In the 1980's, Global Warming due to the enhanced greenhouse effect came to be perceived as a serious threat to the planet's ecological and societal sustainability. This concern was based primarily on estimates of global warming and other climate changes from numerical models of the Earth's climate system. (This perception was reinforced by a few hot, dry summers in the eastern U.S. which constituted for some people the "smoking gun" of climate change.) While the development of models is critical to our future ability to examine what we may be doing to alter the climate of the Earth, many scientists acknowledge that models are still rather simple representations of the complex processes that control the Earth's climate.
The observational evidence for enhanced greenhouse global warming is also less than clearly defined. While all surface-based global temperature data sets indicated warming of 0.3 to 0.6 degrees C since the last century, the complete source of this warming is still unknown. First, the Earth was evidently coming out of a relatively cold period in the 1800's so that warming in the past century may be part of this natural recovery. Data sparseness and reliability are somewhat suspect in the early years of the thermometer climate record and remain a concern even today when the shrinking network of stations is attempting to capture relatively small variations. Local land use changes may also have added additional warming not connected with greenhouse gases.
With this background, scientists recognized that we did not have an observing system in place with adequate means to truly monitor the health of the planet or to provide the data needed to validate and improve the models of the Earth System. One obvious limitation of information about the atmosphere was the lack of true global coverage.
2. The Microwave Sounding Unit data set
I am here to report a success story - a story that involves U.S. Government scientists and managers who collaborated closely and productively with university scientists. In 1989, to test the ability of satellites to monitor the Earth, Dr. Roy Spencer, a NASA scientist, and I began investigating temperatures measured by the existing TIROS-N family of weather satellites (average life span was only four years each). These satellites were designed to provide information for daily weather forecasts, not for answering questions about global climate change.
The instrument of interest to us was the Microwave Sounding Unit (MSU), identical copies of which were flown on all of NOAA's operational polar orbiters since 1979. The MSU measures the intensity of weak microwave radiation emitted to space by oxygen in the air. The magnitude of this intensity is proportional to air temperature, so with global coverage by the satellites we could compute the true globally-averaged air temperature. Two specific layers have lent themselves to accurate measurements: 1) the lower troposphere, or the lowest 7 km of air next to the surface, and 2) the layer at 17-21 km, or lower stratosphere.
Putting together a climate record from multiple satellites involved collecting a huge volume of data and was a remarkable achievement in and of itself. It is a tribute to the current government system and the vision of scientists at the National Center for Atmospheric Research (NCAR) that those data (with little perceived market value at the time) were saved and archived. The MSU data products are now almost priceless in the global warming debate in having established a precise historical record of the Earth's temperature over the last 18+ years.
It was our good fortune that my call to NCAR asking about the possibility of obtaining the MSU data came one week before a previously scheduled, major NCAR project was to begin to copy all satellite data from an old, outdated storage system to a newer one. Thus, forewarned that Spencer and I believed the MSU data were of some unique value, NCAR kindly extracted the necessary data (only 2% of the total) for us at only the marginal cost of the extraction process. This relatively "free and open" attitude concerning data availability was the key to our success in creating the MSU data set, since obtaining the data from a cost-recovering data center would have been prohibitive (the quote was over $1 million) for the speculative value of the MSU data for climate monitoring.
The computing facilities for our own massive processing task were provided by NASA's Marshall Space Flight Center, and we had the enthusiastic support of the Earth Science and Applications Division. After several months of tedious data analysis, we were able to construct various data sets with exceptional precision and continuity. The particular technique we eventually developed allowed the MSU data to be independently validated. In Fig. 1, I show the comparison between MSU temperatures and those measured by radiosondes (balloons) in which a weather instrument package is carried aloft. These two systems (satellite and radiosonde) are completely independent in every way. In Fig. 1 it is clear that both systems are measuring the same variations in temperature to high precision.
For long term variations, I include in the table below comparisons between large numbers of radiosondes and MSU measurements. It is again clear that both systems are telling us the same story on temperature variations since 1979. Note that none of the long-term trends differ by more than ±0.03 degrees C/decade.
Comparisons of trends since 1979 for MSU lower troposphere vs. various radiosonde-based tropospheric datasets which, except for the 850-300 hPa layer temperature, are weighted to match the MSU weighting function.
No. stations used BalloonTrend degrees C/dec. MSU Trend for same region Difference (Balloon minus MSU) Years Global (850-300 hPa) 1 63 -0.06 -0.04 -0.02 79-96 No. Hemisphere 2 250+ +0.01 +0.02 -0.01 79-96 So. Hemisphere 2 50+ -0.11 -0.08 -0.03 79-96 Global2 300+ -0.04 -0.04 0.00 79-96 W. No. Hemisphere 3 97 +0.16 +0.14 +0.02 79-94 1 Angell 1988 and updates. 2 Parker et al. 1997. 3 Stations in an area roughly bounded by Truk, South Pacific to Pt. Barrow, AK to Keflavik, Iceland to Trinidad. This is a comparison of sondes with colocated MSU.
Our datasets begin with January 1979 and continue to this day. We have been fortunate that two of the four MSU channels have performed exceptionally well on each of the nine satellites that were launched at intervals of about two years. It was critical that at least one satellite in functioning condition was orbiting when a new satellite was launched, because we required a period of overlap for precise intercalibration. (Only two satellites are operational at a given time).
3. The temperature of the lower atmosphere
The temperature of the global atmosphere is shown for the lower troposphere and lower stratosphere in Figure 2 (courtesy R. Spencer). Since we live in the lower troposphere, that time series has received the most attention. You will notice that there are large variations, both month-to-month and year-to-year. Because these variations are independently observed by two satellites, we know they are real. The trend in the time series is slightly downward (-0.05 degrees C/decade or -0.09 degrees F/decade). It is this relatively flat trend when compared to surface data (which show warming trends since 1979 of +0.09 degrees C to +0.14 degrees C/decade, depending on which dataset is cited) that has attracted attention to the Spencer/Christy MSU dataset.
Though the MSU temperature record has demonstrated high precision, there is also an element of ambiguity in the measurement. The layers measured by the MSU are several kilometers deep. Any intra-layer variability, therefore, would be masked by the vertical average. For example, a warming trend at upper levels and a cooling trend at low levels of one layer would be seen as no trend in the MSU vertical average.
One of the reasons the surface thermometer data have shown greater warming in the past 18 years is due to the fact that in continental regions the surface temperature responds with greater variation than the deep layer of air above. Over oceans (and in the global average), the opposite occurs. In the past 18 years there has been a tendency for the atmosphere over land areas to show warming (which is greater in the surface air response) while the atmosphere over oceans has exhibited cooling (greater effect in the MSU record). This pattern is thought to be due to natural variations. The net effect in the global average is a relative difference in the trends between surface air and the deep atmosphere. Thus, the uneven warming/cooling distribution of the past 18 years accounts for part of the difference.
Other differences are due to areas poorly sampled or not sampled at all by the surface network, as well as to some urban warming or land-use changes around many of the thermometers. It is a monumental achievement to construct a record of surface air temperatures, and most of these data sets have been subjected to many careful corrections to account for these non-natural temperature impacts.
Because of its precision and true global coverage, we believe that the MSU dataset is the most robust measurement we have of the Earth's bulk atmospheric temperature. At the same time, it is still a relatively short data set for climate studies. As indicated in Figure 2, the data contain both long and short period fluctuations. To be useful in the global warming debate one must understand and carefully account for fluctuations in the data that may be masking or dominating the anticipated enhanced greenhouse signal.
Recently, two colleagues have questioned the precision of the MSU data. They believe the data have spurious jumps in 1981 and 1991 which caused the overall trend to be downward rather than upward as they believe it should be. Their basis for this allegation utilized no observed data from the atmosphere. Since the time their allegations were made public I have shown that the MSU data are indeed precise with independent and direct observations of the troposphere (i.e. I used real data). For example, in the most serious allegation, my two colleagues speculated that the merging of one satellite, NOAA-7, into the time series caused a spurious 0.25 degrees C jump in late 1981 in the tropical time series. I show in Fig. 3 the temperature anomalies of two satellites NOAA-6 and -7 for the tropics during that time. It is important to note that these are completely independently calculated. One can readily see that whether NOAA-7 was included or not, the time series is still the same. Therefore, the addition of NOAA-7 into the dataset did not cause a problem and the claim of my colleagues is clearly in error.
4. The causes of the temperature variations
In a recent study, Dr. Richard McNider, also of the University of Alabama in Huntsville, and I looked for the causes of the natural fluctuations. We found that by accounting for the influence of tropical ocean temperatures (El Ni–o) and the cooling effect of volcanoes, we could explain over 60% of the monthly variations (Fig. 4). These natural, shorter-term fluctuations indicate to us how much the global temperature responds to specific causes. Once calculated and removed, we see that without El Ni–os and volcanoes, the temperature trend of the past 18+ years is upward (+0.06 degrees C/decade or +0.11 degrees F/decade, Fig. 4, bottom. The value varies from +0.05 to +0.10 degrees C/decade depending on certain parameters specified.). What is causing this upward trend? We do not know for sure. It may be the enhanced greenhouse effect. At the same time there could still be a longer term trend in the data due to variations in aerosols, water vapor, or other unknown factors that are masking the true magnitude of the greenhouse effect.
The latest results from global climate models, which include improvements and the cooling effects of air pollution, indicate warming rates for the Earth of +0.08 degrees C to +0.30 degrees C/decade for the latter part of the 20th century. These are about half of the warming rates predicted a few years ago, when only increases in greenhouse gases were modeled. Note too, that according to the latest models there should be more warming in the troposphere than at the surface. Therefore, the MSU is ideally suited to provide information on the layers that should show the greatest change. The present warming rate of +0.06 degrees C/decade observed in the "adjusted" MSU data is just outside this model range, and is not inconsistent with fully natural variations on decadal time scales. Therefore, uncertainty remains as to the cause(s) of the trend the MSU has measured.
Why is there a discrepancy between the models' estimate of global warming and what the MSU data have shown? One must remember that temperature is essentially a response parameter. The MSU data in Figure 2 show us what has been happening to the climate but not why. A key goal of efforts to study the planet from space is to provide heretofore unmeasured data that can provide an understanding of why the Earth system behaves as it does. I believe that new observables such as aerosols, rain structures, water vapor distributions and surface characteristics, when used in conjunction with the MSU data set will provide answers to these questions. Our work demonstrates that satellites can be used to monitor the Earth on decadal time scales and that the vantage point of space offers the only truly global view of the Earth system that can give robust measures of key variables.
The Spencer-Christy MSU data set has been used by some as evidence that global warming is not important, which then undercuts the need and urgency of programs to continue to study the Earth System. I strongly disagree with this interpretation. By showing that the Earth's rate of warming is slower than predicted by earlier models or surface data sets (Fig. 5), it does, perhaps, remove the sense of urgency for those who wish to enact greenhouse gas controls or to shut off scientific debate. But most importantly, the slower warming rate in the last two decades in effect gives us the security of time so that data from future observations and research may be used within the debate.
I believe that honest and open scientific debate with precise data is the key to making sound societal decisions. The cultivation of diversity of scientific thought is critical to vigorous debate. The MSU data set would not have been developed without the competitiveness and entrepreneurial spirit fostered by having separate NASA science centers and a broad university research program. Industry should recognize that good science and good data are their allies, whether in debates on acid rain or global warming. It is now more critical than ever that we study the planet's health with new diagnostic devices. Any delays in doing so may mean that the length of data records available to scientists will be reduced and cannot be used in the societal debates.
The disagreement between models and the MSU simply illustrates how little we understand about the complexities and factors that control the Earth's climate. Every month Roy Spencer and I process the newly arrived data and eagerly look at the month's temperature to see what is happening to the Earth. If we knew everything we needed to know about the Earth's system, we would not be as anxious about the results. I look forward to the time when new data from planned satellite sensors, coupled with an understanding of the Earth's climate system developed under research programs emphasizing global change, make surprises in the MSU global temperature as rare as being surprised by land-falling hurricanes in this era of weather satellites.
5. The temperature of the lower stratosphere
The record of the lower stratosphere is fascinating in its own right. Clearly, here is an example of global change on the scale of years to decades (Figure 2). The two conspicuous warming events were due to explosive volcanic eruptions - El Chichon (1982) and Mt. Pinatubo (1991). The aerosols injected by these explosions high into the stratosphere caused the warming through radiative interactions. Notice, however, that once the aerosols settled out, the global stratospheric temperature fell to levels below those observed at pre-eruption. It is widely thought that the loss of stratospheric ozone, both naturally from volcanic events and from human-generated chemicals, has caused this overall cooling. The increase in greenhouse gases, which will cause stratospheric cooling, is probably a factor as well, though smaller.
The 1996 annual stratospheric temperature was the lowest annual value ever measured by satellite, and March 1997, was the coldest single month on record for the North Polar region. (Globally, the temperatures have rebounded a bit for the first half of 1997.) Something is changing in the lower stratosphere -- the temperature tells us that much, but cannot specifically indicate the cause. (Others have much more experience here.) The extent of the stratospheric cooling trend points to the need to fully understand its cause.
6. Concluding remarks
Continued monitoring of global temperature through the Spencer-Christy method is expected as long as our good fortune holds and the two orbiting instruments do not fail (which almost happened recently). Thus, we should continue to provide the scientific community with precise temperatures for deep atmospheric layers.
In any weather variable, e.g. temperature, rainfall, etc., it is the shorter-term fluctuations (week-to-week) that cause the greatest impact on human productivity. One valuable benefit of a program of escalating Earth observations is the resulting improvement in weather forecasts -- particularly out to two to three weeks and even to seasonal averages. The potential economic impact of improved long-range forecasts would be enormous. Virtually every sector of our economy is sensitive to weather, especially those related to energy production and consumption, agriculture, transportation, insurance and recreation. Improved knowledge of coming weather situations would be used to add value to the products and services generated by these industries.
A strong and continuing program in atmospheric observation and research has this more subtle benefit as well. There will be extreme climate events in the near future because that is the nature of weather and climate. Without a continuing program of research that places climate variations in proper perspective and reports with improving confidence on their causes, we will be vulnerable to calls for knee-jerk remedies to combat "climate change," which likely will be unproductive and economically damaging. We can protect ourselves from such pitfalls by improving our ability to measure what the climate is doing and determine the causes for its variations.
In simple terms, the "Global Climate" is our patient. We have taken its temperature in a few places and have seen just enough change to cause concern. Before prescribing any powerful medicine though, the patient should be given a complete physical as soon as possible, so we may then make the proper diagnosis and chart a correct course of action for the benefit of all.
Angell, J.K., 1988: Variations and trends in tropospheric and stratospheric global temperatures 1958-87. J. Climate, 1, 1296-1313.
Christy, J.R., 1995: Temperature above the surface layer. Climatic Change, 31, 455-474.
Christy, J.R. and J.D. Goodridge, 1995: Precision global temperatures from satellites and urban warming effects of non-satellite data. Atmospheric Environment, 29, 1957-1995.
Christy, J.R. and R.T. McNider, 1994: Satellite greenhouse signal, Nature, 367, 325 (27 January 1994). (Fig. 4 of testimony, updated, taken from this article.)
Parker, D.E., M. Gordon, D.P.N. Cullum, D.M.H. Sexton, C.K. Folland and N. Rayner, 1997. A new global gridded radiosonde temperature data base and recent temperature trends. Geophys. Res. Lett., in press.
Spencer, R.W. and J.R. Christy, 1990: Precise monitoring of global temperature trends from satellites. Science, 247, 1558-1562 (30 March 1990).
Spencer, R.W. and J.R. Christy, 1992: Precision and radiosonde validation of satellite gridpoint temperature anomalies. Part I: MSU channel 2. Journal of Climate, 5, 847-857.
Written Testimony, John R. Christy
University of Alabama in Huntsville