During the past 20 years, global mean surface temperature has been rising at a rate as large as any that has been observed within the historical record.  Such rapid warming at the Earth's surface is in contrast to the trend in the global-mean temperature of the lowest 8 kilometers of the atmosphere (within that portion of the atmosphere referred to as the troposphere) as inferred from measurements of radiation emitted by oxygen molecules (a proxy for troposheric temperature) sampled by the microwave sounding unit (MSU) carried aboard the NOAA polar-orbiting satellites.

Until as recently as two years ago, the latest estimates of this so- called "tropospheric temperature trend" based on satellite data since 1979, were indicating a slight cooling from 1979 onward.  About a year ago, the algorithms used to process the satellite data were modified to take into account changes in viewing geometry due to the decay in the satellite orbits.  As a result of these rather small corrections, together with the extraordinary warmth associated with the 1997-98 El Nino, the satellite
data are now indicating a warming trend, but it is still much smaller than the trend in surface temperature.

In an effort to reconcile these seemingly contradictory sets of measurements, the National Academy of Sciences (NAS) convened an ad-hoc Panel whose members included the developers of the MSU and experts in remote temperature sensing, ground-based and balloon-borne atmospheric
temperature measurements, and specialists in the detection and modeling of global climate change.  Represented among the eleven Panel members were three lead authors of chapters of the1995 and forthcoming IPCC (Intergovernmental Panel on Climate Change) scientific assessments, and
individuals representing a wide spectrum of viewpoints with respect to the greenhouse warming issue.

The NAS Panel assessed the uncertainties inherent in the satellite, balloon-borne and ground-based measurements, and it considered the possibility that systematic biases might still remain despite the numerous corrections that have been made thus far.  It also considered the various technical issues that arise in comparing data sets with different sampling characteristics.  The Panel's task was rendered more complicated by the fact that satellites and ground-based thermometers do not measure the same physical quantityfor a variety of reasons surface temperatures and temperatures aloft do not track one another perfectly, either locally or in the global average

The episodic periods of warmth associated with El Nino events and the periods of global cooling that follow in the wake of major volcanic eruptions influence temperatures at different levels of the atmosphere to varying degrees.  Likewise, humans exert influences on climate that result
in a warming at the surface due to the buildup of greenhouse gases and a cooling in the stratosphere brought on by the depletion of stratospheric ozone.  In the presence of natural climate variability operating on a variety of timescales, a 20-year period of record such as the
satellite-based record of temperature examined by the panel of scientific experts, cannot yet be regarded as representative of the longer-term behavior of the climate system.

The Panel of experts affirmed the conclusion of the 1995 IPCC report that global mean surface temperature has warmed rapidly since 1979; and it noted that the upward temperature trend has continued and accelerated in the years since the 1995 report went to press.  A larger degree of uncertainty remains with regard to the tropospheric temperature measurements, but the
Panel believes it is more likely than not that the troposphere has been warming, but at a rate less than that of the temperature at the Earth's surface.  It is conceivable that there will need to be further adjustments to the estimates of global-mean surface and tropospheric temperature trends
to account for any additional sources of bias that have not yet been discovered.

The Panel stressed that even if the current estimates of surface, radiosonde, and satellite measurements prove to be correct, there is no basis for expecting that the surface will continue to warm at a rate faster than the troposphere in future decades.  The Panel noted that the high
degree of uncertainty inherent in the tropospheric temperature measurements underscores the need for more comprehensive global observations for monitoring climate change.
 

                                                       Evidence in Support of the Findings

Land-based surface temperatures are measured directly, while sea-surface temperatures (SSTs) are used to establish the monthly temperature of the air just above the ocean surface.  Because daily temperature variability of the ocean surface is relatively small, SSTs can be reliably determined with
fewer observations than would be required to establish marine air temperatures.  Coverage increases with time and is better after 1950, and global after 1982, when the capability of satellites to measure sea surface temperatures was added.  Biases occur through changes in observing
practices and changes in land use, such as the urban heat island effect.  The advantages are the long record of ground-based measurements of temperature from the mid 1800s, many independent measurements, several independent analyses, and many cross checks such as Northern versus
Southern Hemisphere values, rural vs urban, global vs land-based vs ocean vs marine air temperatures.  The disadvantages are the less than global coverage in sampling, which changes with time.  For example, underrepresentation of the Antarctic and the Southern Ocean might result in
a slight overestimate in estimates of the present temperature trend. Nonetheless, there is a high degree of confidence that the observed surface temperature trends are robust.

Balloon-borne temperatures of the lower atmosphere, assembled from up to nine hundred radiosonde stations that began operating in the mid-1940s, are at best twice daily, and were standardized in July, 1957.  These radiosonde stations provide good vertical resolution of temperature profiles.  The biases stem from many changes in instrumentation and observing methods, many of which have poor or no documentation.  Known biases occur in some
brands of radiosonde equipment, often due to radiation effects.  The advantages are that each sounding is with a new instrument; there are dozens of instrument types; and a few groups provide independent analyses of the record.  The disadvantages are the dozens of instruments that are
inadequately calibrated, with biases, often unknown, and that change with time; and much less than global coverage.

The satellite temperatures are estimated from microwave radiation emissions from oxygen which are proportional to temperature and are known as the "MSU 2LT" temperatures (otherwise referred to as the Microwave Sounder Unit, Channel 2, Lower Troposphere).  MSU data retrieval requires measuring microwave radiation in the troposphere from a variety of angles in order to
calculate the 2LT temperature record.  Coverage is global over a few days, two or four times per day, and began in December, 1978.  Observation times vary from one satellite to another, and as each satellite drifts in orbit.  Only very broad vertical layers of the atmosphere can be sensed.  The
biases arise from the use of nine different satellites which have been deployed to construct the satellite temperature record; orbital decay affects the 2LT temperature data retrieval because it changes the angles at which measurements are taken; and east-west drift alters the time of day at
which measurements are taken.  A shift in the sampling time from early to mid afternoon, for example, would produce a spurious warming.  Instrument calibration and solar heating of the MSU instrument platform require corrections.  The retrieval process itself amplifies the background noise, which interferes with the temperature signal that one is trying to detect.

The advantages are the long-term stability of microwave radiation emissions from oxygen (the proxy for tropospheric temperatures), and the global, fairly uniform, coverage.  Biases are well determined if there is adequate satellite overlap and millions of observations to help reduce the random
noise.  The disadvantages are that the temperature signal one is trying to measure includes a signal from 20% of the land surface (ideally, 100% of the signal should derive from the atmosphere); contamination by precipitation-sized ice; and biases are not constant.  Consequently,
continuity of measurements across different satellites is an issue, and overlap of measurements between and among the NOAA satellites is inadequate.  Only one group of scientists, for the most part, has processed the satellite temperature data, making it difficult, if not impossible, to
independently check the methodology, the data and the conclusions.

As data sets have been improved, discrepancies among them have been reduced and there seems to be good agreement between the radiosonde and MSU tropospheric temperatures, although the radiosonde record is inadequate in the tropics and Southern Hemisphere.  The fact that the surface temperature and the tropospheric temperature are two physically different quantities is believed to account for a considerable part of the differences between them.  In particular, observed stratospheric ozone depletion cools the MSU but not the surface; episodic volcanic eruptions cool MSU more than the surface; increasing greenhouse gases warm the MSU more; solar effects are
small but make for an added complication; and tropospheric aerosols are changing and have complex regional and vertical profile effects that are not well known.  El Nino and other natural climate variability is likely to produce a larger temperature signal in the troposphere, while day-night
differences are greatest at the surface. Land-ocean differences are also greater at the surface where winds are weaker.
 

                                                                    Biographies

Dr. John M. Wallace  is a professor in the Department of Atmospheric Sciences, and co-director of the Program on the Environment, at the University of Washington, Seattle.  From 1981-98 he served as director of the (University of Washington/NOAA) Joint Institute for the Study of the Atmosphere and Ocean.  His research interests and expertise include the study of atmospheric general circulation, El Nino, and global climate.  He is a member of the National Academy of Sciences; and a fellow of the American Association for the Advancement of Science, the American Geophysical Union, and the American Meteorological Society.  He is also a recipient of the Rossby medal of the American Meteorological Society and the Roger Revelle medal of the American Geophysical Union.  He has served on numerous panels and committees of the National Research Council.
 

Dr. Kevin E. Trenberth is a senior scientist and Head of the Climate Analysis Section of the National Center for Atmospheric Research (NCAR) in Boulder, Colorado.  Prior to joining NCAR in 1984, he was a Professor of Atmospheric Sciences at the University of Illinois and earlier worked in
the New Zealand Meteorological Service.  Dr. Trenberth has also contributed significantly to the Intergovernmental Panel on Climate Change (IPCC), as a lead author for Chapter 1 of the 1995 Scientific Assessment, and as a lead author for the 2000 IPCC assessment.

Dr. Trenberth has published over 280 peer-reviewed scientific articles or papers, including 26 books or book chapters.  He has served on a number of advisory committees and panels, including several dealing with El Nino research.  He is a member of the International Scientific Steering Group
for the World Climate Research Programme's Climate Variability and Predictability Programme, for which he recently (1996-1999) served as co-chair; he serves on the Joint Scientific Committee of the WCRP; he is a member of the National Research Council's Climate Research Committee Panel
on Reconciling Temperature Observations and the Committee on Global Change Research; he is a member of the National Oceanic and Atmospheric Administration (NOAA) Advisory Panel on Climate and Global Change and NOAA's Council on Long-term Monitoring; he serves on the National Science Foundation's Climate System Modeling Advisory Board; and he serves as a member of the ECMWF's (European Center for Medium Range Weather Forecasts) Reanalysis Project Advisory Group.

Dr. Trenberth is a Fellow of the American Meteorological Society and the American Association for the Advancement of Science, as well as an Honorary Fellow of the New Zealand Royal Society.  He is also the most recent recipient of the Jule G. Charney award from the American Meteorological
Society.