A climate modeler with the Scripps Institution of Oceanography, he has written and lectured extensively about climate change. His interview offers an easy-to-follow guide to what science does and doesn't know about climate past, present and future. It also deals with the reliability of computer climate models, why understanding clouds is a top research priority for climatologists, and what's fueling the heated, polarized debate over global warming.

I would say yes. And "concerned" is a good word. Not alarmed, and not nonchalant. So far as we know, this is a phenomenon with a long time scale. On the other hand, we have to keep in mind that there have been surprises in the past. The ozone hole is a wonderful example. There was a theory that ozone would be slowly depleted, but the discovery that half the ozone over the Antarctic atmosphere disappeared every southern spring, that was a huge surprise. And there's a lot of recent evidence that the climate system is capable of behaving like a switch rather than a dial, and producing surprises.

What's the difference between weather and climate?

It's a very sensible question. In a nutshell, the difference between weather and climate is that weather deals with the instantaneous state of the atmosphere. If I say there will be a thunderstorm in London on Thursday afternoon, that's a statement about the weather. But climate deals with longer time scales and with averages and other statistics over space and time. So that if I say London next summer will be drier and warmer than usual, that's a statement about climate. A catchy way to put it is that climate is what you expect, and weather is what you get.

We know enough to say that climate has varied enormously, naturally, regardless of human intervention.

Climate has varied on every time scale to which we have any observational access. Ice ages come and go on time scales of tens of thousands of years, for example. And there are still unexplained shorter-term climate variations. In the United States, the dust bowl of the 1930's is an example of something that had huge societal and economic effects. As far as we know, it was just natural variability.

That might be of the order of 100 years, 200 years?

Yes, or just decades. For example, the El Nino oscillation that takes place on a time scale of a few years is a climate phenomenon. So you're right. Climate changes. It changes on all time scales. What's different between our time and our grandparents' time is that now humankind, which has been a passive spectator at this great natural pageant, has become an actor and is up on the stage. And what we--all 6 billion of us--do can affect the climate.

If we were taking this purely as a detection problem, of detecting the signal of human effects, we'd have to know what was normal in order to be able to detect an abnormal variation. How near are we to knowing what a normal climate should be?

That is an object of very intensive research, even as we speak. The question of detecting a human-induced signal above the noise of natural variability is a very active area. And scientists, as you know, are cautious in their public statements, so that when the Intergovernmental Panel on Climate Change--the IPCC--makes a consensus statement along the lines of, "The balance of evidence suggests that there is a discernable human impact on climate," that's a lot of lawyerly words, but it's meant to convey the idea that we believe we know enough about natural variability to have been able to detect the probable cause of some climate variations in human activities.

But the problem of detection...let's take the variable of temperature. What does the instrumental record tell us about surface temperature? Because this would be the thing we naturally turn to.

It is the natural first variable that occurs to most people, and to scientists as well. We have to keep in mind that climate is very multivalued, that it's not just temperature; it's lots of other variables. But if we stick to temperature, then we know, for example, that over roughly the twentieth century, from the late nineteenth century until today, the climate has changed. The average global temperature, if we restrict ourselves to just that one number, is about a degree Fahrenheit, or half a degree Celsius, more in late twentieth century than it was in late nineteenth century. That's a warming which, by itself, could be natural variability.

It doesn't seem like much, on the face of it, half a degree, because our annual variation is very high, ninety-five degrees, maybe.

People say that all the time. They say, "I moved from Denver to San Diego, and my temperature on average went up a whole lot more than that, and doesn't seem to hurt me at all. In fact, I rather like it." But when you use numbers like that on global average, then smallish numbers have large significance. The difference between an ice age and an interglacial period might be only a few degrees Celsius. And furthermore, we're not saying that half a degree is a serious climate change, in the sense that it's made the world harder to live in. We're simply saying that if the forecasts of climate change over the next century are anywhere near correct --those are forecasts for climate changes of two or three or four degrees Celsius warming--then that does begin to represent a very serious climate change, not because you're a little bit warmer but because that degree of average temperature change has large consequences for things like sea level rise and possibly many other climate variables.

Let's turn to the notion of the greenhouse effect. This idea occurred to scientists in the nineteenth century, about the special optical properties of certain trace gases like carbon dioxide. Could you tell us, in very general terms, what it is, how it's a natural phenonmenon.

Well, you've hit on the important distinction right there. There is a natural greenhouse effect. And it comes about because sunlight penetrates the earth's atmosphere relatively readily. Some of it is absorbed in the atmosphere. Some of it's reflected away. But something like roughly half the sunlight that strikes the earth is neither reflected away nor absorbed in the atmosphere, and it penetrates to the surface of the earth and is absorbed there. And then the earth, like all bodies in the universe, re-radiates that energy. But it re-radiates it at longer wavelengths. The sun's energy is peaked in the visible part of the spectrum, the part our eyes can see. That's because the sun is very hot. The earth is much colder than the sun, and the laws of physics tell us that it will radiate not in the same wavelengths that it received the energy, but in longer wavelengths. And in fact, the earth's radiation is primarily in the infrared part of the spectrum. And those same gases in the atmosphere that are relatively transparent to the incoming sunlight are partially opaque to outgoing radiation from the earth. Gases that are important are, first of all, water vapor, and second, a whole suite of gases that are present in the atmosphere in small concentrations. They include carbon dioxide, methane, nitrous oxide, ozone, the chlorofluorocarbons, and quite a few other gases. And because those gases absorb the infrared radiation and re-radiate both upward to space and downward toward the earth, they keep the earth warmer than it would otherwise be. In round numbers, we know that the natural greenhouse effect keeps the earth about thirty-three degrees Celsius warmer than it would be in the absence of an atmosphere. The moon, for example, is much colder on average than the earth, because it has no atmosphere and no greenhouse effect.

So there's nothing mysterious about that, and certainly nothing alarming about it. It's as real as gravity, and it keeps the planet habitable. We'd have an average surface temperature on the earth below the freezing point of water, were it not for the greenhouse effect.

The concern is that we human beings are modifying that greenhouse effect by adding to the atmosphere gases that increase the natural abundance of these so-called gases, the ones that absorb the infrared radiation. We're adding them through lots of processes, the most important single one of which is burning fossil fuel (coal and oil and natural gas), which releases carbon dioxide.

It's interesting that the main gases in the atmosphere (oxygen and nitrogen) don't have this effect.

That's correct. Oxygen and nitrogen together make up about 99 percent of the dry atmosphere. It's useful to think about the dry atmosphere as a special case, and the real atmosphere as being the dry atmosphere plus water vapor, because water's very variable in concentration. It's plentiful in some places, rare in others. But if you think about the dry atmosphere, nitrogen and oxygen make up most of it, almost all of it, around 99 percent. And they don't have any effect, to speak of, on either incoming solar radiation or outgoing infrared radiation from the earth.

There's a complex physical reason for that, but in a nutshell, it's because the molecules of nitrogen and oxygen have two atoms per molecule. We write oxygen, for example, as O₂, symbolizing that each molecule has two atoms joined together. But the greenhouse gases have three or more atoms. Water vapor for example, H₂O, two hydrogens and an oxygen. Carbon dioxide, CO₂. That's one carbon, two oxygens. And the rules of physics tell us that to absorb energy under the temperature and pressure conditions of the atmosphere, a molecule needs three or more atoms. And that's why it's nitrogen and oxygen that are essentially passive in this role.

But these gases like CO₂ are very tiny quantities. How can trace gases produce a real effect?

The usual answer I give to that question is: You don't have very many germs of horrible diseases in your body, but it doesn't take very many to kill you. The notion that a gas's effect on climate ought somehow to have to do with whether it's abundant or not compared with, say, oxygen and nitrogen--which no gas is--is a false rule. These gases, although they're present in tiny concentrations--carbon dioxide right now on average somewhere around 370 parts per million by volume. That means, if you took a million randomly chosen molecules out of the atmosphere, only 370 of them would be carbon dioxide. But because each molecule is a powerful absorber of radiation in a particular part of the spectrum that depends on that molecule's structure, they can absorb sufficient energy.

And, as I said, a good example is the natural greenhouse effect. Water vapor, carbon dioxide, the other gases that are naturally present keep the climate warmer, much warmer, several tens of degrees warmer on the Celsius scale, than it would otherwise be. And there's no doubt about that. That's not controversial at all. You can't find any greenhouse skeptic who will tell you that there's no natural greenhouse effect.

And, of course, if we wanted just an existence proof of a planet with a lot of CO₂, we could look at our neighbor, couldn't we?

Yes. The planet Venus has a lot of CO₂. You can't find an exact match to earth, because Venus and Mars, which are the nearest planets, are both different in many respects, including distance from the sun. My nearest heavenly body for climate comparisons is the moon, because it's about the same distance from the sun, on average, as the earth. It gets its energy from the sun, as does the earth. But because it has no ocean and no atmosphere, hence no water vapor or any other gases, it has a very different climate. It's much colder on average, and it has a very harsh day-night difference because there's no ocean or atmosphere to buffer the big swings between sunlight and darkness.

Before we started burning fossil fuels very quickly, we had this natural greenhouse effect. Since we've been burning it, what is the evidence that this is ending up in the atmosphere? Because there are other places it could go, aren't there?

There are other places the carbon dioxide could go, and it does go to those other places. Again, in round numbers, something like half the carbon dioxide that's emitted out of our chimneys and tailpipes when we burn coal and oil and gas...about half of that ends up in the atmosphere. There's a very complicated set of pathways, through the atmosphere, the ocean, and the biosphere that carbon, in the form of carbon dioxide and other forms, travels. But roughly half of it ends up in the atmosphere.

And we have very accurate measurements of atmospheric carbon dioxide for about four decades, dating from the late 1950s. The scientific hero of this is at Scripps Institution of Oceanography at the University of California, San Diego. His name is Charles David Keeling. And it was Keeling who saw the importance of measuring carbon dioxide, devised the instrumentation, and invented and built it himself--it hadn't existed before--to measure these concentrations to very high accuracy. And he has, through incredible perseverance, single-handedly at first, and later in conjunction with many other people, made this unbroken record of the carbon dioxide content of the atmosphere. It's gone up, in round numbers, from around 315 parts per million, when Keeling started measuring it, to around 370 parts per million, on average, today. And those numbers again are totally noncontroversial. Everyone accepts them.

It's going up not only because we're burning more per year, but because it's staying there. Is that right?

That's right. It accumulates in the atmosphere. Part of what's emitted into the atmosphere ends up in the ocean. Part ends up in trees on land. There are many pathways and many sinks. And parts of the so-called carbon budget are still the object of very active research. But of the part that ends up in the atmosphere, we have a rather clear-cut record. And the concentrations just keep going up and up and up.

And some of these molecules are going to hang around for quite a while, the residence time?

That's right. Each greenhouse gas has a different residence time, that is, a different characteristic time for which it stays in the atmosphere. And that has to do with the speed of the processes that generate it and that take it away. In other words, it has to do with the nature of the sources and sinks. And for carbon dioxide, the typical residence time is centuries. For some greenhouse gases, like ozone, it can be much shorter--a day or less, for example.

Let's move on to how CO₂ or greenhouse gas is just one factor in this thing. Talk about the idea of forcings. When climatologists talk about climate forcings, what are they talking about?

When you start to talk about climate, the first order of business, so to speak, is to define what you mean by it, and to define the climate system, the part of the natural universe that you think matters to it. And we've begun to think of climate as not simply the average state of the atmosphere, but the average and measures of the variability of the atmosphere and other aspects that interact with it.

For example, the ocean: It exchanges --as we've just said-- carbon dioxide and also heat and energy, momentum, water, salt; then there's the world of living things: Trees also exchange matter with the atmosphere and the ocean, ultimately; and there's the land surface: the world of ice and snow, glaciers and sea ice and snow-pack and so on. All of these are parts of the climate system.

However, there are other things outside the climate system that influence climate. A straightforward example is the sun. Virtually all the energy that matters to the climate comes, ultimately, from the sun. So if the sun can vary--and it does vary--then that variability can affect the climate. You might think of volcanoes as another example of something outside the climate system, in the sense that so far as we know, the climate doesn't affect the volcanoes in any direct way. But if a volcano spews a lot of matter into the atmosphere, that can affect the climate. We have very clear examples of that in modern times.

So, "forcings" connotes the idea of external influences on the climate, such as changes in the sun, or changes in the concentration of greenhouse gases, or changes in particulate matter in the atmosphere from volcanoes. Those are three good examples of climate forcings.

Incidentally, the potential list of climate forcings is very long, indeed. And at some point you can get into questions about whether it's a forcing or part of the climate system. So for example, if the land-surface use changes because people plow under some grasses and change the characteristics of the land surface, you can think of that as human caused, as an anthropogenic forcing. But if the climate changes so that those grasses die off naturally and are replaced by something else, then that's probably best thought of as part of the internal variability of the system.

Climatologists also talk about "feedbacks." When you have a forcing, it can react in different ways. Right?

That's right. We distinguish between forcings and feedbacks. In simplest terms, you can imagine a feedback as being something that goes on in the climate system in response to a forcing, which itself then has an effect on the climate system. Here's a straightforward example. Suppose the climate warms up. For whatever reason, the earth gets warmer. And then suppose some snow and ice melt. Under the snow and ice there are darker substances--land or ocean. And, therefore, in response to the warming, the world has become a little bit darker and, hence, less reflective, so it absorbs more sunlight.

So, you have a chain of events. You first warm the climate for whatever reason, and then the chain is: snow and ice melt, a darker surface is exposed, that darker surface absorbs more sunlight than did the snow and ice that was there before, and, therefore, the climate warms even more. It's a positive feedback. It's as though you had the thermostat in your house set so that when the house warmed up, it turned on the furnace and warmed it up still more. A negative feedback works in the opposite direction, in the way that the thermostat in your house usually works, that when the house warms up, the thermostat turns on the air conditioner and cools it back down.

Give me an example of a negative feedback in a climate system, like a cloud one.

Well, if the clouds changed in response to a warming, so as to counteract the warming, that would be a negative feedback. Here's a hypothetical one: As you may know, cloud feedbacks are among the most important collection of unknowns, things we're still doing research on. But here's a possibility that's been thought of. If, as the climate warmed, the clouds in the warmer climate, on average, contained more water than the ones in the colder climate--because the atmosphere is more humid--then those clouds would reflect way more sunlight, and that would tend to cool the climate system down. If the climate warmed and produced more clouds so that the average cloud cover of part of the earth increased, that would tend to cool the climate back down. So those are possible negative feedbacks.

Feedbacks can only amplify or diminish. They can't reverse the sign of an effect?

In general, that's right, because at least we don't have any examples of cases where the climate system is so stable, so robust against potential changes, that a change in one direction originally starts to drive the climate in the other direction. You could concoct such a thing fancifully. You could imagine it. Imagination's very powerful. You could imagine, for example, a climate in which the world warmed up a little bit and then, for whatever reason, it suddenly became totally cloud-covered, like the planet Venus. And that total cloud cover, in replacing the partial cloud cover we have today, projects a much whiter, brighter surface to incoming sunlight, and cools the earth rapidly and plunges us into an ice age. So you could think that up. But so far as we know, we haven't seen evidence for any catastrophic reverse in feedbacks like that in the system.

Climatologists also talk about inertias in the system, delays.

Climate's complicated, and one of the things that makes it complicated is that different parts of the system respond at different rates. To put it another way, parts of the climate system have much longer memories than other parts of the climate system. And so they provide a kind of inertia, a sort of buffer or flywheel, that means that the effect of a forcing may take time to be felt.

Here's an example: If the climate responds by first warming up the ocean in response to, say, increased sunlight or increased greenhouse gases, then the ocean contains a lot of water, the water has a high heat capacity, and it takes a long time to warm up the ocean by just slightly increasing the amount of radiation it receives. If only the upper ocean is involved, it might take decades. If the entire depth of the ocean is involved, it might take centuries. And so there can be a time lag between when you force the climate system--when you make an external influence on it-- and when you see the response to that climate system.

Also there are things that you sometimes refer to as oscillations within it, which is a kicking back and forth, to do with ice and oceans.

There seem to be some natural rhythms to the climate system. I hesitate to say "cycles" because I don't want to convey the idea that the climate is in any sense completely regular like the pendulum on a clock. But, for example, ice ages tend to come and go, not at exact periodic intervals but at characteristic intervals: tens of thousands of years. They seem to be paced by changes in the earth's orbit. That's a theory that's been around for quite a while, and there's a lot of recent support for it. But that pacing by itself may not be enough to induce the ice age. The changes in the earth's orbit, which take place over thousands of years, slightly alter the distribution of sunlight with season and with geography. But it seems likely that, although those may provide pacing items, some of the feedbacks in the climate system are necessary to switch it from one mode to the next.

Another example is El Niño. On shorter time scales, El Niños come and go every, perhaps, three, five, seven years--something like that. But we've got enough records to know that sometimes it's longer than that and sometimes it's shorter than that. Some El Niños are strong and some are weak, and in fact no two are alike. The Indian monsoon, which is important to the livelihood of millions and millions of people, shows up more or less every summer. But it shows up with different intensities and at different times, and those changes have huge societal and economic consequences.

There are lots of cycles. One could go on and talk about them all the time.

But these generally flick both ways. Is that the idea?

That's right. The idea is that there are natural oscillations. What you might think of the climate system as is a very complicated bell. And if you strike a bell with a hammer, you get a resonance. It rings. And the frequency that it rings at, the sound that it makes, has to do with the way the bell's constructed. And the climate system's a bit like that too. It's got natural modes of variability. And so when you kick it, whether it's with greenhouse gases or changes in orbits, you can, so to speak, excite the climate system to ring along those modes. But they're very complicated, they interact, they're imperfectly understood, and it's probably an oversimplification to think of the climate system as just something with a certain number of limited characteristic variabilities that you can set off.

Talk me through a climate model. Try and tell me what's different from the "back of the envelope" calculations that we could have done in the late nineteenth century.

Climate's complicated, and so climate models are complicated. Einstein once said that everything should be made as simple as possible, but not more simple than that. And so we've resisted in recent years the temptation to overidealize climate to a couple of equations that you can solve with pen and paper.

The foundation of climate models today are really the models that we've taken over from the folks who do operational weather prediction. The weather forecast that comes out every day in the newspaper or the television is produced by a computer simulation in which the entire world atmosphere is the domain of calculation, and the variables involved are the standard weather variables: the winds, the temperatures, humidities, cloud cover, things like that. And for weather forecasting, those variables, all of which interact with each other in complex ways...for weather forecasting, those variables are measured so that we know today's weather, and then we use the model to project tomorrow's weather and the day after that.

And the models basically are a few equations in fluid dynamics?

It's a few equations; it's Newtonian physics, if you like. So it's equations that express fundamental physical principles: the conservation of mass, the conservation of energy, and so on. But they express complex interactions between these variables.

So, for example, if the wind is such to blow moist air into a region, clouds might form. The clouds could affect the sunlight, the sunlight affects the temperature, the temperature affects the pressure, the pressure affects the wind, and the wind affects the humidity and the clouds again. So there's lots of loops in there, and the equations take that into account. And these equations are solved on grids, which means that we give up the idea of knowing the temperature at every single point in the atmosphere and, instead, because we have only a finite amount of computer power, we represent the atmosphere by average variables at points separated by a few tens or a few hundreds of kilometers.

So you divide the atmosphere up into what? Cubes or blocks?

You can think of it as little boxes, or a lattice whose characteristic size might be fractions of a kilometer vertically, and tens or hundreds of kilometers horizontally. That's still thousands and thousands of points--the intersections of these lattices--at which we have to have the variables. And the weather forecasting gets better, in part, because as computers get faster, one can afford to put the points closer together and have more of them and resolve the weather better. When the points are very far apart, important events can essentially pass through this grid undetected, the way an insect could pass through a coarse window screen.

So we start out with an atmospheric model like that, and then we add to it and couple with it models of the ocean, models of the land surface, models of ice and snow, and, ultimately, models of bio-geochemistry--that is, models of the chemical and biological interactions with the physical system. So the model gets quite complex. And, therefore the computing requirements go up. And, therefore, also, sometimes the model gets harder to understand. You run the risk, so to speak, of buying realism at the cost of insight. The model may produce an answer that resembles the real world in some way, but the complexities of the model are such that understanding why the model behaved as it did can be almost as tough as understanding why the atmosphere behaves as it does. So these models, we think of them as one end of the spectrum of theoretical tools and observational devices that we have to help us understand the climate.

So the purpose of the model . . . is to help you think about climate?

The model in the end is a computer program. We can't take the atmosphere or the whole planet and put it in a test tube and do experiments on it. So instead, we simulate it in a computer. That's turned out to be the best way of incorporating all the various complexities. As I said, such a computer simulation or model or program can be complicated enough, so it's hard to understand. So the real way that climate study goes on is that observation studies and theoretical studies give rise to individual simpler models of individual processes:

So, for example, we can create a model of a thunderstorm and study how thunderstorms produce rain, or how clouds absorb infrared radiation or reflect solar radiation. And then the understanding gained from observational studies, from simple theories, from process models like that, gets incorporated into the comprehensive global model of the whole climate system.

Models get more realistic as we understand the physical system better. If you don't understand the basic physics of what's going on, if you don't understand what causes clouds to form--and there's much about that we don't understand--then the model can't be realistic, and solving it on even the most powerful computer with very fine spatial resolution is, in effect, just getting more and more accurate answers to the wrong equations, to equations that describe something that's a little bit different from the real physical climate system.

Skeptics criticize models. When they say models aren't good at reproducing the past...what should we expect a model to be able to do? Are they really "what if" scenarios for exploring different boundary conditions? How should we think about them?

In a way, a model is an incorporation of our best current knowledge about the climate system. And, so, you might think that an ultimate goal of this science would be to produce a completely realistic model. At that point, we will have essentially completely understood the climate system. We'll make weather forecasts as accurate as they can possible be, and so on. That goal's a long way off. That's not reason to regard present day models as useless, or to hold them in contempt. In a similar way, you might say that the goal of medical science is to cure all disease. And the fact that that's not yet been done doesn't mean you should treat your physician with disdain.

Models can do a lot. As I said, the foundation of climate models is the atmospheric model that we use in the daily weather forecast. And although meteorologists have thick skins because their predictions are sometimes wrong, and nobody ever forgets it when they are, nonetheless, weather forecasts are pretty good. They're a lot better than they were only a decade or two ago. And as the same physics that's in the weather models gets incorporated into the climate models, the success of the weather models gives us reason to, in part, trust the climate models. Their veracity and their reliability is higher because, although we can't yet wait a century to see what happens as the greenhouse effect strengthens, we've seen successes on shorter time scales--not just the daily weather forecast but predictions of El Niño, for example, which are also made with coupled ocean-atmosphere models. The ability to simulate variations in the seasons, the fact that not all summers or winters in a given part of the world are alike, is another example where we have partial success.

So the trick, I think, in climate models is to interpret them wisely, to have a feel for what parts of the model are trustworthy and what parts of the model are shaky, and, therefore, to be able to use the model results as guides to policy and really guides simply as to what to expect as climate evolves over coming decades.

I suppose the quarrel that mainstream atmospheric scientists like myself have with the people who've come to be called skeptics --the ones who essentially pooh-pooh the prospect of man-induced climate change--is that although there are uncertainties in the models--nobody knows this better, in fact, than the people who work on the models all day, every day, none of whom are the skeptics--although the models have their imperfections, it seems to me it's unlikely that every imperfection in the model is going to be one that makes the model more sensitive to greenhouse gases rather than less.

If I were skeptical about models, it seems to me a reasonable position would be: Well, climate change is going to occur, and changing the chemical composition of the atmosphere is going to affect climate; the models can't predict it perfectly; but that, to me, means that climate change might be either less severe or more severe than the models, and the models give you, at any moment, you might say, a mainstream estimate--our best guess of what the climate's going to do in the future.

You can't wait 100 years, so how do you validate a model, from a policy standpoint? What about looking at whether models reproduce the past, the instrumental record?

There's a difficulty in reproducing the past, which is that we don't know where to start things off. We don't have measurements of what the whole climate system was doing in 1890 or 1912. And so that keeps you from replicating the climate of the twentieth century perfectly. And there are other theoretical reasons for thinking you might not be able to replicate it perfectly, even if you had a good knowledge of the initial state.

For example, you don't know with perfection how all the forcings varied during the twentieth century, which volcanoes went off, and how much of what kind of matter they put where, and how the sun varied. Nonetheless, these models are able to reproduce, in broad outline, the climate of the twentieth century. That is to say, if you start them off at a reasonable guess of what things were like 100 years ago, and put in reasonable estimates of these forcings, you can produce something like the climate of the present time.

There are many other examples of partial verification. A good one is the eruption of Mt. Pinatubo in the Philippines. These very same kinds of models were used to predict that we would see a global cooling in the neighborhood of one degree Celsius, and that it would last in the neighborhood of a couple of years. And that prediction was made before the climate changed, but indeed a climate change very much like that was observed.

So there are lots of means by which we can gain confidence in the models and, at the same time, learn about the aspects of the models that still need improvement. So neither I nor any other climate modeler would accept a blanket statement that says, "Models are useless, they can't reproduce the past, and we can't trust them for the future." That's just not true.

If you're trying to reproduce this model, and you want to explain why, say, between 1940 and 1970 doesn't reproduce, then you have to talk about other things like solar variance and sulfates and other forcings. And those things have to be captured if the model's to be any good. Correct?

Even if you had a perfect model, and even if you had a very good estimate of what the climate was like at the beginning of the period of interest--say, 1900--then in order to simulate the evolution of the climate over the past 100 years, you'd still need to know some things that we don't know well and may never know well, which include how the sun has varied over that time, and the production of aerosols in the atmosphere--both natural aerosols, such as from dust or volcanoes, and manmade aerosols, such as sulfates from industrial products. You'd still need all of those forcings that we have reason to think can influence the climate over decadal time scales.

And lacking that, there's a limit to how well you could even hope to predict the climate. We're still learning what that limit is. That is, it's still an object of active research to go back and say: Well, if we think the climate did this over the last century, and if a model says it did that, how far apart can this and that be for us still to trust the model? That's still active research.

But the basic concept is important to keep in mind, which is that we don't know the predictability of climate perfectly. That is, climate, like weather, can surely vary naturally on its own, independently of forcings. So you could imagine a climate system in which the sun were absolutely constant, there were no changes in the chemical composition of the atmosphere or the surface of the earth, and that climate might still vary. And it might vary in an essentially random and unknowable manner.

But quite aside from that, the part of the climate that's predictable potentially, if we knew the initial state and the forcings well enough, is still imperfectly predictable if we have errors in the initial state and the forcings, quite apart from errors in the models. So there's a whole slew of reasons why we can't expect to "hindcast" the climate of the recent past perfectly.

Other things that skeptics say...one argument might be that humankind could be producing a forcing in the opposite direction. And we've touched on this. So sulfate aerosols producing either directly scattering or indirectly through clouds, and so forth. Couldn't that, in principle, keep up with the greenhouse gases?

There's a fallacy in the concept that sulfate aerosols can counteract the greenhouse gases. Sulfate aerosols are not just little negative greenhouse gas molecules. They have very different properties. And one of the properties that's very different is that they tend to be rather short-lived, and therefore they get removed from the atmosphere--they get rained out and removed in other ways--relatively quickly. Characteristic time might be in the neighborhood of weeks. So that if you were to draw a map of the world and indicate on it where the sulfate aerosols were at any given moment, you would find them primarily in the industrial regions--if we're talking about manmade sulfates, small particles--and just downwind of those industrial regions. So you see them in Europe, you see them in North America, you see them in Japan, other areas of Eastern Asia, and the regions just downwind of those regions. You don't see them uniformly distributed.

Carbon dioxide, on the other hand, because it's in the atmosphere for centuries, is more or less uniformly distributed. The winds mix it around so that the concentration of carbon dioxide is pretty much the same everywhere. For radiative purposes, you can, essentially, assume that the concentration doesn't vary with location.

That means that to the extent that the sulfate aerosols counteract the greenhouse warming by providing, for example, a more reflective part of the atmosphere for sunlight to bounce off of, they do it only in limited regions, and therefore they complicate the picture. They don't simply counteract it.

Another argument asserted is that if the climate was much less sensitive than you're assuming--since most of the bang comes from the water vapor feedback--then there'd be much less concern.

It's quite amazing to me. I don't mind talking about skeptics, but there are a very small number of them, and I sometimes wonder why the media, in some perverse sense of fair play, seem compelled to give the same amount of air time or newspaper space to half a dozen skeptics as to thousands of scientists who would essentially agree with the consensus. But although this will contribute to that imbalance, I'm willing to talk a little bit about skeptics.

Most skeptics don't actually do research. They comment in a highly selective way on research that other people do. Their own research tends to be very limited, and limited to a very few processes. You don't get anything like a balanced view from skeptics. They tend, as a group, to approach the problem rather like lawyers, making the best case for a client who has a preconceived position, rather than like scientists, which is to examine the climate system with the idea of figuring out how nature works, not to substantiate a preconception that one comes in the door with.

So the notion that the water vapor feedback might be weaker than climate models think it is, is a perfectly reasonable subject to investigate. And the suggestion by those that water vapor feedback is poorly understood is being investigated. It has led to a great deal of research. A perfectly good outcome of skeptical science, when done by first-rate scientists like Richard Lindzen, is that it provokes other people to do research. The result so far is that there are still some unanswered questions, but the great bulk of climate scientists have not yet been converted to the view that the way the models treat that feedback is very wrong.

So the jury's still out. It's an area of uncertainty. And you know, science isn't a democracy. It doesn't take but one scientist being right. But it is very much a cooperative process. And experience shows us that in cases like this, it tends to be the exception rather than the rule where one voice crying in the wilderness turns out, in the end, to prevail over conventional wisdom.

So I can't sit here and tell you that I or anybody else is 100 percent certain that the water vapor feedback isn't over-estimated by models. I'm just saying there are a lot of reasons, and not simply prejudices and gut feelings, but a lot of research results that suggest that the models have got the water vapor feedback pretty much right.

But the water vapor really accounts for the range in predictions, doesn't it? Primarily, from the low end to the high end?

The biggest single factor in accounting for the range in predictions of climate models, and, more precisely, in accounting for the factor-of-three difference in the sensitivity of climate models to greenhouse gases--if you measure that sensitivity by global average surface temperature change--the biggest single reason for that is clouds. And so if you want to think of clouds as being part of water, because they're made of water, then that's included. But clouds aren't water vapor. Clouds are solid and liquid water. And so that's not generally thought of as part of the water vapor issue.

But we do know enough about cloud feedbacks now to know that they do dominate the models' response on global average. And so that, for example, if you were to take the cloud part of one climate model and transplant it into another climate model, you replicate the sensitivity of the donor model in many ways. So they really do have a dominant role to play. They're still poorly understood. And they are the top research priority among currently investigated topics in the physical climate system, for that reason.

So the biggest way of reducing uncertainty is to better understand clouds?

That's right. And in particular, I think that means going back to observations. For a long time, climate models contented themselves with very simple treatments of clouds, saying for example, that the cloud cover in a region would be proportional to the relative humidity there. We've now made enough measurements on real clouds to know that real clouds don't quite behave that simply, and there's lots more complications. And so some of the most exciting ongoing research today consists of going out in the field, making intensive measurements on all aspects of real clouds, correlating those with the local meteorology, and improving, thereby, the way models treat clouds.

Let's assume that the world will continue to grow in the way that it's doing, the amount of carbon emitted by the world will increase steadily, and therefore the atmospheric concentration of these gases will increase. If we took the middle range for a doubling or a trebling from the models of four or five degrees--varying from two to nine degrees Fahrenheit--would that be a big deal? What kinds of things might we expect?

I'm going to start with a comment on your question: "If we keep putting carbon dioxide... ." That's a very interesting premise. And I think not enough attention is paid to it in the climate debate, and, in particular, on the policy side of the climate debate. Because it's rather unlikely that things are going to go on exactly as they have, for a number of reasons, one of which is population.

Most people don't realize how rapid recent population growth has been. In 1999, the population of the world reaches 6 billion people, on best estimates. And in 1930, only seventy years ago, it was 2 billion. So within one biblical lifetime of seventy years, the global population has tripled. That's not happened before. And many people think it's not likely to happen again--that a

6 billion-person world is conceivable, and maybe 10 or 12, but if it tripled again to 18 billion, then for many other reasons, we'd have serious problems.

The other thing to keep in mind is that the climate system is not something that responds to simply a static human input. Humanity changes too, and technology changes rapidly on these decadal time scales that we're talking about. So it's quite possible that the world, over the time period that we're discussing--roughly the next century--will have evolved technically to something very different than it has. One has only to think back 100 years to see how little of what we take for granted in present-day world --automobiles, airplanes, pharmaceuticals, nuclear weapons--was not there at all and was imagined only by visionaries.

And so I tend, philosophically, to reject the notion of business as usual carried on indefinitely, because I find on many grounds, even though my specialty is meteorology, that it's unreasonable to think that technology will stand still and that rate of growth of population won't change.

The second question is a pure climate question. What's the climate like if it's four or five degrees Fahrenheit warmer than the present climate? And the short answer is, we don't know. There are some things that we're pretty confident of. One is, for example, sea level will rise. That is pretty straightforward physics.

... It rises basically for two reasons, both of which are well understood. One is simple thermal expansion. A given amount of water takes up more volume when it's warmer than when it's cooler. The other is that the warmer climate causes some of the snow and ice on land to melt. That's happening today, by the way. Glaciers are retreating. And so sea level rises.

It's not quite that simple, because in the short term, the hydrologic cycle speeds up. That is to say, the rate at which water is evaporated from the ocean and precipitated as rain or snow on land and sea speeds up. So that, for a while, you have a competing process. And a given glacier in, say, Antarctica, might be gaining mass because more snow falls on it at the same time that it's losing mass because icebergs are breaking off or some of it's melting. But in the long run, the rising sea level catches up and prevails.

And, in fact, we have very clear geological evidence of this. In the ice ages, sea level was much lower than today. Not just a little bit lower, but tens of meters lower. So it's quite certain that if you warm the climate and wait long enough, you get a higher sea level. That is exactly the reason why the countries who are most concerned about greenhouse climate change are low-lying island states, who feel in many cases their very existence is threatened by rising sea level. The Netherlands--

Does this happen with a doubling of CO₂?

Yes, this happens with a doubling of CO₂. Part of the trickiness comes in here when you talk about doubling, because the climatic consequences of doubling CO₂ occur some time after the actual doubling takes place. So we're talking about a lot of processes at once here, and we're using shorthand language. In the first place, when we talk about CO₂, we're talking about several greenhouse gases. In the second place, when we are talking about doubling, we're talking about not an instantaneous doubling and then waiting until the climate system comes to equilibrium, but we're talking about a gradual increase in the concentrations, at some point at which doubling is reached. And after that doubling is reached, the climatic consequences of the doubling become more and more apparent. As we said, there are time lags built into the climate system. It takes the ocean a while to warm up; it takes the ice a while to melt, and so on.

But let's talk about the effect of a doubling at some time in the future.

But the effect of the doubling, occurs later, we having committed ourselves to a certain degree of climate warming or climate change, in response to greenhouse gases already put into the system. So that even if you were to stop emitting greenhouse gases right away--today, this morning--you still have climatic change because of the greenhouse gases already added.

So, after that occurs, then sea level rises, and then a whole range of other possibilities come into play, which we are still in the process of learning about. We think, for example, that the strongest climate change will occur in the far north, because of the ice and snow feedback. Ice and snow melt as the climate warms, the reflectivity of the surface is lowered, more sunlight is absorbed, and the warming is therefore enhanced. It's a strong positive feedback locally, but it's only important where there's ice and snow that can melt. And so most climate simulations show the strongest warming occurring in the Arctic.

And part of the fingerprint, in fact, that leads us to think that the balance of evidence suggests a discernible human impact on climate already, is that we see a number of those changes occurring in the Arctic. By themselves, they may not be conclusive, but taken in together with all the other evidence, they're suggestive of the kind of climate warming that the models predict.

Climate scientists find themselves in the middle of this policy debate. On the one hand, people coming out of the White House try to present an idea that the science is in--we don't need to worry about that. On the other hand, there are skeptics arguing that the climate science is so uncertain, we shouldn't do anything.

As soon as one starts to talk about policy, I think that a responsible scientist has to become very clear about differentiating between the research results, on the one hand and the policy implications, or the role of the research in influencing policy, on the other. You, so to speak, take off the white coat and step outside the laboratory, and you mix in, together with scientific expertise on the climate system, a whole range of personal convictions, political views, in some cases even religious statements.

But with that caveat understood, it seems to me that the policy picture right now is very complicated, that you find in various parts of something like the United States government people who are quite convinced that the science is solid enough to justify immediate action. And so, for example you might think of the Kyoto Protocol as an example of that sentiment. That is a rather formidable-looking agreement in which the nations of the world have vowed to reduce the emissions of carbon dioxide to a few percent below what they were in 1990, which is a rather largish percent, below what they would otherwise be under "business as usual" scenarios by 2010 or 2012, which is the timeline.

The Kyoto Protocol expresses the idea that, yes, enough has been known, and even though we can't settle exactly what the climate consequences will be, and we can't establish exactly what the level of carbon dioxide and the other gases ought to be to avoid adverse anthropogenic effects on climate, nonetheless the precautionary principle or some combination of judgment and wisdom and political ideas leads us to say, "It's time to start slowing down, at least, the rate at which we put CO₂ into the atmosphere."

There are other people, as you know, who can be found also in the United States government, who don't think the science is anywhere near solid enough. And when they mix in with their reading of the science their own convictions about politics and economics and personal values, they fear the economic consequences of what they would term draconian cuts in emissions with consequent cuts in standard of living and in energy production. And there's the whole spectrum of views in between those, and in fact, a few outside those. And it's hard to know, for a scientist, where to come down on that.

I think, simply through talking to scientists and reading what they write, that many scientists would tend to subscribe to what have been variously described as "no regrets" or "win-win" or "insurance" policies. So that, for example, if you drive a car that gets better fuel economy--more miles to the gallon or more kilometers to the liter--then you have done a little bit toward decreasing the emissions of carbon dioxide into the atmosphere.

But even if climate turns out to be not the issue we think it is, even if climate turns out to be more robust than we presently think against human effects, then you've done a number of other pleasant things for yourself. You've saved some money, you've improved the air quality in the town where you live, you maybe have reduced the dependence of your country on foreign oil supplies, so you might have improved national security and the balance-of-payments problems. There are a lot of reasons for driving a car that gets better fuel economy. And so I think measures like energy efficiency and energy conservation are widely advocated. Switching to renewable energy where practicable is something that, seems to me, has many other things going for it besides its benefits to the climate system.

But one does come across this paradox: that people who are already convinced that the science has been done don't think more research is needed. And people who think that scientists are out not to give objective studies of how nature works but to push a preconceived idea that a climate catastrophe is looming oppose further research. And, so, for many scientists, to whom the need for further research is not simply self-serving but also obvious, because we see so clearly where the holes are in our present knowledge and where the uncertainties are in our model predictions, for us to find natural friends in the political spectrum who will share our sense that research is not only urgently required but actually rather cheap compared with the climate consequences of not doing it makes the political process bewildering and sometimes frustrating.

Final question,: Are you an optimist or a pessimist? Because given that 90 percent of our energy comes from fossil fuels at the moment, these win-win kinds of solutions are easy for people to get on board with. But many policy skeptics might say they won't make much difference in this problem--that you're accumulating the stuff in the atmosphere...very hard to avoid a doubling or a trebling. No shortage of fossil fuels. If you include the clathrates, they can go on for hundreds, thousands of years. Given that, are you an optimist or a pessimist that we will turn this around before it strikes?

I think that when you ask a scientist whether he or she is an optimist or a pessimist, now we've got no further reliance at all on our comfortable laboratory and our white coats. You're really just asking one human being how they feel. And I look at the ozone hole issue as a useful paradigm in this case.

Remember the story of the ozone hole? There were theoretical reasons in the 1970s to think that stratospheric ozone would slowly be depleted by manmade chemicals, chlorofluorocarbons. But the discovery of the ozone hole in the 1980s was serendipitous and a huge surprise. It was a gradual diminution of ozone. Half the ozone over Antarctica disappeared every southern spring.

And it was that sudden climate catastrophe (if you want to think of it as climate), that totally unexpected smoking gun, that led very rapidly to international agreements--the Montreal Protocol and subsequent ones--to the development by the chemical industry of ozone-benign substitutes for CFC's, and you might say, a very successful outcome. If our current understanding is right, the ozone layer will gradually heal itself in coming decades.

Therefore, you could imagine an optimistic scenario in which the Kyoto Protocol --which only slows down the rate at which we emit greenhouse gases, only buys us some time, you might say, doesn't solve the problem--leads to international cooperation, leads to further development of nonfossil fuel-based energy sources and buys time for a deeper understanding of the climate system and for advances in technology, and lays the foundation for international cooperation so that we reach a safe level of greenhouse gases in the atmosphere sooner rather than later. I don't know whether I expect that to happen or not. It's certainly within the realm of possibility.

Pessimistically, everyone agrees that if you keep adding greenhouse gases to the atmosphere, you eventually change the climate in a serious way. If a skeptic says that he or she's not worried about doubling, ask them about tripling or quadrupling, and ask them about all the other gases besides carbon dioxide, some of which are increasing faster than CO₂.

There comes a point where you can't escape the idea that you're having serious climatic consequences. And so the issue becomes one of guessing whether we get wise before that day, or whether technology bails us out, or whether we have to wait for some perhaps quite unanticipated climate surprise that wakes us all up. I very much hope that the optimistic scenario is the one that develops.