Because I noticed that the first publication of the two of you is in German, and is he, was he based in Germany at that time?
No, no. Joe — I have another paper which is, [???] also had it too at that time. But anyway, what happened is I was struggling with radiative transfer, development of radiative transfer scheme. Basically though what I tried to do is to combat existing radiative transfer computation scheme which was developed for graphical calculations into computer algorithm. And then use that in testing radiative transfer calculation, and at that time radiative equilibrium computations. When Fritz Mueller arrived, Fritz Mueller looked at my scheme and he said — he's a pioneer in that anyway — he said this isn't, a certain approximation I made is not good enough. So he was very helpful like that. And then he looked at the —
Why did he come? What was his —?
Joe invited him. Because he realized radiative transfers [???] very important. But Fritz Mueller originally thought of doing it sort of highly parameterized radiative transfer scheme [???] explicitly calculating it. So you know, it's some kind of linear [???] like simple linear equations express radiative transfer into that form, and then that is a physically based way. So he transferred physically based scheme into linear regression like equations. And he wanted to do it that way. But then as soon as he saw what I was doing he thought, "Hey, this is a great idea. Why not do it a fundamental way?" So then he and I started collaborating, and he immediately find out the shortcoming of my scheme. And then that German paper is the one which he and I were struggling with [???]. I worked there and he quickly wrote a paper. One weekend he wrote that paper, like this. I have very little to do with it, the writing of that paper. And but anyway, so that collaboration was great. I owe him greatly because to straighten out some of my calculations and so forth. But even after I left there, I continued and then I started putting, in order to test it radiative equilibrium became highly unrealistic. What is very important is a convection between, convective heat exchange between earth's surface and the atmosphere. And so that what I come up with, and so once I get radiative convective equilibrium, I can test this radiative scheme's ability to simulate similar structure of atmosphere very effectively. And that paper, done in 1964 with Strickler.
Right. Tell me about Strickler and what his role is in this.
He did two things. One is he helped me in programming that radiation cord, particularly when you put a cloud cover in there. The radiation cord algorithm becomes very complicated. And he was very, very good at that. And but then this [???] or consequence of forcing him to do this complicated programming computer algorithm, he said he'd had enough.
Yeah. And so he actually originally worked with Ryerson [?] developing aria primitive equation model when he was at UCLA before he came, joined us. And so he rather wants to go back to hydrodynamical equations. So he moved to another branch of our laboratory, which is experimental prediction branch, after this. But he did marvelous job. He did two things. One is he developed radiative convective equilibrium program, and then he also did help write a program for [???] convective adjustment. And so that his programming was very reliable and it really saved me at that time. I could sleep. Because it was much, much harder to check computer program at that time. And it was extremely difficult, and it's so much easier also to make a mistake at that time. And so that was Strickler's job. And —
So your relationship with these professional programmers who worked there, they would write the code, but then you would have to debug?
What I, the skill I gained at that time is when I look at — The radiative convective equilibrium was a very good thing to do, because it's essentially linear. So that whatever change you make or whatever Greenhouse gas you put in, you are able to think physically what effect that should have, and then you test it by the model. So that these radiation transfer algorithm, you can, based upon physical reasoning, if you got the result which shouldn't be that way then you can spot it. And I was sort of, I told you earlier I was getting better and better at sort of thinking, logically reasoning based upon physical reasoning, and I was doing that in the physics department very well toward the end. And so when I started doing the radiative transfer, even though I was trained as a dynamic meteorologist I was learning a new trade, but I quickly saw smoke out of any unreasonable result based upon physical reasoning. And that's how I was able to identify many of the programs. But originally I am sort of more or less more of the honorable person rather than visitor mind [?], and therefore I have a terrible time adjusting to you make some mistake, you know, one person Joe originally assigned, he make, when he makes one change he makes three more mistakes.
Yes. That's the problem with programming.
Yeah. And so I got a nervous breakdown almost, very close to a nervous breakdown. But once I get this Bob Strickler, working with Bob about radiation and hydrologic cycle. And another person, Leith Holloway.
Leith. L-e-i-t-h. Same as Chuck Leith's last name, but this time it's first name. Leith Halloway. And he is a typical digital person. He check everything. Then he check that you checked, and checking and checking and checking and checking, and double checking, triple checking, opposite of [???] there. So I got these two very good checking persons, and by the time I moved to downtown location for Washington for Stretch, shortly thereafter my depression and nervous breakdown gradually cured.
And because they such a good programming. And then so the original motivation of studying Greenhouse Effect has very little to do with my concern over environmental problem, but the Greenhouse gas, as I explained some few minutes earlier, is that Greenhouse gas is the second most important factor for climate next to the sun. Sunshine of course number one importance, but Greenhouse gas on this planet changes surface of earth by as much as 30 degrees Centigrade. So that in order to test the radiative computation algorithm, how effective it is to simulating the thermal structure of atmosphere, you have to put this Greenhouse gas in and then try to evaluate what is the effect of that on climate. So —
Yeah. Now, speaking of this, when did you first become aware of the work of Arenius [?] and —?
Arenius I didn't know for a long time. However following Arenius there is a person like Calander [?].
This is UK engineers, and Calander, and then next there is a guy named Plass [?], and then —
These are in the '50s and they [???].
Fifties. [???]. If you look at my debut which I sent you, you can see all these developments there. And so these people there. But you know —
Did you know about their work when you started working on this?
Was this the stuff that you found?
Actually the person who introduced me to this work is Professor Yamamoto, same as this [???] same name but it's unrelated, because Japan has so many Yamamotos there.
The same as whom?
Yamamoto, [???] Yamamoto, who was the Admiral of entire Japanese fleet, [???] Yamamoto there. But unrelated. Anyway, so Professor Yamamoto is a very, very, probably the person I respected the most among Japanese scientists, and he happened to come to Weather Bureau in order to advise Sigmund Fritz at that time, Sigmund Fritz who was inaugurating weather satellite laboratory.
Okay. Spell that name for me if you would.
And the first name is?
Sigmund. Sigmund Freud.
Sigmund Fritz. And he also advised us about solar radiation algorithms at that time, because he belonged, a short while he belonged to general circulation research section along with Fritz Mueller who came out with visiting scientists, similar visiting scientist. Fritz Mueller actually was, uh, Sig Fritz was actually part of our general circulation research section, but —
Okay. When did he come and how long had he been there —?
He was just a short while, because the head of Weather Bureau research at that time, a guy named Harry Wexler, and he at that time had a vision of sending weather satellite. And Sig Fritz is a person who makes this happen. And Sig Fritz and I happened to be in the same room, shared room there. So —
And what period was this when he was part of your group and then —?
1958 to '59 to '60 or something like that.
And let's see, why I started talking about Sig Fritz?
Right, Yamamoto, yeah. So anyway, so then come back to Yamamoto, Sig Fritz also there advised my radiative transfer, but Yamamoto is the one who introduced the importance of carbon dioxide and so forth. And so I get, and he also enticed my interest in radiative equilibrium too. And so even when we are in Japan through his textbook he wrote. And so I was —
So you already knew about this from him.
Yeah. In Japan.
In the mid-'50s in Japan.
Yeah, yeah. And also I was interested in radiative convective equilibrium.
But then my job requirement to try to develop radiation algorithm come very handy to do this thing there. So I again opportunist. I took this opportunity to do this problem. But then a funny thing happened. Fritz Mueller, who was working with me, wrote a very interesting paper around 1963. And so I think he worked on it even before he came USA. Or maybe around that time. He found out that all these methods like Callander, Plass and everything, all these methods start failing when you start putting, when temperature warm up in the atmosphere. Okay. When temperature warm up on the atmosphere and then more water vapor in the atmosphere. Okay. Because it [???] saturation vapor pressure [???], more water vapor. And then you, that means more downward flux of radiations. So when Fritz Mueller started putting what we call water vapor feedback, these methods which was developed before started failing miserably. You started getting any answer. Sometimes when you double CO2 you get a cooling of 10 degrees, depending upon temperature, and another temperature you double CO2, you get 15 degree warming. Starts getting all kinds of crazy results. Mainly because including Arenius, all the pioneers' work of the Greenhouse Warming think about only the [???] dation of radiation balance of earth surface. They never think about perturbation. If you increase the radiative gases in the atmosphere, how the heat exchange by boundary [???] between atmosphere and the surface changes. Which is equally or even more important component. But almost all of pioneers just discuss radiative balance perturbation to earth's surface without considering. And so that that's why they are failing. And the radiative convective equilibrium give a perfect means of putting that convective equilibrium. So Fritz Mueller say, "Oh, why don't you put your radiative convective equilibrium which you developed after I left from —
By then he had visited as a visitor. He came back to GFDM. And then we have a discussion.
Wait. Run that by me again. He went back to Germany?
Went back to Germany, as a visitor, just a temporary visitor, and he said we have a nice chat together because he come back you know, so I say glad to see you again, and we have a nice discussion about it, and then we mentioned about these recent papers about failure of this radiation perturbation approach at the earth's surface and how to solve this problem.
And this is what year?
Has to be '64 or something, 5, maybe '64 or 5, somewhere around that time. His paper was published '63. So it must be somewhere thereabout. So he came back and we have this discussion, '64 I bet. So and then he said he listened to my new radiative convective equilibrium rather than radiative equilibrium which I was doing with him. So look at the radiative convective work approach. Ah, this is a perfect tool to resolve this problem as a result to our discussion. And he very much encouraged me to do this problem. And I was more than anxious of course to do that. And that's how the papers in 1967 with Wetherold [?] came out. That is a 1967 paper. And to resolve this tradition of radiative perturbation approach of earth's surface to include the perturbation in the convective heat exchange between the atmosphere and the earth's surface. But in order to do that, you can't just discuss heat balance of earth's surface, you have to discuss heat balance of atmosphere and exchange between atmosphere and space. And so that's what sort of we are doing at that time. So that's how I get to this 1967 paper. Now around that time —
Before you go on with this, to do things like this you need to know a lot about the existing, you know, you need some data about what the thermal structure of the atmosphere is actually like, and one of the things you mention in these papers in this period around '63, '64 is that there are new data from high altitudes that have become available. And one of the things I am curious about throughout all of the things we have talked about so far is you know your interest has been mostly theoretical all along.
But to what degree have you been learning about observations of the atmosphere during this period and what kinds of things were important.
Now, I don't know exactly which thing you saw in my paper, but you know by then of course we know the summer [?] structure of atmosphere enough, because of radio sound and all these things. Now there is one thing which wasn't too clear was water vapor concentration in the stratosphere. The people send, the troposphere and the lower stratosphere people know pretty much. Oh, even though stratosphere water vapor we may not know enough. Temperature we know quite high altitude, and middle stratosphere we didn't know, and there was all kinds of results presented due probably to the contamination. If you send balloon or something the evaporation from balloon you may be measuring water vapor which comes out of balloon and all that kind of thing. There was all [???] instruments, frost point hydrometers and — And so what saved us was this measurement by Mastenbrook.
Yeah. Mastenbrook. He's a naval lab research lab scientist. You send the frost point hydrometer, he put the balloon there, and then he have a string of about few hundred meter long strings and put the frost point hydrometer, and it went up and down, and then checks the accuracy. And so we are able to put a nice looking water vapor in the stratosphere. Mastenbrook has done marvelous. And then of course UK, the Royal Air Force, a guy named Goldsmith [?] send a lot of Cambra [?] aircraft.
Okay. Spell that please?
Goldsmith. G-o, yeah. And he measured lowered stratospheric water vapor beautifully. And so these things really helped the time in what kind of water vapor profile to put in in the upper atmosphere. Troposphere first we put to observe the absolute humidity, but later in order to put the water vapor feedback in this model I assume relative humidity assumptions.
I think I understand this, but let me [???] to be sure.
This is because if you are going to deal with changes in temperature, the water capacity of the air grows at the temperature rises.
So you have to use relative humidity rather than the total water content.
Yeah, yeah. Now, if it were a three dimensional model you don't have to do that because hydrologic cycle you put in the model would take care of itself.
But in the one dimensional model you have to assume the relative humidity, the question, and this is a crucial point because the other Greenhouse gases other than water vapor is long lived Greenhouse gases. Residence time is very long [?]. So that essentially atmospheric concentration of these gases, the timing by how much source and how much sinks and the long term balance and gradually, slowly changes. On the other hand water vapor residence time is a matter of a week or so, so that essentially not determined by intensity of sink and source but determined by the moisture holding ability of the atmosphere, which is saturation vapor pressure. And so, but on the other — in the three dimensional model it takes care of itself rising air, saturate and precipitate, falling air, [???], and that process is automatically taken care of and the other, as you get about 50 percent humidity everywhere. And so that's — But then if you want to do it one dimensional model, you have to simplify, and that's what the simplification is. So hydrologic cycle, I replace the hydrologic cycle by one assumption to fix relative humidity. And that was a very, in my opinion, very effective simplifications which I managed. And that made the one dimensional model useful for the study of sensitivity of temperature to Greenhouse gas concentration. And so that is a very interesting thing there. So what I'm concerned, sort of my developmental model is practically inseparable from the physical inquiry. But now new generation of modelers, they just do modeling as if it were objective by itself and don't seem to be as interested to try to understand the nature.
You know, we've been going for a couple of hours now.
What time is it now?
I think it's about noon. My watch has actually stopped.
Yeah, okay, good. Okay.
Maybe we should stop and take a break for a little while.
Yeah, yeah, yeah, yeah. Let's see. My wife, where she is now? [after break...]
I think we are covering pretty effectively. You are an excellent question and asking.
Well, one thing I think is this Japanese group like numerical weather prediction groups, it's a sort of a very close knit group. Of course you have University of Tokyo group and the meteorological agencies or groups and so forth, there is a different subgroup there. But [???] we are kind of very close. Everybody knows what everybody else is doing, and also sometimes you will feel lonesome doing something independently from the group because —
I was going to ask you about this. I mean, partly I'd like to hear a little bit about what it was like for you to come to the United States and —
Yeah. This is the thing that I tried to tell you, is that to be in such a group you feel secure by doing something collaboratively, but you are sometimes afraid to be alone to go different direction from the group. Feel secure just the group. It's maybe same kind of feeling as some kind of religious sect. Once you get in there, sometimes it's very difficult to come out.
And so by coming to the United States and then sort of working away, independent of group, I didn't have that much contact with the group, although Gambo came there visiting again. I guess he came to numerical weather prediction groups as a visiting scientist, but that's a short while, and most of the time I sort of almost completely detached. And in the U.S. I felt like I can do anything I please almost. Although I was collaborating with Joe Smagorinsky, but he more or less left me to do the things that I am responsible for.
Do you think this is partly a cultural difference also?
That America has always been about individualism and Japan is more a group in society.
Yeah. So you get credit as a member of group rather than try to learn differently on your own direction. And once I come to United States, I felt so relieved, sort of being by myself and do whatever pleases — although I am in yet another group, obviously. But the way it goes is sort of Smagorinsky is very careful in creating responsibility clear so that you don't have to interact so much in order to do your own job. And so you know I was doing radiative transfer and so forth. Well eventually I may try to help him to get this model running smoothly, incorporating Arakawa's ideas and so forth so that hydrodynamical system can run stably even without nonlinear viscosity in it or small, relatively small nonlinear viscosities. And so I was doing something to help him, but I always had my own project which I can enjoy, do my own pace and so forth, and this is I think very, very important in sort of do, opening up new territories or getting into new interdisciplinary fields. And I think this may be one of the important differences between U.S. and Japan here. Here they let you go your own way, whereas Japanese societies are in a very narrow place, so many people living together, and so in a sense I felt relieved coming out of the numerical weather prediction group. Now, some of the Japanese scientists you interview may feel differently, and they are more sort of Japanese in that sense. I probably have as much Japanese as any of them in some sense, but maybe I felt relieved for some reason. And —
How good was your English when you first came here?
It wasn't bad. I went to, um, there was some group, tropical cyclone international conference or something, so you know we went to conversation school in which conversation school and in the office we decided to discuss our issues in English to prepare for conference and so forth. So by the time when I arrive here — Also, you know, for entrance examination you have to be pretty good in English grammar and so forth, reading and writing. May not be good in terms of speaking. So if you started speaking, then when I get here I have still a problem understanding other people saying, but I had absolutely no problem reading Washington Post or any of these things. So the English was not any problem. However, there is another benefit working in the United States in contrast to Japan is that the — Now in Japan also they now start course write in English on paper, but here you have to express much more specific way. You have to say things more clearly, whereas Japanese your sentence very often is not complete. You leave sentence blank. So Japanese models, there is many dots there, and sentence is partially, the sentence is incomplete. So then in Japanese literature class in Princeton, examinations sometimes fill in the, answer the part of the sentence.
Yeah. And also sometimes subject is abbreviated in order not to be too specific. Verbs [?] is eliminated, or sometimes sentences suddenly stop in the middle. Leaves the other listeners to fill in the blank. And so it's not a very suitable culture to communicate. And also another thing is that in order to accommodate each [???] very nicely so that the — in Japan they don't try to articulate what is the difference in opinion, what we agree you and I, and what we agree/disagree. And to try to make that agreement and disagreement very clear, then you can begin to see where is an unsolved problem in science, but by so doing sometimes you have to [???] the danger of offending others. So that these things, so that when I started writing the paper here, even though I was writing paper in English in Japan, but nobody to scrutinize me. But here I can, I have to send in the paper in the journal, then have to be reviewed and all that kind of thing. If English is too bad they just return your manuscript practically. So —
Does that mean there was no system of peer review in Japan?
The peer review is there in Japanese journal. However, it is usual, even though that may not be true, but people believe that they are able to identify the reviewer, because of the small group of people involved, and by reading from the tone of sentence you say hey, this must be the guy. Okay. Now sometimes that could be a wrong person, even in Japan. However, the problem is they think they know who reviewed it, okay, so that then it may create unnecessary friction, so unwarranted friction, between some people who actually didn't review. Okay? But anyway, [???] whether they are correct or not doesn't matter. The fact that people think that they can identify the reviewers. And then if you write a critical review, then you, you know, you're sort of dishonest or your face is not saved sort of thing. And so that is a problem. The peer review system can never work there. And so that it's, among scientists' interaction you can't be very frank. Being frank usually helps each other. And if you disagree and one is right, another one is maybe incorrect. But you know usually things are not that black and white, and so never one is absolutely right, another is absolutely incorrect, but certainly to try to make clear that what point of agreement and disagreement, the first step to go to next step. And even though you are wrong, very often you become famous by being wrong. Right? If you stimulated the interest of people. Like my friend Wally Broecker or something in Columbia. He is sometimes wrong, he is sometimes right, but when he is wrong still he is helping others to get issues [???]. Right? So it doesn't matter really to try to articulate a difference, you know, among people's differences. But so these are some of the hindrances, some of the factors which hinders Japanese science. Even though many, many brilliant people get into the field, but so in terms of brilliance are maybe so-so in this country, the people like that, sometimes they achieve a great deal because of this system here. And the people are willing to disagree with you. And so like this Greenhouse science debate and my politics debate is something else. Science debate, people disagree. So a friend of mine like Dick [???] disagrees like this. And if he disagrees, originally I felt very unknowing. You know, my friend is disagreeing, my water vapor feedback is wrong, and this is wrong, that is wrong, you know. And so it's kind of annoying. But then the people get, start studying more about water vapor feedback, observational evidence for or against his point. And this progress entire field. And so one have to enjoy the disagreement and the debate which comes out of it in order to enjoy scientific research, and don't afraid maybe being wrong. And might well enjoy the debate.
Now, when you first came here, did you think that you were, did you know that you were coming to stay for a long time?
No, I wasn't sure. This is another good thing to come from one country to another. Particularly [???] there with no job in Japan. I left graduate school as soon as I get Ph.D., so I didn't have a job in Japan. So I came here and if Joe Smagorinsky decides that I'm not good enough then I have nowhere to go. Okay? Although the contract is sort of temporary appointment, and so I get a special kind of result from — And Joe could fire me anytime, or go back to Japan, somehow find a job. Probably I can find some job in Japan probably, but I didn't feel like that. I felt that I was, tried to demonstrate I could do something.
So you must have worked very hard.
Very hard. And this is another thing that the U.S. system is very good, because no matter how much you achieve, but when you really don't get any praise, if you stop producing in this country you can't sit on yesterday's achievement. You know, as soon as it stops producing and grant money stopping or people say, "Oh, you better get out because younger one have to replace you" because, you know, you have to, organization has to run full speed in this country, so that many good people are waiting behind you. Right? So that you are comfortable in your job as long as you are producing the research, but yesterday's work doesn't help you to stay in your job in a comfortable way. And that's another thing I think is, system here is good for science productivity. You never feel you are comfortable based upon what you did yesterday. And so that way in Japan when up to a certain age when you get to something you work about ten years or so, after 10-15 years after Ph.D. you write a few good papers, and then you become a professor at the University of Tokyo. Then you made it. And then, about that time, you suddenly you know are invited to all important committees, because professors at the University of Tokyo have to be in their committee in order [for] that committee to be respectable. So then he has no time to do things. Furthermore, because of the system in which professor, associate professor, assistant, [???] assistant professor, it's a hierarchy there. Professor sits on top. Europe is the same system. Whereas U.S. is more networking. And this system, you don't have to do your own work. The people you supervise can put your name on the paper long after you stop producing it yourself. Right?
And so still that, but it's a practice slowly disappearing in Japan, but still it's a very prevalent practice. And so these are some of the factors which makes U.S. science great. You know, sometimes they say crisis like the missile gap and so on, these gaps, or now they talk about the modeling gap. And they talk about that if you know oh U.S. is going to lose this, these things, you know. U.S. isn't going to lose. Because yeah, in some sense maybe you know because of U.S. restriction about the computer or something, so that U.S. scientists may not get the best computer anymore. But that doesn't matter. It always comes up. And computer looks like this, but then you know like the high resolution television set, you know, Japanese started getting a nice, high resolution, emanation, then said, "Oh no, let's the rule of game. We are going to go digital." Right?
Things like that. Or Japanese [???] this what they have to compete is to create a bigger and more and more powerful processor. Then U.S. say oh no, don't put a processor, but the cheap processors, thousands of them. And see what happens. And so this is I think that diversity of ideas and different thinking. You know they encourage a different way with thinking, and they encourage people to go their own directions, and then if that direction happens to stink, then that's your choice. Right? And so that I think is very good. Of course in my case of course Joe had a very nice vision there, so that I was helped getting started there. Now once —
Now did you have much contact with the other members of the Japan numerical weather prediction group that came to this country? Arakawa and Kasahara —?
Kasahara at the beginning, shortly after all these three groups doing something, producing something, except Chuck [?]. Yeah, yeah.
You mean NCAR.
Yeah. Chuck Reese. No, NCAR wasn't in there yet, so at that time Joe's group, and then UCLA started shortly thereafter, you know maybe quite a few years after, and then after that — No, Chuck was studying about the same time, Chuck Reese. And so then, after all three groups start producing something, then NCAR, Kasahara came here.
And so I remember I went there to help him, or discuss him the issue of modeling and what we are doing or we are not doing. Because Kasahara always a nice person to meet. When I first came to United States, I first went to his house before I see Joe here. And he was always —
He was already here?
What was he doing then?
He was doing, he was at the University of Chicago at that time, and doing some numerical prediction of hurricane with Joe Depratzman [?] over there, who happened to be one of the, that young [???] at Institute for Advanced Study.
But he went back to Chicago, because of that temporary [???] there for Joe Depratzman. But he was very good in helping everybody by applied math when they tried to integrate equation motions and so forth. Joe Depratzman was an excellent person to have around. And so Arakawa [sic; Kasahara] was working with Joe Depratzman, and then after that he went to NCAR and to take care of general circulation models. And Chuck is the one doing, putting hydrology. He didn't put radiation very much in there, but he tried to, radiation I think his radiation was highly parameterized then. But he tried to, you know, I think maybe yeah, the high, uh, low resolution atmosphere. And he is amazing person; he's a very bright physicist. We are doing by group. He was doing it by himself.
That's right. I interviewed him [???].
He is a brilliant person, and suddenly said goodbye, you know, Edward Teller was his mentor or whatever, I don't know exact relation. And then okay and I have done this, and now let me do whatever I please, and then he started doing this project. He did it himself. He [???] this thing and it started running. Now two things, the problem there is that his finite [?] difference system was computational unstable. So it keeps blowing up, so he had to put enormous viscosity in it. His model was swamped with viscosity. And the moisture convection scheme is some kind of diffusion parameterization, frequent [?] diffusion parameterization, something, you know, depend upon something like [???]. And I am not sure exactly how he did it, but he didn't face up to this problem of how to handle this instability of first kind, that there is a convective instability coming from unstable stratification. When it's unstable what happens is if the atmospheric static stabilities are unstable, then you get of course convective motion of course. However, scale of convective motion, smaller the scale is, the more unstable it is. So if you leave the unstable stratification dry or moist in a large scale, then you are bound to get convection at the scale smaller than grid scale. Because smaller the scale is, more unstable. So finite difference system cannot handle this kind of static instability. So if you leave the unstable stratification. So what I did was to neutralize the lapse rate, thereby eliminating. Now Jules Charney and [???] tried to do it with [???] pumping, [???] boundary layer convergence and relay [?] this heat of condensation proportional to boundary of convergence. And so various people tried to do different way to avoid this what we call convective instability of first kind, or static instability, static or convective instability. And so I devised moisture convective adjustment, and which Arakawa and Kasahara was talking, and Arakawa first you remember.
Yeah. He can send you his manuscript, I'm sure. And so that was an early development, so I tried to solve this by neutralize the lapse rate in [???] scale. Because whatever unstable stratification there, then small scale motion grows in [???] to scale. Eliminate that instability. So that was the physical basis of convective adjustment there. So I eliminated that, so ever since our model went very nicely, and then we had this nonlinear viscosities in there also shortly after we borrowed from Arakawa that the energy conserving system, famous energy conserving system, we put that in. So our model runs very smoothly. Whereas Chalk's [?] one is numerically his [???] scheme, but numerically, computationally unstable. He didn't handle that static instability problem either, so that he had to swamp the model with viscosity. And about the time he was ready to write a paper, he realized, bright guy, brilliant mind, he's a bright guy, realized this whole system is unstable. So that's why he never published the paper. So that's —
When did you first find out about him and what he was doing?
Oh, he's a good friend Joe Smagorinsky.
Yeah. So we are sort of — It's interesting. You know, here are three guys, three honchos here, [???] Mintz, Joe Smagorinsky, and Chuck. Now [???] and Joe is both Jewish persons and very competitive. So you know they are sort of fighting each other. Joe said you know I started to think you know primitive equation was I started this and I take care of linear viscosity. Now, [???] come behind, but then [???] quickly tried to do everything we tried to do. Joe actually at the beginning went to UCLA, tell them how he is doing these things, and then quickly [???] started following and tried — And then shortly after [???] started getting Arakawa's help. And so these finite differences, theirs is better. I can't compete with Arakawa. Okay? Right? Theirs is better. Right? So I have to just bow [?]. But fortunately Doug Lilly, as soon as he heard Arakawa created the energy conserving system, Doug himself figured out how to do it here. So he immediately put in our — He's an excellent applied mathematician too. He has a good physics mind, but — So we got this imported. About that time I think Chuck ran out of steam. He got so tired after a while. Imagine doing this by single person?
Yeah, right. And he did all the programming.
Yeah. But you know his personality is so nice. You may have noticed it. Very, very nice person. So neither [???] nor Joe, you know, hated him. They actually liked him — even though they are competing. So [???] and Joe got adrenalin flowing by competing with each other. And Chuck just simply worked hard. And he is a friend to both of these guys.
Do you remember Chuck's film of his model?
Yes, it's a beautiful film.
Yes, it is, and I've actually got a copy of it.
Yeah. And so then [???] coming like this, but [???] very, very viscous world [?]. And I don't think there is any cyclone waves much in there.
Right. What was your reaction to seeing that film when you first saw it?
I was really; particularly I was interested in the way tropical rain belt pulsating diurnally, like this. And I thought that it's a sort of you know, it's a very nice sort of vast try at things. But we are waiting for his paper to come out. But never came out. But a good reason. One is his system was unstable from convection viewpoint, and then advection [?] calculations. And so we end up being the first one to publish hydrologic cycle.
Right. Did you then, or have you ever since then wanted to make a model that had a, make model visualizations that moved like his, like that movie?
Yeah. But the thing is that at that time to make that kind of thing is not easy. And I was so up to here, even though it would be nice I thought to do that kind of thing, there was no way I could get the thing done. And the time when I started making a movie was the middle of 1970, long after Chuck did these beautiful movies at the beginning. And he has a kind of personality, liked by everybody, you know, and I think he gave us a nice stimulus there. Yeah. But you know, in hydrologic cycle he really, Joe comes in from, he started doing this rainfall prediction. And then I put a convective adjustment and make it work. And Chuck never get into beyond making movies there. So in a sense movies may be a good thing, but in the other sense sometimes, you know even now they do it, they make a beautiful movie, but stop digging.
Right. Now, when I talked to Arakawa, he said that his impression was that Mintz and Smagorinsky communicated for a while and then Smagorinsky, as you say, got competitive or maybe both of them did, and stopped pretty much.
Yeah. That's right. At the beginning Joe went there and spills all the beans. Okay. But then after a while Joe gets sort of conservative in doing that, and [???] now died. It's sort of he is a person like sort of teenagers at age of 60-something. It's a sort of very enthusiastic about the things and once he is interested in something keep digging in, digging in, digging in, and [???] no end, you know, continuous bombarding of questions and never give you any chance to respond to his questions. He sort of overwhelms everybody.
Now I am a sort of very talkative personality, as you have already noticed, but even I have no match with [???].
I know one other person who is probably very similar, and that's Lou Schneider [?].
Who has got to be one of the most talkative people in the world. [laughs]
Yeah. So you know, the way, you can never interrupt him because he says something and then he make a conditional statement and, and, and but this is, but under this circumstance this doesn't work, but, but this, you know, and he kept going on and on. And then you think that he will stop at that point, and then he starts a new sentence, and then this sentence never ends, you see. And [???] is exactly like that. But main difference is [???] is even much more, ten times more aggressive than Steve Schneider. So no wonder Joe became a paranoiac. Right? And but what happened is that we are doing the things which they are not doing. That is, putting radiative processes, putting a hydrologic cycle in, and by putting radiative process in we started doing interaction between stratosphere and troposphere, trying to understand why the stratosphere behaves the way it does. As a result with interaction. And that's the first paper. Next paper is we discussed the hydrologic cycle simulations. Now [???] which UCLA group was doing. Their model was dry, their model was two levels, they don't have a stratosphere, and but Arakawa made a major contribution there by putting this finite difference which don't need a nonlinear viscositive to keep it stable. And we borrowed from that. So he made a great help to us, and eventually we probably influenced everybody by starting putting [???] radiative transfer and putting a hydrologic cycle into the model. And so that way we have affected and they have affected the field. And —
Yeah. So, tell me exactly what you got from Arakawa and approximately when.
That was a time when I was learning, we are learning GCMs, this 9-level GCM successfully. However, I —
This is the early '60s, '63, '64, '65?
Yeah. Yeah. We are running, we published that paper '65. Okay. And we are running that around '63 or so already successfully. However, I don't like the fact that nonlinear viscosity is dumping too much kinetic energy. And so when I find out Arakawa's finite difference system, I immediately borrowed. Fortunately I don't even have to bug Arakawa, because this Doug Lilly who is [???] with that immediately figured what they did. Once you know, you can do it. It's easy to copycat.
And so we immediately put the energy conserving system in, thereby reducing the viscosity.
...that you can benefit from your friend, you know, who is collaborating with you or people who are friendly to you, friendly scientists. However, most benefit you get from those who are competing with you. And [laughs], so this is why I'm sometimes kind of reluctant like family tree things. You know that. The reason why it is so, is this kind of communication going apart from family trees. Because there is no such thing that, no evidence we copycatted from Arakawa, but really we did! We got it. Or other people. Okay, Arakawa didn't use convective adjustment, but you know at the end of his scheme he does convective adjustment. You know that? And so, you know, and basic philosophy of convective adjustment is a very good one, so that I affected just about everybody in a very subtle way. Or explicit radiative transfer and all these things, or having explicit radiative transfer, then discuss the interaction with the stratosphere. And hydrologic cycle so we can, we begin to do beautiful simulation of [???] general circulation. I think we are the first to do that. That is the consequence of putting hydrologic cycle in. So it looks as if, in terms of family trees, there is absolutely no relationship. But it doesn't matter. The influence goes differently.
Right, right. Now around this time, the middle of 1960s, one thing that starts to happen is that there is a concern about these proposals for the supersonic transport.
And you did some work I think on the possible effects of water vapor release in the stratosphere by a supersonic transporter and maybe other things.
Yeah. Now, at that time now you have to — this is again related to this, but you know we had this 9-level model which has a down [?] stratosphere. It's a sort of [???] finite difference level in the stratosphere. So these two papers we published, Joe Smagorinsky and I collaborated. One paper he senior authored, and the next one I senior authored. That was two papers there. And the papers, first paper he senior authored is the one in which we discuss the interaction between stratosphere and troposphere and why the stratosphere behaves that way. And so that's how we started getting into stratosphere-tropospheric interactions in the atmospheric general circulations. And then around that time I get an interesting letter. I guess Joe got interesting letter from Australia. Named Barry Hunt [?]. And he proposed to do ozone, interaction between ozone photochemistry and atmospheric general circulation. Way ahead of time. I mean, it become Ozone Hole and everything. And so he proposed to come to GFDL, use a GFDL model to study stratospheric tracer study including ozone for the kind of study dynamics interaction. He's a master, just got to master thesis or something in Australia. But he is an interesting fellow, trained in physics. And I liked his proposal, and Joe liked it, Smagorinsky liked it, so we decided to invite him over. And then he developed from 9, extended from 9- to 18-level model, which sort of treats stratosphere much more explicitly than the first run we made. And so middle 1960 we started getting into stratosphere, why stratospheric temperature distributes the way it does, and also we tried to put some kind of tracer, like nuclear fission material, put in the stratosphere, or we put ozone-like tracer and tried to see how stratosphere moves things around. And this eventually led to Barry Hunt and I collaborated, but he's senior authored about stratospheric tracer study by using this kind of model. But after a few years Barry said, "Syuki, collaborating with you is very nice, but you know everything that begins has to end. I want to go back to Australia to do my own stuff." Okay. Typically Australian. He wanted to go back to Australia. "I'm not going to work with you forever. Thank you, thank you for a nice collaboration," go back to Australia. And so then I have to hire somebody, so I consulted with [???] who he has no laboratory but has a close connection with Department of Energy, and —
Before you go on with this, did Hunt take your model with him back to Australia?
He did, but that's a later version of models there. The Australian connection has two [?]. One is I created some of the simpler version of model with lower resolution, and he took that with him. That's one way we influenced Australia. Later our GCM are used by Miakoda, who wanted to do weather forecasting. So we gave to Miakoda to use our GCM for weather forecasting purposes, because now it has a nice explicit radiative transfer, which was very important, and then also hydrologic cycles are treated very nicely. So it is ideal model, so we exported expertise from numerical weather prediction. Now we are able to export back our expertise back to the people who don't want to extend the range of numerical weather prediction. So we gave that to Miakoda.
Okay. And Miakoda was where at the time?
In our laboratory. One of the Japanese scientists who came after me. Okay. And he is a guy who Joe Smagorinsky also originally named one of the two, and I happened to come here. And second one was Miakoda. So eventually he came here too.
Do you remember when that was?
1963 once, and then '65 the second time or something. And then he stayed on. So now he came in and he was doing extending numerical weather prediction by using now GCM. So originally we exported their thing. Now GCM is being used by — So that —
To your knowledge, is that the first use of a GCM in [???]?
Yeah. Yeah, I think so. So this is — And then Miakoda was collaborating with Smagorinsky, and they have a lot of collaborative papers. And then, and then he became more and more independent from Smagorinsky and he has his own group, and eventually what Miakoda did, inspired, or Miakoda did activity inspired European center to be established too. So this weather prediction which uses general circulation model now went back to weather forecasting and then started stimulating extending range of weather forecasting, which instigated European center.
How? Who decided it? Who was inspired by this?
Because one is of course Joe Smagorinsky by now is an important member of joint organizing committee of GOP [?], and Miakoda wrote, Smagorinsky and Miakoda wrote these papers that these weather forecasts can be extended more than a few days. They demonstrated feasibility. And this is how they — And at the beginning of European center establishment, Miakoda went and they explaining their thing, and people like Joe and Jules Charney and all these people encouraged them to establish such an establishments, such institutions. So what is very nice is, again this is not quite family trees, but original European center used many of our innovations. They got very depressed with other calorimeterizations. They actually get different people the parameterization, and then they created this. But the fact of the matter is that GCM actually, we benefitted from numerical weather prediction, but now GCM is good enough to affect the extended range. Now one of the important things is of course rainfall forecasting, which was in GCM. Another one is radiative transfer. Because the old numerical weather prediction model doesn't have a cooling into space. Okay. So, it's almost like running engine without a cooler. Okay? So after a few days the model fails. You know — And so that's why having explicit radiative transfer, not the detailed radiative transfer code but the explicit radiative transfer is extremely important. And so that's how this GCM activity. And then another thing is that Miakoda keeps testing these forecasting models which we gave him. And then Australia sent a few young fellows to GFDL and brought that model to Australia for their weather forecasting.
Who were they? Do you remember?
The guy named Burke. Not Burke, uh, not Burke. There is another guy; he is now deputy chief of Australian Weather Service. I keep forgetting these names. Gauntlett. Gauntlett.
Can you spell that?
G-a-u-n-t-l-e-t-t or something.
Gauntlett. Something like that. It is a standard name. And Doug Gauntlett, Douglas, Doug Gauntlett. And at that time he is a young boy, but now he is a big wheel over there. And he brought that model to Australia, so some of the early Australian weather forecasting models are exported from GCM, uh, they are exported from GFDL. And when Chuck Reese first want to run the — He fed up, he is by now he is the director of atmospheric dynamics division.
And so he thought, and he was instrumental in getting people like Steven Orsdag to develop this spectra transform model at NCAR, and also it was a code developed by, independent but a code developed by this spectra transform method.
Can you spell those names?
O-r-s-d-a-g or something, Orsdag.
Orsdag. Steven Orsdag. He is a professor at Princeton now. Okay. Applied math, or maybe aerospace. But anyway a professor at Princeton. And Chuck was instrumental in getting Orsdag to do this for spherical coordinate, inviting him as a visiting scientist. So he wanted to do this spectral model rather than finite difference. He knows it's more accurate. So NCAR always wanted to go from grid model to spectral model, and further it didn't have — this is now we jump to middle 1970 — doesn't have to, for me that [?] was showing that I used the spectral model, I got the beautiful result in some kind of Miss Universe Contest of models which is held in Washington at that time, and they decided to give me the Miss Universe award. But by the time they called my name, I was already on the train and coming back, and I missed receiving this award. It's a kind of joke like award anyway.
It seems so.
Is that really what they called it?
Yeah. But you know it's not [???] they could have sent me the award after I got, but they never did. So Chuck decided at the beginning to import this spectral GCM from Australia with all the physics package we had. So the beginning the GCM which they are developing since Katahara's [?] never seemed to work very well. And so Chuck finally fed up with it. And Bob, Chuck wanted Bob Dickenson to develop a new model, but that's also [???]. So he said okay, heck with it. Syuki, I'm going to get the spectral model in, he said. Okay? We are good friends. So [???] said okay. And so he got this model from Australia, and that's was the CCM-0 [zero] is.
Yeah, yeah, right.
CCM-0 came back from Australia. But [???] our GCM model was used for extended weather prediction by Miakoda, and then that was exported by Miakoda to Australia through that Gauntlett, and that was the way that Chuck broke in as a CCM-0 [???]. And but you know there is a difference there, because when they went to Australia it was a finite difference model still.
Was there any controversy at your laboratory about giving away a model like that to the Australians?
No. No. Australia is fine. That's another hemisphere. So Joe probably never wanted to give model to [???] or Mintz, but [???] never wanted it. He thinks GFDM model stinks anyway. Okay. So [???] never wanted it anyway. And [???] sometime you know say you know, "Why you need GFDL? We take care of everything." [laughs] So it make — I never forget. This is a joke, probably not true, but oh maybe not. And you know one day this storefront IBM office which we [???] from IBM, the front glass window was huge, shattered, broken in the morning. So Joe came in the morning and looked at this. "[???] must have that." [laughs] Yeah, so anyway, that is a sort of yeah, yeah, I get in a tangent now, but so there was three [?] model. And well eventually everybody started having higher particle resolution, and eventually everybody started having hydrologic cycle, eventually everybody started having radiative transfer, different [???]. And so I think we did our own share of contributions, which we some of the time we benefitted from Arakawa's contributions and Chuck also stimulated us into — I don't know, our hydrologic cycle development was parallel I think about that time. I didn't learn particularly about the way we did hydrologic cycle, because in hindsight I can tell ours is definitely much better. But he always stimulated us, because there is a sort of triangular competition going from there. So in a sense that was very, very nice, and his personality is such that he never make you feel like he is competing with you.
Yeah, yeah. Okay. Well, can you go over to the best of your memory, at least maybe for the first ten years or so the stages in the development of the main GFDL GCM.
Yeah. Now the best thing —
There's a 9-level model and there was an 18-level model.
Yeah, but 18-level was mainly done for stratospheric study.
But then what happened is then now that we have this hydrologic cycling place, primitive equation is running well thanks to nonlinear viscosity of Joe originally, which came from Von Noyman, and then Arakawa then at the conserving system. Put both together, so it's running beautifully. And because it's primitive equation, we can extend this all the way to equator. And so middle 1960 we start simulating global model. And around that time Krehara, who was my classmate, came to GFDL. And what happened is originally we had a global model here at GFDL like this, and one hemisphere projected to stereographic prediction this way, Southern Hemisphere predicted another stereographic projection this way. We patched together these stereographic at the Equator. In the process of patching we keep losing mass to nowhere. And so we keep losing mass in the tropics continuously by computational reason, because stereographic projection is — Northern Hemisphere is losing mass, Southern Hemisphere is losing mass, and accumulating nowhere.
Explain a little bit more of what you mean by a stereographic projection.
Stereographic projection is usually the —
[???] [???] [???] projecting on a flat surface?
Yeah. So flat surface. So you put a light in the South Pole or something, you put the [???] [???] here. So know that he put it this way. And the Southern Hemisphere put it this way. But what you do, if you put in a two plane [?], Northern Hemisphere this way, Southern Hemisphere this way, what you do air pressure [?] move one plane to another. Right?
And so in the process of patching these two together, we get a numerical truncation error which will make us lose mass. So that also must completely distort atmospheric circulations. This was one of the reasons why we couldn't, after we get the $8,000 a day machine at that time, we couldn't continue running because of loss of mass. And the tropics are the biggest absolute angular momentum. So if you lose a mass, entire atmosphere instantly starts blowing everywhere. Past [???], westerly blows, but after it easterly. And this is purely due to loss of mass from the Equator. And so out of desperation I put — In the tropics there is a great war [?], [???].
So our model is a hemispheric model at that time. So we had [???] UCLA because of Arakawa the [???] everything running smoothly, we've got the only hemisphere model. But what saved us is these two things. One is radiative transfer hydrologic cycle. So we can write a very nice, decent paper. But then Krehara came in and he developed this model which is entire globe, its radical [?] harmonics models for GFDL. And so —
So you were using originally a rectangular coordinate system, and he is moving into a spherical.
A spherical coordinates, yeah. And though then I can run the global GCM, and Krehara is much more meticulous than I am, so he can do. He is one of these original numerical weather prediction group. He is my classmate, but his brain is so exact and brilliant. He can figure out all these things very nicely. So he got, co-created this global spherical harmonics grid system. So now I can then global general circulation model. So then I started doing fast successful simulation of tropical general circulation, faster successful simulation of global hydrologic cycle. It's sort of tropical circulation. Oh, it is a marvelous simulation. And now that's sort of one of the foundations for El Nino study now and everything. Beautiful [???]. I couldn't believe it. And I tried to dig out this morning this letter from Fred Saunders [?] of MIT who is a diagnostics analysis person there. He looked at our simulation. He couldn't believe his eyes, how good this simulation of tropical general circulation is. So he went to Jules Charney and showed this, you know, they have produced this beautiful simulation of tropical general circulation and tropical rainfall, rain belt and all this. Beautiful. How come? Jules said oh their convection parameterization stink. Right? I don't know whether Jules or not. He didn't say Jules Charney, but I am pretty sure. So we get it for wrong reason. Some resident expert said that he said. May not be Jules, but that's what he said. And but I think this is a moment of great triumph. We got a beautiful simulation of tropical GCM and the global hydrologic cycle and towards the end of 1960. And finally seasonal variation papers published middle 1970. But I think that's the first successful simulation of geographical seasonal distribution of rainfalls and tropical general circulations. So one season simulation was done by 1970, and seasonal variation was done by middle 1970. And I think in terms of simulation together with [???] 18-level model, stratosphere simulations, hydrologic simulations, and tropical general circulation. I think these are landmark simulations. We are first one to be successful. And by using, mind you, using this convective adjustment which everybody criticized. And so if you shut that thing I can show you how it is.
Sure. [tape turned off, then back on...]
And we are using very simple parameterizations, but despite these simple parameterizations we get very good simulation. Because now tropics and many of the rain come out of disturbances, cyclone waves and particularly middle latitude. If you have a convergence [???] and the cyclone waves, then you get a rainfall like this morning. So if you get the cyclone waves simulated right when you time average, then you've got the rainfall distribution right. Now the tropics, it's somewhat easier, because you can see here, because this is sea surface temperature distribution. This SSD distribution essentially dictates where tropical rain belt is. So this is what I found out early on using very simple convection parameterization. Jules Charney and everybody said, "Oh, that's too simple. Too simple." But use that and then I was able to capture, because it doesn't depend upon that detail. And so I keep using this parameterization, and then everybody criticized because everybody else working on improving that parameterizations. But I knew that early on I made this simulation I know what kind of simulation I can get. So, for certain purpose of problem solving I know it's good enough.
Well, that's a good point to go back to. I have a question as to a while ago. We got kind of off into the Australian connection. Which is the SST controversy and your role in that sort of political discussion about the water vapor release in the stratosphere.
Yes. I mean the SST is, you know, the Greenhouse Warming or something? Or what SST?
Well, yeah, there was, I think it was a National Academy of Sciences study in 1966 or it started in 1966. It was about the possible effects of the SST, which mostly I think —
Oh, Super Sonic Transport. Oh, oh, SST, oh, I mixed up.
Oh, SST, not Sea Surface Temperature, yes.
Yeah. Super Sonic Transport. So we got sidetracked there. But stratospheric circulation model. So Barry Hunt went back, and but he didn't do anything about SST. So we have to hire somebody else. That what I went [???]. Okay. So the guy we hired was a young fellow who looks awfully young at that time. His name is Jerry Marmon [?].
Yeah. And so is the director of GFDL now. So Joe and I went through all the lists of the young people who can continue stratospheric tracer study. And so Jerry came to GFDL, took over stratospheric circulation project. At the beginning I was working with him, but after a while he says, "Syuki, I rather want to do it myself so I can get all the credit." [laughs] So then I realized yeah, I worked very hard on this stratospheric problem, and it breaks my heart to get out of stratospheric circulation modeling, but I realized I can't do everything. I hire other people in order not to do everything, right? So I quickly realized rather than spreading myself thin in the stratosphere and the tropical general circulation and then by then gradually Greenhouse Warming things. This is around 1970, okay, so I got into Greenhouse Warming issue too. So I was doing all these. I just can't do everything. So I said okay, you've got it. So Jerry Marmon started doing, around 1970, the stratospheric general circulation modeling, and he get involved in originally Super Sonic Transport, Sire [?] Program they called it, and it later became the —
[???] CIAP, Climatic [???] Assessment Program?
Yeah. A guy named Grobeck [?] or something running the show. And then later it become an ozone hole issue, so that GFDL equipment or contribution to stratospheric issues from the circulation viewpoint rather than chemistry started from Barry Hunt and then passed over to Jerry Marmon who then sort of used his model to sort of collaborate people, chemists and all these people to try to put the circulation aspect of these things in the stratosphere. But this again originated this thing to this original thing, the explicit treatment of radiative transfer so that we can get stratosphere into it. And so one of the things we, yeah, so that's, the history goes back to again 1960s. And then so that is, so simulation at that time, this simulation is good because its grid size is 250 kilometer by 250 kilometer. Okay.
That's very small. What was the time step in this?
Yeah. Oh, it was very short too.
This is 1973 [?] where Manabe, Hahn [?] and Halloway paper.
Yeah, yeah. Ten-minute time step or something. Okay, very short. And you remember the computer is much, much slower than desktop, and we are running this. So it took two and a half years to run, it took two years to run two years of experiment.
And we are using about more than 50 percent of GFDL computer time, monopolizing it. We created two thousand magnetic tapes, many of which we couldn't read first time. So I had a nightmare on this.
And of course I didn't anymore, but the people like Leith Halloway and Doug Hahn and all these people really had a nightmare to try to read this thing. So it's a sort of numerical experiment way ahead of time, but because of that we get a big headache. But standard with simulations again very excellent. And so that's what we had there. So in a sense that these developments continue from early 1960 throughout 1960 and just about every paper written by 1975, so that this, Joe's dreams kept on going on and on, and went up to 1975. This is a tropical circulation paper. We have a similar paper one year later about hydrologic cycle, how well we simulate hydrologic cycle and these bion [?] classification simulations. And later people did that 15 years later, but we had done it. This published 1975. And with seasonal variations. And so that's about sort of the big project Joe started went to 1975 and culminated into this. And I know pretty well what kind of parameterization I used. You know, this simple parameterization can guarantee pretty good simulations. And that's what explained my attitude. Sometimes I don't change my parameterization. People criticize me like anything. But I know what I get and what I don't. And so then, so that's one development. But then another thing started in middle '60. That is Joe's dream also, and I think that we have influenced him too, but this climate system cannot be studied by atmosphere alone. And you have to have an ocean. So Joe invited Kirk Bryant [?] from Woods Hole. There was another Japanese scientist we tried to invite. He died [???]. But he instead went to Scripps. So we went to Bryant, Kirk Bryant, who was at Woods Hole Oceanographic Institution. He is originally a meteorologist. And we got him from, Joe got him from Woods Hole and then started developing ocean model. He came here around 1963. You can check the date. And started developing, immediately started developing simple barotropic ocean model, become a baroclinic [?] ocean model, famous Bryant-Cox paper of bioclinic ocean model published 1967. And by then already Kirk and I started collaborating developing coupled ocean atmosphere model. And the best paper —
By 1967 you're saying?
Yeah, yeah. And the best paper is early on which we wrote early on is this very short paper, and still I am very proud of it. It's a pretty good paper here. This is '75, this is yeah, I may not have it here.
I might have it myself. Let me see the —
Yeah. This is the paper which Kirk Bryant write, I write, published in 1969. 1969. It says "Coupled Ocean Atmosphere Model." Maybe you turn this off [???].
Yeah. [tape turned off, then back on...]
Yeah, which sort of discusses this 4-page paper with Kirk Bryant, this climate calculation with combined ocean atmosphere models there. And then we put this ocean circulation model, this idea like geography, put the ocean circulation model in here underneath here.
But another one, we put a swamp ocean, that no circulation, and then comparing them we begin to see what ocean current does, oceanic heat transport does change the climate. And so, and has some [???] circulations and which later we have a hard time getting it back. But this paper had just about everything — narrow [?] Gulf Stream, because we put more resolution in here, in the ocean, and it is an excellent paper. It's almost like you know Greek and Roman times to Medieval Age and next, throughout most of 1970 we are sort of struggling, struggling, couldn't get much better result, so that we didn't use this coupled model for Greenhouse Warming study, even though we started doing coupled modeling late 1960, we didn't use for Greenhouse Warming purpose until late 1970 when we started using it for Greenhouse Warming and so it took a long time. But the good thing is that coupled modeling started late 1960 in collaboration with Kirk Bryant, and this paper I am really proud of. Probably one of the best papers along with this 1967 Manabe-Wetherold paper. They are probably two of the best papers I wrote in my career, even though this is pretty short. There is much [???] paper of this 3-part paper which occupies entire issue of Monthly Weather Review and [???] let me do this. But this is pretty good paper. So that another happened so that it's sort of simulation of tropics and hydrologic cycle by atmospheric general circulation, going all the way to middle '70, and then coupled ocean atmosphere modeling, and then stratospheric circulation model which started this getting into really a stratospheric problem, SIAP [?], ozone, and all these things. They all started in 1960s. And by middle of 1970 I started getting pretty tired of doing these simulations. Because all these thousand tapes took a toll. And also simulation experiment, [???] very slow computer. You just do one experiment. You can't do that strip show analysis, or you can't do, you can't run many experiments, and so that I start going back to lower resolution model. Which simulation is somewhat worse. So then I induced a lot of criticism. People say, "Hey, his simulation is not that good," you know, and "He is doing all this, but he may not get the right answer." Everybody worried about for him. And but that started from somewhere latter half of 1970. But then, because I am so tired of doing things with doing these big runs [?], now I start focusing on the much more sort of scientific issues rather than model development and simulation. But I started asking more of the specific scientific questions. So 1967 we had this one dimensional radiative convective models of the Greenhouse Warming and so forth and then Greenhouse Effect as a whole, and then for the first time we discussed the water vapor feedback very quantitatively. Because before, as I told you, Fritz Mueller showed the traditional approach doesn't work when you put water vapor feedback. So it's a pretty nice thing, and furthermore we, that '67 paper started predicting cooling in the stratosphere. That's probably the first one to say increasing Greenhouse gas cools the stratosphere. And so that's what we did. So middle 1970 then I decided to use GCM to re-study these problems. And so that was in 1975 papers. So I [???] most of the journals if you — Yeah, this is a '75 paper in which this was the first one to use CO2 thing for doubling the carbon dioxide.
That was the one, that's a citation classic. That's the one which cited 300 [???] times, and this is —
Now, speaking of this, was this the first time that you had decided to use doubling of carbon dioxide as the thing to study?
No, we did it in a '67 paper, the radiative convective equilibrium.
Okay. What — I mean, it seems in a lot of ways like an obvious choice, but [???] decide to study doubling as opposed to you know quadrupling or some other —?
Yeah. Quadrupling, I thought it very far-fetched. Right. Doubling I thought it feel possible. But you have to realize —
Were you thinking even then of human induced [???]?
Yeah. No. This time already, yes. Okay. This time it's already yes, but —
[???] 1975 paper.
Yeah. But not '67. Because Dave Keeling [?], who is coming to this symposium, he started measuring CO2, when was it, '58 or something, somewhere around that time.
Yeah. Around the IGY sort of. Yeah.
Yeah. So '58. And then this is '67 but you know somewhere, so it's not even ten years, so you already see this annual cycle of CO2 increasing by the time you get this. But when I started doing radiative convective, radiative equilibrium or something, people don't know. CO2 is even — there was the [???] in that paper, you know Aria's [?] CO2 measurement he punched together and some rising trend, but the thing is that Murray Mitchell found out that temperature increased from the time of century to 1935 or so, '40 maybe, and after that temperature started cooling down in Northern Hemisphere until 1975. So around '75, many people, including Steve Schneider, started —
Yeah. Cooling. And so when we met around — But then, irrespective of that, people thought that things are sort of in general warming up. So there was a study of man's impact on climate.
Right. What was your involvement in that?
Steve was one of the young students at that time, and he went there, this guy is, um, um, MIT's professor now.
Karo [?] —?
Yeah, Karo. Yes.
Is his first name.
Karo is the first name, uh last name. Karo is the last name I think. Wasn't it?
No, it's —
Yeah. I drag you into forgetting.
It will come to me in a minute.
Yeah. Anyway. So Steve was chosen as a bright young student who can help edit this thing, so he went there. I was probably at that time only maybe modeler who was doing this. And the —
That study is interesting because it talks at some length about GCMs and their [???].
I wrote just about everything about the GCM, because I just — And then I already wrote [???] of this paper, and which eventually published '75.
So you already had a draft of this at that time?
At that time I had a result. A result. So I talked about it at that time.
And you can see the little bit of evidence that I have the result of this. And so that was a '69 or '70, somewhere thereabout, in Stockholm. And the —
The publication is 1971, but I think it was actually —
'70. Yeah, '70.
Yeah, 1970 it must have been.
And Steve is very interesting. At that time already I think of the — I told you Fritz Mueller started putting water vapor feedback. He started getting crazy results. And Steve and Laplass [?] Rasool, R-a-s-o-o-l, he [???] the Institute of Space Study at that time. Steve was there too. And they worked together and they got much lower sensitivity due to CO2 increase, but they all have some kind of back of envelope calculations. Not by radiative convective equilibrium. And I remember inviting Steve over here to GFDL. See, why we get the result which is different. And after that Steve wrote the paper which is titled " CO2Confusion." One of these confusions, many of them are coming essentially from this similar, somewhat related to this trouble of Aria estimate, and in that confusion paper Steve tried to articulate why some result this way and so forth, and we smoked out most of the reasons why his results are different from mine. As a result he was — He also had me to co-author that paper, and I said, "Steve, you wrote it. Why don't you write it there?" So that confusion paper comes out. And shortly thereafter this, I don't know thereafter or not, but not too sure exactly. The confusion paper may be after that, this big meeting. Anyway, he was a very young guy and had just come out in the plasma physics field or something coming into this field. And so he started around that time, and study of man's impact and I was the only one who was doing the GCM or even radiative convective model at that time, so I have a free hand writing. [???] gets upset because I tried to put his result in, simulation result into it. And then they credit me for producing that work that [???] produced. But other than that, it's pretty well written actually. I was doing on the jet lag, day and night, and so that's how. And I think that is first comprehensive report. We even talked about the effect of the aerosols. And it's really, really nicely done report.
Now another thing that that report calls for like let's see the study of critical environmental problems, [???], which was the year before and organized by the same guy, Karo whatever his name is.
Uh-huh. Did it at MIT. Yes.
Both of them say, and say a considerable amount about this, that one of the things that's very important is a better network of global monitoring of environmental things. Do you remember discussions about that? Was that —?
No. The way they did was you know they had this talk by various people. And after that everybody went to their homes and then they look at it. You know, in order to write a book like in four or five days, imagine that thing, and I was writing entire chapter by myself. I was young and broken English, but I guess other people straightened my English out I guess, and writing, writing, writing. So very little mutual discussion.
I wrote my section and the Charlesons [?] and — Not, not Charleson. Little guy named John Toomey who wrote the aerosol section, which is still cited a lot. And so forth. So at that time the people begin to worry about this kind of thing may be serious, and Karo is School of Management at MIT he is a professor, and at that time he was enamored with a thing like collaborative [?] loams [?] and so forth.
Yes, right. He's actually a member of the [???].
Yeah. And they tried to do this idealistic model which almost invariably end up with catastrophe.
Right? And so they sort of, many of the estimates at that time probably how CO2 are going to increase in the future, and probably overestimate. And even though some people talk about the aerosol effect, like John Toomey, nobody pay attention to it. It's written in the report, but they never — And then shortly thereafter you know what happened is what happened in this energy crisis. And suddenly consumption of oil started coming down and so forth, and so, and the middle, around first half of 1970 people begin to think these increases of carbon dioxide which is projected by SMEC report may never be realized.
Oh, of course, yeah.
Yeah. Oil crisis dumped to this now on these things about global warming. But then around the middle 1970 this guy Ramanathan , who is a student of Stony Brook, State University of New York, and student of Bob Sess [?], and Ramanathan begin to, his Ph.D. thesis was "Effect of CFC on the Greenhouse Warming." CFC have a powerful Greenhouse Effect which he discovered using radiative convective model. Yeah, yeah, yeah. You know, now by then it's become not only planetary atmosphere but also everybody has begun, a lot of people in NCAR using it, their own version of course. Not exactly the same radiation code, but it's — everybody started doing this. But Ramanathan is good enough to realize this part [???] tritium [?] concentration of CFC can be important for Greenhouse Warming. And then around that time Jim Hanson [?] and other people begin to realize not only CFC, there are other Greenhouse gases other than CO2 and H2O which we discussed earlier. Methane, nitrous oxide, and all these things, when you [???] up, even though CO2 increase may not be as large as which we originally projected, but when you add all the other Greenhouse gases maybe actually as large, so we should still be concerned about it. So that was sort of towards the end of 1970 people started worrying about it again. And so there is various international conferences here and there and so forth. And still there was no concerted effort arguing against Greenhouse Warming, like [???] or anything like that.
That happened I think around toward the end of 1980 I guess, and but anyway this Greenhouse, other Greenhouse gases, trace gases, could be adapt to CO2 so that this could be still as much a problem as the SMEC people thought it would.
And then people — and about that time various groups started doing these various models, relatively simple ones, atmosphere coupled with mixed layer oceans and which we pioneered using it I think for CO2 problem. 1979 or '80 or somewhere thereabout. And then as a group start using it, GISP [?] group, NCAR groups, and all these groups started using atmosphere coupled with mixed layer ocean. All seemed to get 4 degrees Centigrade increase of temperature equilibrium response for doubling of CO2. So everybody said, "Oh my God, this is large. This is large." And furthermore, all these modelers agree in terms of magnitude. Our model also came out to be other than what we had before. Temporarily it went up to 4 degrees. Okay. So everybody got, around that time, about 4 degrees. But when you analyze them carefully, they are all got 4 degrees for different reasons. One had one hemisphere much larger, another had another hemisphere. So if you look at detail [???] disagreement is just pure superficial global mean temperature happen to agree. Everything else is different. But you know, so around that time it's getting, everybody thought it's a very, very large effect. When was that? Somewhere around toward the end of 1980 I think. And then big drought of 1988 came, which is I'm sure purely due to natural variability of climate, nothing to do with Greenhouse Warming, but big single cause. So we testify in the Congress, and Jim Hanson [?] made a big statement about 99 percent certain that this is a Greenhouse Warming and so forth.
And 1992 or something, '90 or something, there is LEO [?] conference?
'92. But then in 1990 you have to publish IPCC Report. And by then, by the time LEO conference comes, opponents of Greenhouse gas theories or whatever start lining up these very articulate people, try to argue against us. And furthermore we begin to realize that even though our models seem to get larger sensitivity, the warming in 20th century never as large as what model projects. And then between 1990 report and the 1995 report, then people begin to realize maybe this is overshooting by model. Maybe partly attributable to sulfate aerosols which cool the atmosphere. And that is the difference between 1995 and 1990 report of IPCC. So that's where we stand now, and if you put the aerosol in there, although you know there is many kinds of aerosols, so you know maybe lucky coincidence, but we seem to simulate the warming of 20th century pretty well. But so that's where I stand about the Greenhouse debate that this time. And yeah.
I want to talk to you a lot more about the IPCC.
Yeah, later. Later, yeah.
Let's do it tomorrow, because that's a long discussion.
Yeah, yeah. Now I think what happened is the first half of my career is just about covered, and so now starting from latter part of 1970, second half of my career actually begin to use coupled ocean atmosphere model for the study of Greenhouse Warming.
And or for the study of natural climate variability. And so that is more or less so that second half is a period which I began to use coupled model for the study of Greenhouse Warming.
So you sort of, the first half is characterized mostly by model building.
The second half is using models [???] interpreting natural phenomenon and trying to do some experiments.
That's right. So that's why. And but then almost every problem I found you need atmosphere coupled with the ocean, so that first Bryan helped me learn about ocean modeling, but then toward the end I started doing a lot of things on my own. So Kirk has been very generous to let me do that kind of thing. Originally we are splitting our job into below surface of ocean, it's his and mine, but you just can't do it that way. Each other had to close to each other's territory in order to be effective in studying coupled system. And but because you will find out that Kirk Bryan is a very gracious person, and let's you collaborate in any way you like to. Of course he is a human being, he has an ego and everything, but he's a very, very nice guy, and so is able to really enjoy the sort of climate variability and climate change by using coupled system. And one of the gimmicks I used is this flux adjustment. Oh, which I got enormous critique about it. I could talk about it. Again, it's a matter of modeling philosophy whether you want to do it. In order to get realistic equilibrium state prematurely, you can use the flux adjustment or you could tune like the NCAR people are now doing, tune each component of model so that you achieve the same thing. Tune the parameterization to achieve the right you know realistic surface conditions. Okay. But another way is admit your model has an imbalance here. Okay. Counterbalance imbalance by putting heat flux in. Okay. That's what I'm doing. So the question is whether you want to do it —
The complicated way or the easy way.
Yeah. Whether you want to do it by tuning, but the danger with tuning is they overcompensate for the shortcomings of other components of model. Okay? So that even though you get the right balance, that doesn't mean your tuning is right.
Okay? So on the other hand, if you are doing flux adjustment you do best to guess bodies or use parameter and then stay with it, and if still there is an imbalance you just put flux adjustment. And that's a difference in the philosophies.
Right. Okay. Well maybe this is a good place to stop for today.
Because we've done a lot, and [???].