Page images
PDF
EPUB

ADDITIONAL QUESTION FOR THE RECORD

JOINTLY FOR DR. WATSON AND DR. NIERENBERG

Please provide a coordinated, joint response to the Committee for the following question. If agreement on specific issues cannot be reached, you may provide a single response which clearly identifies differing points of view.

During the course of the hearing, the point was made that the lifetime for CO2 is complex in nature and may be composed of several time constants ranging from 2 years to 500 years.

b.

d.

Recognizing that a single time constant may misrepresent the actual case, what fraction of CO2 that will remain in the atmosphere after say, 5 years, 50 years, and 500 years? How does this depend on the growth rate for emissions?

What was the consensus CO2 lifetime used in the IPCC assessments? Are there models or data sets which may argue strongly for a different basis?

Of the amount of anthropogenic CO2 put into the atmosphere today, what fraction will essentially never go away (i.e. be there 1000 years from now)?

What additional work must be done to better characterize the CO2 lifetime?

What fraction of long residence CO2 (say, that remaining after 500 years) would suggest that actions be taken now rather than a "wait and see" policy?

EXECUTIVE OFFICE OF THE PRESIDENT
OFFICE OF SCIENCE AND TECHNOLOGY POLICY
WASHINGTON, D.C. 20500

April 3, 1996

The Honorable George E. Brown, Jr.

Ranking Democratic Member
Committee on Science

U.S. House of Representatives
Washington, DC 20515-6301

Dear Representative Brown:

Thank you for inviting me to participate in the Committee's recent hearing on global change modeling. Enclosed are answers to the additional questions you sent me to clarify certain points raised in the hearing and to develop additional information for the Committee's use.

Please let me know if I can be of further assistance.

Sincerely,

Robert T. Watson (RB)

Robert T. Watson

Associate Director for Environment

Enclosure

ADDITIONAL QUESTIONS FOR THE RECORD
FOR DR. WATSON

1. During the course of the hearing, Dr. Michaels provided a chart which purported to show that a key model used by the IPCC did not provide an adequate representation of temperature data in the 5,000-30,000 ft. layer gathered by satellites over the past 20 years. Please provide your own views on the methodology and interpretation used in this chart and its significance, if any, to the validity of the IPCC models.

Answer:

The chart which Dr. Michaels included in his testimony as Figure 1 compares two quite different measurements of temperature (surface observations and satellite observations of the middle troposphere). The differences between the two records provide useful scientific information and also illustrate some basic shortcomings in the logic of Dr. Michaels in expecting the records to be identical, or to be directly comparable to the model results of Manabe.

There are three problems: (i) a seventeen year record cannot be used to derive a long-term trend in the Earth's temperature because the temperature fluctuates too much on such a short time scale due to a variety of natural phenomena; (ii) the observations cannot be compared to a model that does not include all natural and anthropogenic phenomena that affect temperature; and (iii) trends in surface temperature may be different from trends in mid-troposphere temperature.

A seventeen year record cannot be used to derive a long-term trend in the Earth's temperature because the temperature fluctuates too much on such a short time scale due to a variety of natural phenomena;

The Microwave Sounding Unit (MSU) satellite record is a measurement of the temperature of the free troposphere (generally the middle tropospheric layer from about 5,000 to 30,000 feet, but with some influence from even higher layers); Dr. Michaels curve shows the average from thousands of these measurements over the two hemispheres over about a 15 year period. It is important to recognize that the surface and satellite observational records show a response to all of the factors that have caused temperature change over this period, including not only increases in the concentrations of greenhouse gases and aerosols, but also of volcanic dust injections (particularly from the El Chichon eruption in 1983 and the Mt. Pinatubo eruption in 1991), of El Nino variations in the Pacific Ocean, of changes in stratospheric ozone concentration, and of other natural and anthropogenic influences. Like the surface record, the MSU record is quite variable and attempting to determine a long-term trend from this relatively short record is fraught with statistical uncertainties. Trend analyses can give very different results depending over what period the calculations are made and what region is selected. It is generally preferred that records be 30 years or longer and that account be taken of special events such as volcanic eruptions that may affect the record. When the mid-tropospheric temperature record is extended back to the 1960s using radiosonde observations, the longerterm trend shows a significant warming, broadly consistent with ground-based observations and roughly as expected from the newest climate model simulations that incorporate the affect of aerosols. Dr. Michaels does not seem to want to mention this support for the predicted global warming trend.

Observations cannot be compared to a model that does not include all natural and anthropogenic phenomena that affect temperature:

The observations cannot be compared to the model results used by Dr. Michaels because the model simulation assumed that the only factor affecting climate was increasing atmospheric concentrations of carbon dioxide (assumed to be 1% per year). No attempt was made in the model to account for the temperature influences of volcanic eruptions, changes in anthropogenic aerosols, or lower stratospheric ozone depletion during this period, all of which

would be expected to have had cooling influences on the troposphere. Dr. Michaels had to adjust the model results for a number of reasons: (i) the original simulation was not for any particular year, (ii) the actual rate of increase of carbon dioxide is less than that assumed in the simulation; and (iii) other greenhouse gases were not simulated. The approach used by Dr. Michaels to adjust the model results is highly questionable. Finally, given the natural variability of the climate, Dr. Michaels should know that comparing the results for a particular observational period and a particular model simulation is not statistically appropriate. It is essential, if a comparison is to be made, for there to have been an ensemble of simulations, each including the variety of factors influencing the global temperature.

Trends in surface temperature may be different from trends in mid-troposphere temperature:

Dr. Michaels next contends that the surface temperature trends should be the same as the MSU trends. However, before noting a number of reasons for why these records may not be identical I would note that the trends deduced from satellite data over land are not too dissimilar from the trends deduced from land-surface observations. The lack of an observable trend in the satellite record is because there is a warming over land areas and a cooling over the oceans. I would also like to note that if the affects of El-Nino and the Mt. Pinatubo eruption are taken out of the record them even the satellite data shows a warming trend.

There are a number of reasons why these records may not be directly comparable.

First, recent analyses by Hurdle and by Jones and Santer confirm that, at particular locations, temperature variations at the surface and in the middle troposphere can be quite different. While tropospheric and surface temperatures are connected under many weather regimes, they become disconnected under the many weather regimes when a temperature inversion forms (e.g., on quiet winter nights when the surface temperature cools well below the temperature of the overlying air and throughout the subtropics where the descending air in the troposphere creates a near surface inversion). Analyses of the global patterns of these differences suggest that the differences are largest for just those circulation patterns when warming due to greenhouse gases is most likely.

Second, differences occur because sea surface temperatures are relatively stable. Under wintertime conditions over the oceans, the mid-tropospheric temperature measured by the MSU instrument can vary strongly (e.g., in the region of the Aleutian low pressure system where Pacific storms form), while the sea surface temperatures vary only fairly slowly. Thus, whereas land surface temperatures can vary more than tropospheric temperatures, sea surface temperatures will vary only a little. Along with the finding that over land, especially in the winter and on summer evenings, the surface temperature can vary much more than the air temperature, these differences mean that surface and tropospheric temperature measurements are recording quite different quantities.

Third, lower stratospheric ozone depletion is likely to have affected the mid-tropospheric temperature trends more than the surface temperate trends. The counter-arguments of Dr. Michaels on this issue are again highly questionable because the geographic region of maximum ozone depletion does not have to coincide directly with the geographic region of maximum affect on tropospheric temperatures.

I are quite confused about why Dr. Michaels plots the two curves with an offset--he somehow seems not to have adjusted to a common reference level. If this is to show that there is an offset between greenhouse gas only simulations versus the new simulations including sulfate aerosols, I agree. When the cooling influence of aerosols is accounted for, the two curves come into broad agreement (recognizing the differences in the two quantities discussed earlier).

To more properly investigate the accuracy of model simulations, I would invite Dr. Michaels to prepare a comparison of an ensemble of the newest model simulations with the extended record of surface temperature changes. It is such a comparison that has been relied upon by the IPCC and led

to their conclusion that the agreement between model results and observations is suggestive of a "discernible human influence on the climate." I concur with this IPCC finding.

2. During the course of the hearing, Dr. Michaels provided a chart which purported to show that all of the temperature change from 1965 to 1994 occurred in one year, a feature which models cannot predict. Dr. Michaels implied that the inability of models to predict such behavior called into question their use for policy purposes. Please provide your own views on the methodology and interpretation used in this chart and its significance, if any, to the validity of climate models.

Answer:

Both the simple climate models and the General Circulation Models used by IPCC are suitable for policy formulation. The fact that they do not simulate every bump and wiggle in the observational record is not surprising given they do not attempt to simulate every natural phenomena that affects the Earth's climate on short time scales.

First, I want to note that the record that Dr. Michaels showed was of the radiosonde record, which is a measure of the free troposphere temperature. As indicated in my answer to the previous question, this record may not be exactly the same as the surface temperature record and it has likely been affected by the cooling influence of the lower stratosphere ozone depletion suppressing warming in the last part of the record. With respect to the statistical analysis, as Dr. Michaels frequently points out, there is considerable natural variability from one year to the next, hence one needs to be very careful of creating break points at particular time periods, because this can distort the interpretation of the record. Interestingly, this is precisely what Dr. Michaels has done in his analysis of the chart in question. What is interesting is that the temperatures stayed elevatedunlike natural fluctuations, the average did not return to its pre-jump level, hinting that this is likely a human-induced warming.

Given that many factors influence the Earth's temperature, and most models do not attempt to simulate all of them, temperature discontinuities should not be unexpected. Observations suggest that major changes in circulation can occur over short periods and persist for some years. Thus, the climatic regime can shift into drought or high precipitation conditions in California, the Sahel, and other regions and last for many years or more. There are suggestions that these shifts result mainly from changes in ocean circulation, which can shift rather dramatically over short periods (e.g., EI Nino events in the Pacific Ocean and interruption of the thermohaline circulation in the North Atlantic Ocean). Therefore, at certain times a number of factors (e.g., onset of an ENSO event, recovery for a volcanic eruption, etc.) could all affect temperature in the same direction. These factors would be superimposed on the long-term warming trend caused by greenhouse gases. It is true that climate models that include only the atmosphere do not reproduce large interannual shifts in the climate, but the new models that include a fully dynamic ocean representation are showing rather significant shifts in the climate are possible over short periods. That the natural climate can seemingly relatively easily shift modes should not, however, make us complacent, rather, we should be very concerned that human influences on the climate might lead to shifts in the climate of the future into new modes to which society is not well adapted.

3. You testified that the IPCC assessment finds that aggregate global food production under projected climate change conditions should be able to keep pace with population growth and nutritional needs. In making this pʊjection, how does the IPCC take into account the increasing air pollution in many developing countries, soil erosion and degradation, competing demands for land and water, growing populations, the need for fertilizers, sea level rise onto productive river delta lands, and other factors that may limit the availability of agriculture to shift or increase production? What is the range of uncertainty of this projection and what are the chances that the actual consequences could be much worse?

« PreviousContinue »