Page images
PDF
EPUB

atmosphere radiation budget in order to see how it changes in response to the observed changes in global mean temperature, thus yielding an observed value for the sensitivity. A preliminary attempt along these lines was made over a year ago by M.-D. Chou of NASA/Goddard, and he found a sensitiivity that was about a tenth of what models suggested. However, the situation is not that simple. Chou used ERBE data and considered averages over only the tropical Pacific.

As I noted in a paper to appear in the Proceedings of the National Academy actual measurements of sensitivity would have to consider global means, and even then, the sensitivity will depend not only on the change in global mean temperature but also on its pattern. Nevertheless, I suggest that one should, at least, be able to obtain an upper bound on sensitivity. Moreover, one will inevitably get fundamental diagnostic quantities for testing models. With respect to the last item, C.Covey of DOE's Lawrence Livermore Laboratory has performed a preliminary comparison of model outputs with Chou's observations, and has confirmed the model tendency to overestimate sensitivity. At the moment, the main problem in going ahead with a better analysis is the absence of suitable data on surface emissivity over land. There is some possibility that data exists from which this may be extracted, but it is uncertain as to whether accuracy will be sufficient. In any event, there will be a small meeting in April at MIT where a few scientists from NASA, NOAA, DOE, JPL and various universities will gather to see what data is currently available to make such a study and to critically assess the proposed methodology. I am hopeful that the effort will succeed, but if it does not, we will know more exactly where we must improve our observations.

I should add that according to the work by Chou and by myself, the crucial factor in climate sensitivity seems to be the behavior of atmospheric water vapor (far and away the most important greenhouse gas, and completely natural) in dry cloud free regions where a 4% change in relative humidity leads to a change in the radiative flux of 4 Watts per square meter. Until very recently, our knowledge of the behavior of water vapor was severely restricted by the absence of reliable data; uncertainties and errors exceeded 20%. However, in the last two years a number of things have changed this situation significantly. The instruments used on meteorological balloons have improved as has their calibration algorithm. NASA is obtaining upper tropospheric humidities from SAGE II limb sounders, and the Department of Defense's 183 GHz microwave sounder on SSM/T-2 is allowing Roy Spencer and Dan Braswell at NASA/Marshall to prepare beautiful daily maps of water vapor over the whole earth. What we see is differing in important ways from what models are producing. A crucial aspect of the difference was identified by Sun and Held at NOAA's Geophysical Fluid Dynamics Laboratory where they found that the variations in water vapor at upper levels and at the surface were far more tightly coupled in models than they are in nature. This almost certainly points to numerical problems. Nevertheless, the same huge global programs which argue for the inclusion of ecologists, economists and physicians rarely point to the far more fundamental need for applied mathematicians to put the models on a sounder basis. Fortunately, this does not mean that mathematicians will not involve themselves; the problem is challenging enough to attract attention even without programs.

Turning to the second scientific item I mentioned at the beginning of my testimony, the fundamental problem here is to account for the major changes in climate that are known to have occurred in the past. Were such changes necessarily associated with changes in net radiative forcing, or are there basic mechanisms whereby such changes can occur independent of any net radiative forcing or of sensitivity to such forcing? In 1990, I published a paper in the Bulletin of the American Meteorological Society which pointed out that the atmosphere was very inhomogeneous with respect to its main greenhouse gas, water vapor, and that the earth's surface did not cool primarily via radiation but rather via evaporation and motion. The motions act to carry heat both poleward and upward where diminished water vapor permits heat to escape more readily via radiation. The obvious consequence of this is that if we do not accurately model the dynamic heat transport, we cannot calculate the mean temperature of the earth. No one in the atmospheric sciences would argue with this; it is absolutely basic. Rather,

members of the modelling community have argued that the models do well with such transports, and that there is no major problem here. However, extensive model intercomparisons conducted through DOE's AMIP program have shown wide differences among models and between models and observations. These differences also represent uncertainties and errors greatly in excess of the contributions from doubled CO2.

A consequence of the mean temperature depending on dynamic transport is that there might be climate change in the absence of mean forcing. Motions depend on horizontal variations in heating rather than mean heating, and such variations occur for a variety of reasons ranging from ENSO events (dependent on the interaction of the atmosphere and the oceans) to variations in the earth's orbit. The motions responsible for carrying heat consist in a large scale circulation in the tropics, known as the Hadley Cell, and transient eddies in the extratropics. In a pair of papers in 1988 and 1992, A. Hou (of NASA/Goddard) and I established that changing positions and patterns of heating could greatly alter the intensity of the Hadley Cell; it was also noted that the Hadley Cell could be a major source for the extratropical eddies. Since then, W. Pan and I have established that orbital variations strongly modulate the Hadley Cell providing a possible link between orbital variations and ice ages. A. Hou and E. Chang at MIT have established that variations in Hadley intensity can alter polar temperatures, and Hou has dramatically confirmed this with NASA/Goddard analyzed data. The alteration of Hadley intensities by ENSO events has been established, and the relation of Hadley intensity to the intensity of extratropical planetary scale eddies has been observed by Hou as well as by Chen and van den Dool at NOAA. J.M. Wallace and his students at the University of Washington are finding that significant parts of observed global warming may, in fact, be associated with ENSO patterns. In a recent paper, I have shown that the mixing by extratropical eddies strongly conditions the response of the atmosphere to stationary forcing (as is provided by major elevations like the Himalayas and land-sea differences) which, in turn, determines storm paths. Strong evidence exists that existing models are failing to replicate this behavior, and efforts are beginning at NASA/Goddard to see how this can be remedied. P. Stone, at MIT, has quantified the failure of models to mix properly, and C. Giannitsis at MIT has developed a possible simple diagnostic by showing that mixing determines the position of a major circulation feature, the subtropical jet. In brief we now can be quite certain that the atmosphere (especially when coupled to the oceans) can undergo significant variations in mean climate even without external forcing. In ascertaining this, important possibilities have emerged for improving models for both climate and weather.

What can we conclude from the above. Although I have only focussed on two basic questions which I am intimately involved with, it is evident from these examples that significant advances in our understanding of Global Change are occurring with substantial, and generally unplanned, cooperation among a variety of scientists. It is equally clear that much of the progress is occurring quietly in areas that are not amenable to easy popularization. Most important, I hope to have demonstrated that the scientific community has the capacity to focus in prioritized manner on the basic science of Global Change regardless of whether the large planning bodies do so. The main contribution that the government can make to this science is to maintain the health and integrity of the scientific community - a task which is distinct, in many ways, from the maintenance of specific programs. A healthy scientific community can make progress even without major programs, though in some cases the resources required will benefit from larger efforts. Large programs without a healthy and creative community are almost certain to prove wasteful, and program planning from above by individuals not closely and personally involved in successful research is likely to miss those seemingly esoteric details that form the foundations of science. It can never be forgotten that instruments and programs do not answer questions; individual scientists do.

Page 4 of 4

Chairman WALKER. Dr. Balling? And do I understand that you need the screen as well?

Dr. BALLING. That's correct, sir.
Chairman WALKER. Okay.

Dr. MACCRACKEN. And I will, too, sir.

Chairman WALKER. Okay. All right. Everybody needs the screen, so I'll put the screen down and leave it down this time.

STATEMENT OF DR. ROBERT BALLING, OFFICE OF CLIMATOLOGY, ARIZONA STATE UNIVERSITY, TEMPE, ARIZONA Dr. BALLING. Well, thank you very much.

What we're looking at on the screen is the latest model run from the United Kingdom. The red line shows what happens to planetary temperature if we disturb the climate by only increasing carbon dioxide. The orange line represents what happens if we disturb the system by increasing carbon dioxide and sulfates. And the yellow line represents the thermometer network readings from around the planet.

When this was introduced in Berlin one year ago, it appeared that the climate debate was solved. It did seem that the model was replicating the climate system with some accuracy.

Those of us who have been skeptical on this issue I think were immediately charged with explaining to the world why the model now does replicate reality, and that's something we had in fact been rooting for for some time.

And I would argue today that that model run still leaves us with so many uncertainties, that it's not the least bit clear where we're headed to climatically.

This is a depiction I'm sure we'll see in a minute or two from Dr. Christy. It shows the satellite record which I believe is the most accurate record of planetary temperature available. It begins in January, 1979, extends to the present, or near present. And over the entire period of record, we do see that the satellite, looking at the low atmosphere, despite all the publicity, has shown statistically significant cooling.

If we then look at the past 20 years of the satellite records shown in blue, the thermometer records shown in yellow, and compare that with the model run from the United Kingdom, one could make something of an argument that the gap is actually widening.

Despite all of the publicity associated with a model run that replicates reality, reality in fact has been cooling and the model run that we see that apparently was going to replicate reality becomes even more distant from the satellite record.

There's no question that our scientific colleagues today are debating why the surface network shows this warming, but the satellite record appears to show this cooling.

We know that over the past 17 years, that the numerical model suggests we should have seen something on the order of 0.3 degrees per decade. We know that the thermometer network is something close to that.

And yet, if we look at the satellite date labelled MSU, or the radiosonde data which come from balloon launches around the world, we get cooling. And of course, this is a big part of the debate, why

these different data sets show different patterns that should be so clear.

Well, my geographic training also allows me to ask questions about where this warming should be located. Models throughout the world tell us that the bulk of the warming should be in the northern hemisphere, in the Arctic. Even a very simple model we run at Arizona State University makes this same prediction.

And so we can go look to see what's happening in the Arctic, and if we take that same satellite data from Spenser and Christy and look only at the Arctic, we again get statistically significant cooling. I've stratified the data set by season. We're told that the winter season is the time when we should get the greatest warming. And yet, there is cooling in the winter season as well.

If we go back and look at the thermometer data that would be available for the Arctic for the past 50 years, we again see that there's absolutely no warming in the Arctic. This is quite troubling when this is the part of the planet where we think we should see the greatest amount of warming at the present time.

I've also been involved in determining trends in temperature at different locations around the planet. The bluest of the blues here would represent places where the temperature has cooled by over one degree Centigrade in the past 16 years.

And what you see is that the Hudson Bay area, which the models predict to have the greatest amount of warming, is in fact the place where the satellites see the greatest amount of cooling. And if we stratify the data by latitudinal bands, we really get nothing in the satellite system that looks consistent with what should be present given the model outputs.

Our group in Tempe has also been involved in a number of regional studies where we have looked at temperature records from the United States and from Europe and from South Africa and from Eastern Australia, and even the Middle East. And we do our best to gather the finest records we can, not contaminated from urban growth.

And in virtually every case, when we look at the model prediction versus reality, we continually find the lack of a linkage.

I would be the first to admit the models very poorly replicate climate patterns through time at regional scales. But once a scientist begins to pick one region after another, and continually finds little agreement between observed climate patterns and predicted patterns, you soon believe in the data and not so much in the model.

So my bottom line, I guess, is that no scientist would ever come here and say we need less data. In reality, here would be any number of data sets that could be constructed to help resolve this conflict between model predictions and observations.

Thank you very much.

[The prepared statement of Dr. Balling follows:]

Testimony Prepared for the U.S. House of Representatives Committee on Science

Hearing Date: March 6, 1996

Robert C. Balling, Jr.
Office of Climatology and
Department of Geography
Arizona State University
Tempe, Arizona 85287-1508
Phone: (602) 965-6265
FAX: (602) 965-1473

Introduction

In March, 1995, delegates in Berlin attending the First Conference of the Parties to the Framework Convention on Climate Change received a document showing the latest results from a numerical climate experiment conducted in the United Kingdom (Climate Prediction Group, 1994). One of the plots in that document showed the model output for global temperatures from 1860 to 2050 (see Figure 1). These results came from a numerical simulation that included both the warming effects of increasing greenhouse gases coupled with the cooling effects of sulfate aerosols. The plot also showed the actual planetary temperatures from 1860 to the near present as measured from thermometers around the world.

The scientists who prepared the report claimed that for the first time a climate model "has been able to replicate in broad terms the slow rise in global temperature since the middle of the last century." The obvious implication is that a model capable of simulating past conditions should be more reliable in its climate predictions for the future. By 2050, this model run shows a rise in global temperature of 1.5°C; warming in the Arctic exceeds 3°C, and a temperature increase in the United States over 1°C. The rate of warming is "probably twice the rate which some of the more sensitive ecosystems can tolerate."

« PreviousContinue »