See the Index to Climate Change Fact Sheets: for the original documents.

Last revised 1 May 1993 by the Information Unit on Climate Change (IUCC), UNEP, P.O. Box 356, CH-1219 Châtelaine, Switzerland.Tel. (41 22) 979 9111. Fax (41 22) 797 3464. E-mail

1. An introduction to the science of man-made climate change

Certain gases in the atmosphere, such as carbon dioxide, play a crucial role in determining the earth’s climate. Although other factors are important as well, the composition of the atmosphere to a large extent controls our climate. Levels of so-called "greenhouse gases" are particularly important.Greenhouse gases affect the "energy budget " of the climate system. The earth receives energy continuously from the sun. It must get rid of this energy at the same rate by sending it back out to space. Greenhouse gases affect the ability of this outgoing energy to pass through the atmosphere. As their concentrations rise, the climate must somehow adapt, or change, to keep the energy budget in balance. One probable change (among others) is a warming of the earth's surface and the lower atmosphere.Atmospheric concentrations of greenhouse gases are rising rapidly, mainly because of human activity. By burning fossil fuels and deforesting the earth, mankind is increasing carbon dioxide levels. Intensive agriculture, coal mining, and leaky natural-gas lines are major sources of methane. Industrial products emit chlorofluorocarbons (CFCs). Nitrous oxide and low-altitude ozone levels are also increasing rapidly, for reasons that are less clear. Less than 200 years since we began making major emissions, greenhouse gas concentrations are rising to levels higher than any yet seen while humans have existed on this planet - and they will rise much further in the years ahead.Changes in greenhouse gas concentrations have been associated with dramatic climatic changes in the past. The last time greenhouse gas levels changed as much as they are changing now was when the earth emerged from the most recent ice-age. There is strong evidence that greenhouse gases played a significant role in that post-ice-age warming. The current increase in greenhouse gases will affect the climate - but we don’t yet know exactly how. The different components of the earth’s climate system interact in complex ways, causing natural climate variations, many of which are still poorly understood. But even if we understood the present climate much better than we do, the future could still hold surprises. Due to the unprecendented rise in greenhouse gas emissions, we are entering into a new, hitherto unexplored, climatic regime. Climate models indicate that one of the main effects of greenhouse gas emissions will be global warming. Assuming that no action is taken to reduce emissions, computer models of the earth’s climate indicate that global average surface temperatures will rise by 1.5-4.5 C over the next 100 years. This rise is larger and probably faster than any such change over the past 9,000 years.Climate models are far from perfect, and they rely on projections of future greenhouse gas emissions that are far from certain. But most scientists believe they provide the best estimates we have of future climate change. Emission scenarios and model predictions may overstate the risk, but they are equally likely to underestimate it.There is some evidence that this warming has already begun. Average world surface temperatures appear to have risen by 0.3-0.6 C over the past 100 years. But although many climatologists believe that this indicates a real change, the historical temperature record is poor. Moreover, the climate varies naturally, and this observed warming is still within the range of natural variability. Nevertheless, this warming is also broadly in line with what models predict should have resulted from emissions to date. Past greenhouse gas emissions have already committed us to more climate change in the future. The climate does not respond instantly to emissions, and most greenhouse gases remain in the atmosphere for decades after being released, continuing to influence the climate. This built-in delay increases the risks in waiting for more conclusive evidence before acting to reduce emissions. If no action is taken to reduce greenhouse gas emissions, the consequences for many of the world’s societies and ecosystems may be serious. Average sea-levels may rise, which would affect coastal communities through more frequent flooding and increased ground-water salinity. Changes in rainfall patterns and soil moisture levels are probable, but still difficult to predict. Both would have significant implications for agriculture.Those most at risk from climate change are likely to be those least able to adapt. Not all climate change impacts will be negative. But for natural ecosystems and subsistence agricultural societies that have evolved over centuries to suit the present climate, any rapid change is likely to be traumatic. Although uncertainties remain, we know enough to say with confidence that the risk of climate change is genuine and serious. There is much that we still do not understand about the climate system and our impact on it. But the level of uncertainty in climate models should not be exaggerated. It is no greater than the uncertainty in the economic data and models on which equally far-reaching policy decisions must be based.

2. The role of greenhouse gases

Almost all minor constituents of the atmosphere absorb some infra-red radiation - but some absorb more than others.Drawing a clear line between "greenhouse gases" (GHGs) and "non-GHGs" is therefore not easy. But which are the gases that really matter?All GHG concentrations are determined by a balance between "sources" and "sinks". There are two ways mankind can increase atmospheric concentrations of GHGs: by increasing the strength of GHG sources (processes that produce GHGs) and by decreasing the strength of GHG sinks (processes that remove GHGs). Man-made sources are generally the easiest to quantify, but both sources and sinks are important. The main greenhouse gas for the present climate is water vapour. In the lower atmosphere, however, water vapour levels are determined by the natural balance between evaporation and rainfall. They are therefore not directly affected by human activity (although they are affected indirectly through an important feedback mechanism). Some GHGs are increasing as a direct result of man-made emissions. The most important of these are carbon dioxide (CO2),methane, and chlorofluorocarbons (CFCs). The main source of "new" CO2 is fossil fuel emissions (figure A). Deforestation may also be significant, but it is more difficult to quantify. Once in the atmosphere, CO2 is chemically stable and lasts for many decades.  Carbon dioxide is removed from the atmosphere by a complex network of natural sinks. Most estimates suggest that about one-third of the CO2 being released at present is absorbed by the oceans. Another important and related sink is photosynthesis by vegetation on land and by plankton in the sea. Most of the CO2 absorbed by photosynthesis, however, is released again when plants and plankton decay or are eaten by animals. Only a small fraction is removed permanently. Now that atmospheric CO2 has risen well above its natural level, many aspects of this complex carbon cycle are changing. The cycle may also be affected by the destruction of forests and by feedbacks between global warming and chemical and biological processes in the oceans.Human activity affects methane levels in several ways (figure B). Converting land to agriculture, particularly rice paddies,releases methane. So do deforestation, coal mining, and the extraction and use of natural gas. Arctic regions may also act as a source if development (or even global warming itself) causes the release of methane frozen into tundra. Because natural sources of methane are not completely understood, methane’s role in climate change is still very uncertain. Unlike CO2, methane is destroyed by reactions with other chemicals in the atmosphere and soil, giving it a life-time of about 10 years. Human activity may also be interfering with these "sink" reactions. The increase in methane seems to have slowed recently, although scientists do not yet know why.CFCs 11, 12, and 13 are more straightforward. There are no natural sources, so all CFCs in the atmosphere are there because of emissions from aerosol propellants, refrigerants, foam production, and solvents. As well as attacking the ozone layer, CFCs are very powerful greenhouse gases and their levels are rising rapidly (figure C). Although concentrations should level off as the Montreal Protocol on Substances which Deplete the Ozone Layer is implemented, CFCs have long life-times, and their effects will be felt for many decades to come. By destroying ozone (itself a greenhouse gas) in the stratosphere, CFCs also affect the climate indirectly in ways that are still not fully understood. Important GHGs, including low-altitude ozone and nitrous oxide, are increasing as an indirect consequence of human activity. Ozone levels are falling only in the stratosphere (the famous "ozone hole"), where ozone is needed to protect us from ultra-violet radiation. Meanwhile, they are rising in the lower atmosphere. The gases responsible for creating low-level ozone are carbon monoxide, the oxides of nitrogen (all found in car exhaust fumes), methane, and other hydrocarbons. Nitrous oxide is also increasing, but for reasons that are still not well understood. Other GHGs include HCFCs, other hydrocarbons, and halons. Emissions of these gases have been less significant to date but might become more important as industrial processes change. This is especially true of HCFCs, levels of which are expected to increase rapidly as they replace CFCs as a result of the Montreal Protocol.Molecule for molecule, some GHGs are much stronger than others. As an example, figures B and C show, on their right-hand scales, concentrations of methane and CFCs in terms of "CO2-equivalent". This is the additional concentration of CO2 that would have approximately the same effect on the radiative properties of the atmosphere - and thus the same direct effect on climate - as the concentrations shown of those GHGs. Note that the CO2 -equivalent scales are equal. Carbon dioxide increases have dominated the enhanced greenhouse effect so far, but together the other gases now contribute to over 40% of it.

3.  An introduction to the climate system

The climate system is complex. It is governed not only by what happens in the atmosphere, but in the oceans, the cryosphere (glaciers, sea ice, and continental ice caps), the geosphere (the earth's solid surface) and the biosphere (living organisms in the oceans and on land). The interactions among these various "spheres" are difficult to predict, not least because their respective processes occur on widely differing time scales. The typical equilibrium response times of the climate system’s various components range from a single day to millenia. Solar radiation is the only significant source of energy driving the climate system. Because air is largely transparent to incoming short-wave solar radiation, this radiation does not directly heat the atmosphere very much. Instead, it warms the surface of the Earth. The surface then re-emits long-wave radiation that, because it can be absorbed by certain gases, warms the atmosphere.  The amount of warming that results from solar radiation depends in part on the nature of the earth’s surface. Ocean and land surfaces warm at different rates, and land covered by vegetation absorbs and reflects solar energy differently than do deserts or ice-caps. In this way, surface variations create complex patterns of surface energy distribution. The oceans have an important influence on our present climate. For example, wind-driven surface currents like the Gulf Stream carry large amounts of heat from the tropics to colder latitudes, which warms the atmosphere above. Other surface currents carry cold water towards the equator, which cools the atmosphere. Elsewhere, notably on the west coasts of continents, the winds blow away surface water; colder water from below wells up to replace it and cools the atmosphere. Deep-ocean currents affect long-term climate variations. Over most of the oceans, surface water is warmer, and so less dense,than the water beneath it. This discourages surface water from sinking downwards into the deep ocean. Only in certain regions,notably in the Antarctic and northwest Atlantic Oceans, does a combination of evaporation (which increases the water's salt content) and wintertime cooling make surface water dense enough to sink all the way down. This process of "deep-water formation" is still not fully understood, but it is clearly important. It is the primary mechanism whereby heat and dissolved carbon in surface water is transported down to the ocean depths, where they may remain for a thousand years or more. Changes in deep ocean currents may have caused natural climate fluctuations in the past, and their role in storing or releasing "excess" carbon could interact with man-made climate change in the future. Ice reflects a significant amount of incoming solar energy back out into space. So any changes to the amount of ice and snow on the earth - some climate models suggest that the Arctic Ocean’s ice covering would all but disappear in a warmer climate - would affect the amount of solar energy absorbed by the earth's surface.The biosphere’s role in the climate system is not yet well understood. The biosphere is made up of living organisms on land and in the seas. It helps to regulate climate through its role in the carbon cycle. Furthermore, land vegetation has a significant effect on surface reflectivity, heat, moisture, and energy. Because of the complexity of the biological processes involved, scientists can make only very general estimates of the biosphere’s role in the climate system. Much more research will be needed before the biosphere’s contribution to climate variation can be quantified. Much still remains to be learned about the atmosphere as well. Although the atmosphere has been widely studied and modelled (particularly by weather forecasters), large uncertainties remain concerning climate variability. One of the greatest unknowns is the role of clouds. Do they act to cool the earth by intercepting and reflecting solar energy, or to warm it by reducing outgoing terrestrial radiation? Satellite observations indicate that they probably do both, but their net effect on the present climate is very uncertain. It is even less clear how this net effect will change in response to global warming. Changes in cloud amount and cloud types might increase warming (positive feedback) or reduce it (negative feedback). A further unknown is exactly how the exchange of heat and gases between the atmosphere and other components of the climate system takes place. Much more research is needed to enable scientists to better predict how climate change will come about. According to the Intergovernmental Panel on Climate Change (IPCC), to improve our predictive capability, we need to understand better the various climate-related processes, particularly those associated with clouds, oceans, and the carbon cycle. We also need to improve the systematic observation of climate-related variables on a global basis; to investigate further past changes; to develop improved models of the Earth’s climate system; to increase support for national and international climate research activities, especially in developing countries; and to facilitate the international exchange of climate data.

4. Radiation, climate and climate change

The climate is controlled by the flow of energy to and from the earth. The earth is kept warm by the energy it continuously receives from the sun. It avoids getting too hot by simultaneously giving off energy as well. We can understand the climate, and how it might be affected by human activities, in terms of these two related processes. Radiation has a central role in climate because it carries energy through space. Most people associate radiation with nuclear weapons, but the term refers more generally to light, which we can see, and to related emissions that are invisible to the human eye but which can be detected by other means. For example, we can feel infra-red radiation from a fire as heat and, while we can't sense ultra-violet radiation from the sun directly, it can nonetheless give us sunburn. All radiation carries energy from where it is emitted to where it is absorbed. Radiation occurs as a wave motion that behaves somewhat like waves on the ocean. For this reason, the different forms of radiation are often labeled by their 'wavelength', which is the distance between the 'tops' of two waves. Different colors of light correspond to different wavelengths, and invisible radiation has either longer wavelengths (infra-red) or shorter wavelengths (ultra-violet) than does visible light. The earth absorbs short-wave radiation from the sun, mainly in the form of light, and emits long-wave, infra-red radiation to outer space. Because radiation carries energy, the absorption of radiation tends to cause warming while the emission of radiation tends to cause cooling. How much radiation a particular body emits depends on its temperature: other things being equal, radiation is emitted faster at higher temperatures. The flow of energy between the sun and the earth and between the earth and outer space is strongly affected by the atmosphere. Gases and clouds in the atmosphere absorb, emit, and reflect radiation. Each gas absorbs at a number of different wavelengths. The presence of atmospheric gases and clouds makes the passage of radiation to and from the earth's surface much more complicated than it would be otherwise. Furthermore, air movements in the atmosphere can also transport energy. On average, the energy 'budget' is balanced both at the top and at the bottom of the atmosphere. At the top, the total incoming radiation is equal to the total outgoing radiation. At the bottom, the total energy absorbed by the earth's surface is equal to the total energy transported away. In other words, the total heating equals the total cooling. Both these budgets are kept in balance because the rate of cooling increases as the temperature does. Any variations in total heating result in temperature changes that rapidly bring the budget back into balance. The energy budget is altered when the atmosphere's composition is changed - by, for example, mankind's emissions of greenhouse gases. The result can be a change in climatic temperatures. Increased amounts of gases such as carbon dioxide make the atmosphere absorb long-wavelength radiation from the surface more strongly and also emit more radiation back down towards the surface. The energy budget at the surface is therefore upset. For the balance to be restored, some warming must occur.

5. Is the Earth warming up yet ?

All major climate models agree that the clearest symptom of climate change would be a global warming. The way the atmosphere absorbs and emits radiant energy has already changed significantly because of greenhouse gas (GHG) emissions over the past 150 years. Models suggest this should have substantial climatic consequences, the clearest of which would be a sustained rise in the Earth’s average near-surface temperatures. Models still differ over whether we should yet be able to detect a significant temperature rise. Slower-acting components of the climate system, such as the oceans, may delay any climate change by up to several decades. Because the length of this response time is difficult to predict, it is unclear how much of the climatic response to recent emissions should have happened already, and how much is still to come. Also, recent research suggests that man-made sulphur emissions may be cooling the atmosphere enough to compensate for a significant fraction of the GHG-induced warming (although pollution controls may soon reduce this effect). Nevertheless, detecting a significant rise in global average temperature over the past 100 years would be strong evidence that the climate is indeed changing as the models predict it should. Global temperature records exist only for the past century or so. The data available to researchers looking for temperature trends are land-based air temperature measurements and marine air-temperature and sea-surface temperature records. All begin around 1860. Before then, data collection was not systematic enough for a global record. Land-based records come mainly from meteorological stations. Marine records rely principally on observations made by merchant ships. In both cases, therefore, more observations have always been made in some regions, such as densely populated areas or well-travelled shipping lanes, than in others. Data is very scarce for certain regions, particularly in the period before 1900.1 For example, ships avoid some regions in specific seasons; very few travel to the Southern Ocean during the winter even today. All data must therefore be analysed carefully to ensure that such patchy coverage does not introduce systematic biases. Since 1900, the change in spatial coverage does not seem to have affected land records significantly.2 Before then, however, even careful analysis may lead to long-term averages that are either too warm or too cold by up to 0.1 C.1  Records may be affected by changes in the way observations are made. Observation methods, such as the time of day that air temperatures are taken, affect what is recorded. If methods were applied consistently throughout an observation period, this would not matter. One meteorological station might record temperatures 0.5 C lower than another in a neighbouring country but, provided it did so consistently, both stations would record the same trend. The problem is that the methods themselves change, and the proportion of observations taken using one method in place of another also changes. The effect of some outmoded methods, such as the pre-1940s use of uninsulated canvas buckets to gather sea-surface temperature data, can be compensated for, but only approximately.1 Alternatively, data sources known to be heavily "contaminated" can be avoided. For example, climate researchers use only night-time marine air temperatures, as day-time records are too erratic.Changes in the local environment may also introduce erroneous trends. For example, cities have a significant warming effect on their local environment. Since many meteorological stations are located in or near large cities, these "urban heat islands" might introduce a spurious trend into temperature records.3 This is the most serious possible source of systematic error to have been identified in land-based data. Considerable effort has gone into quantifying and correcting for this effect by comparing data from rural and urban stations. Generally, the remaining uncorrected effect from urban heat islands is now believed to be less than 0.1 C, and in some parts of the world it may be more than fully compensated for by other changes in measurement methods.4 Nevertheless, this remains an important source of uncertainty.The warming trend observed over the past century is too large to be easily dismissed as a consequence of measurement errors. Some errors might exaggerate the apparent warming, while others might cause it to be underestimated. Their net effect is thought to be substantially less than the 0.3-0.6 C warming observed since 1860. Figures (a) and (b) show combined land and sea temperature trends since 1860 for the northern and southern hemispheres respectively. Both trends are very uneven. In the northern hemisphere, a substantial warming occurred between 1910 and 1940. Thereafter, temperatures fell by about 0.2 C to the mid-1970s, and then increased sharply again. The trend in the southern hemisphere shows a clearer warming trend beginning around the turn of the century, but it is still very uneven.The size of the observed warming is compatible with what climate models suggest should have resulted from past GHG emissions. It is also small enough to have been caused instead by natural variability (although it is also plausible that natural variability has temporarily masked some of the warming caused by GHG emissions). Other explanations, such as a link with the solar sunspot cycle, remain tentative suggestions. Among the various possible explanations of the observed trend in recent global temperatures, the GHG hypothesis has the strength that it is both compatible with the available evidence and is supported by a plausible physical mechanism. Nevertheless, according to the 1990 IPCC report, "the unequivocal detection of the enhanced greenhouse effect from observations is not likely for a decade or more"

Notes:1 C. K. Folland et al., in "Climate Change: The IPCC Scientific Assessment", J. T. Houghton et al., eds., Cambridge, 1990.2 P. D. Jones et al., "J. Clim. Appl. Met.", 25, 1987, pp. 161-179 and 1213-1230.3 R. C. Balling & S. B. Idso, "J. Geophys. Res", 94D3, 1989, pp. 3359-3363.4 P.D. Jones et al., "Nature", 75, No. 6289, September 1990, pp. 169-720.

6. How records from the past climates support the case for global warming

One of the key problems in predicting climate change is determining how surface temperatures will respond to a given rise in greenhouse gas (GHG) concentrations. This "climate sensitivity" not only depends on the direct effect of the GHGs themselves, but also on natural "climate feedback" mechanisms, particularly those due to clouds, water vapour, and snow cover. Each of these may change in response to global warming and might therefore act either to enhance or to suppress any temperature rise. Climate models suggest that water vapour and snow cover, at least, are "positive feedbacks" and so should enhance the warming. Nevertheless, an estimate of total climate sensitivity that considers all feedbacks is crucial for checking these model results. Past climates have left records in ice and ocean-sediment cores that provide some of the best available evidence.1 A couple of kilometres beneath the surface of the Antarctic and Greenland ice-sheets lies ice which has been there for tens of thousands of years. Ocean sediment records go back even further. As sediments form on the floor of the ocean and snow piles up, trapping air bubble into ice, they store information concerning the climate of their day and the factors which affected it. The figure below shows that temperatures have varied closely with GHG concentrations for most of the past 100,000 years. The problem then arises, was this because GHG levels caused the temperature to change, or the other way around? Scientists can attempt to answer this question, but as yet only tentatively. Scientists have learned that several factors affect the earth’s climate. The various "climate-controlling" factors operate on very different time-scales. The slower-acting factors are the earth’s orbital movements around the sun and the expansion and retreat of the polar ice caps. The faster-acting ones are atmospheric dust; changes in ocean circulation; feedbacks due to water vapour, clouds, and snow; and the concentrations of greenhouse gases.Variations in the earth’s orbital movements around the sun create very slow climate cycles. The earth "wobbles" slowly on its axis, rather like a spinning top which is about to fall over. Called "Milankovich cycles", these wobbles affect the amount of solar energy the Earth receives and where that energy is deposited. This in turn affects the climate, introducing regular cycles with periods of up to 100,000 years. They are the main factor behind the onset and retreat of the ice-ages. Scientists can tell from basic astronomical theory how Milankovich cycles have evolved over the past hundreds of thousands of years.The ice-sheets affect the climate by reflecting sunlight. All other things being equal, the more ice on the earth, the colder it is, because ice reflects sunlight better than does any other surface covering. So during an ice-age, the global temperature is cooler than it would be due to Milankovich cycles alone, because the ice-sheets reflect solar energy. Scientists estimate the past volume of ice-sheets in the following way: As water freezes, different isotopes (types of chemicals) tend to freeze out at different rates. If a lot of water is bound up in ice-sheets, oceanic concentrations of the slowest-freezing isotopes rise. By looking at the concentrations of these isotopes in marine sediments, scientists can work out how much ice was around when they were formed. Three other factors - dust, water, and greenhouse gases - have a much more rapid impact on climate. The amount of dust in the atmosphere depends mainly on volcanic activity. Dust levels in past climates can be estimated from the ice surrounding bubbles trapped in ice-cores. Greenhouse gas concentrations may be found by analysing the bubbles themselves. This leaves all other feedbacks including changes in ocean circulation, water vapour, clouds, and snow as the undetermined factors in past climate changes. Ice-cores can also tell us about past temperatures. Because different isotopes of water freeze at different rates at different temperatures, ice composition depends on the temperature at which it formed. Analysis of Antarctic ice tells us about past temperatures in the snow-formation region above the Antarctic. This isn’t the same as knowing the global temperature, but there is some evidence that Antarctic temperatures and global temperatures rise and fall together.Knowing how these factors have changed, and knowing the temperature response, scientists can deduce the role of the remaining factors. This assumes of course that there is only one equilibrium temperature corresponding to a given array of "settings" for these climate-controlling factors. That might not be the case: first, because there might be another factor which scientists haven’t thought of, and second, because the temperature might not depend on the current settings of the climate-controlling factors alone, but also on previous settings.  Making these assumptions gives a value for climate sensitivity consistent with that found by climate models. These models suggest that if the net effect of ocean circulation, water vapour, cloud, and snow feedbacks were zero, the approximate temperature response to a doubling of carbon dioxide from pre-industrial levels would be a 1oC warming. However, after taking into account changes in the three other climate-controlling factors that we know about - orbital changes, ice caps, and dust - a comparison of past temperature changes with past changes in GHG concentrations indicates that such a doubling of CO2 would actually result in a warming of 3oC or more. Assuming that scientists haven't left out anything vital, this suggests that the net effect of water-based feedbacks is positive and would amplify GHG-induced warming by more than a factor of two.Many assumptions have been made, but the historical evidence increases our confidence in model results. Ice-core record analysis suggests that changes in GHG concentrations are associated with short-term (century time-scale) temperature changes, strongly amplified by cloud, snow, and/or water vapour feedbacks. This is in remarkably good agreement with the picture given by climate models. Notes:1 C. Lorius et al: "Nature", 347 (1990), pp. 139-145.

7. Measuring the "global warming potential" of greenhouse gases

The "strength" of different GHGs varies enormously. In 1990 CO2 accounted for more than 98% by weight of the total emissions of the five main GHGs (low-level ozone is not considered here or elsewhere in this sheet because its impacts, although large, are still difficult to quantify). The figures on the back page show that, whatever time-scale we consider, the contribution of CO2 to the total effects of 1990 GHG emissions was much less than 98% because, ton-for-ton, it is the weakest of the main GHGs. The immediate effects of an increasing concentration of a particular greenhouse gas depend on the properties of the gas itself. They also depend on how much of it (and how much of certain other gases) is already present in the atmosphere. The molecular properties of a greenhouse gas (GHG) determine how much infra-red radiation it will absorb and in which wavelengths. This is clearly important, but we also need to know how much energy the gas is likely to encounter in those wavelengths as it drifts around in the atmosphere. Think of an analogy: if you add a little mud to a clear swimming pool, the effect is immediately apparent. If you put the same amount of mud into a murky pool, you wouldn’t notice any change: the pool was opaque already. Carbon dioxide (CO2), for example, occurs naturally, so the atmosphere is already partly opaque to wavelengths absorbed by CO2. This reduces the direct impact of CO2 emissions. So unlike CFCs, which did not exist in the atmosphere until man introduced them, each additional kilogramme of CO2 has slightly less effect than the last one as the relevant wavelengths slowly "black out" (as the pool gets muddier, extra mud has less effect on its appearance).Present GHG emissions will affect future GHG concentrations in different ways, depending on the particular "life-cycle" of each gas. CFCs have the simplest life-cycles. The only way the atmosphere gets rid of them is through slow destruction by sunlight in the stratosphere. The average "lifetime" of CFC-11, for example, is 55 years. The other CFCs have lifetimes ranging from 90 to 400 years. Crucially, the lifetimes of the main CFC replacements (the hydro-chlorofluorocarbons, or HCFCs) are much shorter: typically about 15 years, which reduces their long-term effect on the climate. Lifetimes of the other GHGs are somewhat harder to define, since their life-cycles are too complex to be characterised by a simple decay process. Approximate lifetimes are 50-200 years for carbon dioxide, 10 years for methane, and 130 years for nitrous oxide. Some idea of the relative importance of different GHG emissions is given by their Global Warming Potentials (GWPs).The direct GWP of methane, for example, is defined as the cumulative direct effect on the atmosphere's energy budget resulting from a one-kilogramme release of methane, relative to the direct effect of a one-kilogramme release of CO2. In calculating this cumulative effect, it is necessary to specify the time horizon over which we are interested in the impact of a particular gas. The choice of time horizon strongly influences the relative importance assigned to current emissions of each greenhouse gas. The Intergovernmental Panel on Climate Change (IPCC) calculates GWPs for three reference time horizons: 20, 100, and 500 years (see figures below). The 20-year horizon is relevant to short-term impacts, such as changes in weather patterns; the 100-year horizon applies to longer time-scale changes, such as sea-level rise; while 500 years represents the longest time-scale it is felt reasonable to consider given our current knowledge. On short time-scales, 1990's CO2 emissions contribute over half the direct effects of 1990's total GHG emissions, and methane almost 30%. However, because methane has such a short life-time, the relative importance of 1990 methane emissions becomes much less for longer time horizons. At present, scientists can only calculate GWPs for the direct climatic effects of a gas, but the indirect effects may also be very important. GHGs interact with each other and with other gases in the atmosphere. For example, the chemical reactions that destroy methane also produce water vapour, which can have a significant warming effect, particularly when the vapour occurs at high altitude. This may increase the total climatic impact of methane emissions by 5-40%. On the other hand, CFCs destroy ozone (a GHG) in the stratosphere. By doing so, they may largely compensate for their own direct impact as GHGs. These various indirect effects are still not understood well enough to be quantified in terms of GWPs.2 Estimates of GWPs based on current knowledge must be regarded as indicative only. They must therefore be used very cautiously in formulating policy. The impact of policies which involve trade-offs between one GHG and another (such as replacing coal with natural gas, which would reduce CO2 but might increase methane emissions) is especially uncertain, since current models of both gases' life-cycles(and thus their relative GWPs) may need to be revised in the future. Notes:1 K.P Shine et. al., "Radiative forcing of climate" in Houghton, et. al., "Climate Change: The IPCC Scientific Assessment",Cambridge (1990).2 I.S.A. Isaksen,, in Houghton,, "Climate Change 1992, the Supplementary Report to the IPCC Scientific Assessment",Cambridge University Press, 1992.

8.  Why three hot summers don’t mean global warming

The global temperature is rather like the level of unemployment: hard to define or measure, and easy to misunderstand. Whenever the northern hemisphere has an unusually hot summer, newspapers attribute it to the greenhouse effect. Whenever  it has a particularly cold spell in winter, the same papers ask "what happened to global warming?" Both lines are misleading. Greenhouse gases are not the only factor affecting global temperatures. Some of the other factors are also due to human activity; industrial emissions of sulfur dioxide, for example, may have a cooling effect. Still others are quite natural, such as volcanic eruptions (which can cool the climate temporarily) and variations in the energy output of the sun. Almost all climate models suggest, however, that the world's average surface temperature ought to have warmed somewhere between 0.4 C and 1 C since pre-industrial times as a result of the greenhouse gases emitted so far. So one of the clearest signs of man-made climate change would be the detection of such a warming. Figure A below, which graphs the global annual average temperature from 1861 to the present, does indeed seem to show a warming trend.1 But such data must be interpreted carefully. Natural climate variations make it difficult to distinguish long-term trends. If there happened to be a natural temperature cycle which was at a minimum in 1861 and near a maximum at present, then plotting out temperatures over this time period and drawing a straight line through them could give a misleading impression of the trend. Take an analogy: if you made a series of hourly measurements of the brightness of the sun, starting at noon and ending at midnight two-and-a-half days later, you might well conclude that the world was getting darker - and you would be wrong. Of course, modern statistical techniques assume nothing so naive as a straight line. Instead, they attempt to explain observed variations in terms of a long-term trend (which may or may not be a straight line) as well as a set of fairly regular fluctuations about that trend. They thus reduce a complicated signal to a small number of fairly simple patterns. State-of-the-art statistical methods indicate a global warming of approximately 0.45 C since the beginning of this century . The thick line in figure A shows the underlying trend in global average temperatures obtained using such a "pattern-recognizing" statistical technique.2 It isn't a straight line, but it clearly indicates a warming trend. but substantial fluctuations have occurred around this underlying trend. This is particularly clear if we consider the most recent 30 years of the series (shown in figure B) and use these statistical tools to "filter" the data (in an effort to reduce the amount of random noise). The thick line in figure B shows the underlying trend again, and the thin line shows how the filtered signal fluctuates about that trend.Natural temperature fluctuations seem to be occurring on several time-scales. Some of these fluctuations may be due to natural climate oscillations and may therefore be fairly predictable. Others may be completely random. Data which has been filtered in this way must be interpreted with caution. But whatever their origin, figure B suggests that these natural fluctuations conspired to make the late 1980s particularly warm, much warmer than would have been expected on the basis of the underlying trend alone. It is quite possible that the world will cool over the next few years as the system "swings back" from the hot 1980s, even though the underlying trend remains upwards. These are very recent results, and aspects of them are still under debate.3 But the basic message for non-statisticians is clear:Three hot summers don't mean global warming, nor do three cold winters mean a new ice age. The most statistics can tell us at present is that there does appear to be a genuine warming trend in figure A. Whether this trend is the effect of greenhouse gas emissions or of a natural fluctuation due to some as-yet-undiscovered mechanism cannot be determined from an analysis of the global mean temperature alone. Such natural, century-time-scale fluctuations appear to have occured in the past (although none in the last 9,000 years was as large as the projected change over the coming century). Unambiguous detection of climate change is likely to be a painfully slow process, involving much more detailed comparison of climate model results with observations.4 There is no climatic counterpart to the Antarctic ozone hole. We must not expect a single, dramatic discovery to confirm"global warming" once and for all. If we wait for that discovery, we will wait for a long time - until well after it is too late to do much about it. Notes:1 D. E. Parker & C. K. Folland, Proc. U.S. D.o.E. Workshop on Greenhouse-Gas-Induced Climate Change, Elsevier, 1990.2 M. Ghil & R. Vautard, "Nature", 350, March 1991, pp.324-327.3 J. Elsner & A. Tsonis: "Nature", 353, October 1991, pp.551-553, and M. Allen, P. Read and L. Smith, "Nature", 355, February 1992, p.686.4 T. Wigley and T. Barnett in J. Houghton, et al (eds.), "Climate Change, the IPCC Scientific Assessment", chapter 8, pp.245-255, Cambridge, 1990.

9. Why "climate change" and "global warming" are not the same thing

Recent accounts of the scientific debate on climate frequently misrepresent what is being argued about. They suggest that scientists are still discussing whether or not the climate is changing in response to greenhouse gas (GHG) emissions, as if there were a simple yes/no answer. So if a scientist questions the adequacy of present climate models, or fails to find conclusive evidence for global warming in a particular data-set, he or she is often reported as claiming that "there isn’t really a problem". However, in most scientific circles the issue is no longer whether or not GHG-induced climate change is a potentially serious problem. Rather, it is how the problem will develop, what its effects will be, and how these can best be detected.The confusion arises from the popular impression that "the enhanced greenhouse effect", "climate change", and "global warming" are simply three ways of saying the same thing. They are not. No one disputes the basic physics of the "greenhouse effect". However, some of the consequences of the basic physics, including higher average temperatures due to "global warming", are less certain (although highly probable). This is because the fundamental problem concerns the way GHG emissions affect the flow of energy through the climate system, and temperature is just one of many forms  energy takes.In the long term, the earth must shed energy into space at the same rate at which it absorbs energy from the sun. Solar energy arrives in the form of short-wavelength radiation; some of this radiation is reflected away but, on a clear day, most of it passes straight through the atmosphere to warm the earth’s surface. The earth gets rid of energy in the form of long-wavelength, infra-red radiation. But most of the infra-red radiation emitted by the earth’s surface is absorbed in the atmosphere by water vapour, carbon dioxide, and other naturally occurring "greenhouse gases", making it difficult for the surface to radiate energy directly to space. Instead, many interacting processes (including radiation, air currents, evaporation, cloud-formation, and rainfall) transport energy high into the atmosphere to levels where it radiates away into space. This is fortunate for us, because if the surface could radiate energy into space unhindered, the earth would be more than 30 C colder than it is today: a bleak and barren planet, rather like Mars. By determining how the air absorbs and emits radiation, greenhouse gases play a vital role in preserving the balance between incoming and outgoing energy. Man-made emissions disturb this equilibrium. A doubling of the concentration of long-lived greenhouse gases (which is projected to occur early in the next century) would, if nothing else changed, reduce the rate at which the planet can shed energy to space by about 2%. Because it would not affect the rate at which energy from the sun is absorbed, an imbalance would be created between incoming and outgoing energy. Two percent may not sound like much, but over the entire earth it would amount to trapping the energy content of some three million tons of oil every minute. The climate will somehow have to adjust to get rid of the extra energy trapped by man-made greenhouse gases. Because there is a strong link between infra-red radiation and temperature, one probable adjustment would be a warming of the surface and the lower atmosphere. But it is important to realize that a warmer climate is not the only possible change, nor even necessarily the most important one. The reason for this is that radiation is not the only energy transport mechanism within the lower atmosphere (although it does play a vital, controlling role). Instead, the surface energy balance is maintained, and surface temperatures are controlled, by that complex web of interacting processes which transport energy up through the atmosphere. In contrast to the case of radiation, it is very difficult to predict how processes like cloud-formation will respond to greenhouse gas emissions. Global warming is a symptom of climate change, but it is not the problem itself. It may be the clearest symptom we have to look for, but it is important not to confuse the symptom with the disease.The fundamental problem is that human activity is changing the way the atmosphere absorbs and emits energy. Some of the potential consequences of this change, such as sea-level rise, will depend directly on how the surface temperature responds. But many of the most important effects, such as changes in rainfall and soil moisture, may take place well before there is any detectable warming. If a scientist argues that the warming may not be as large or as fast as models predict, he or she is not suggesting that the problem of climate change should be ignored. The point is only that this particular symptom - global average temperature - may be unreliable. We know that the air’s radiative properties are changing, and we know that the climatic effects of this change will be profound. All climate models (fact sheet 14) indicate that the most significant change will be a global warming. But even if it isn’t, other effects, equally profound, are inevitable. We are altering the energy source of the climate system. Something has to change to absorb the shock.

10 CO2 emissions and  the "missing carbon" problem

Carbon dioxide is the most important greenhouse gas directly produced by human activities.Present emissions of carbon dioxide (CO2) account for about half the short-term climatic impact of man-made greenhouse gases (GHGs). In the longer term, CO2 may be even more important because, unlike several other GHGs, it is not destroyed by sunlight or chemical reactions in the atmosphere. Models indicate that the effects of current CO2 emissions will continue to be felt for centuries. To understand how mankind's CO2 emissions affect the climate, we need to understand the "carbon cycle". Large amounts of CO2 and other compounds containing carbon are continuously exchanged between the atmosphere, the oceans, and the biosphere (animals, plants, and soil). We must understand these exchanges if we are to predict how CO2 emissions will affect future atmospheric concentrations of CO2.Exchanges of CO2 within the natural carbon cycle are many times greater than man-made CO2 emissions. However, these natural exchanges are also almost exactly balanced (see figure). Atmospheric CO2 levels varied by less than 5% between the end of the last ice age (9,000 years ago) and the beginning of this century. Largely as a result of human activities they are now rising at a rate of over 4% per decade. Not all man-made CO2 emissions can be easily accounted for. During the 1980s, fossil fuel use and cement manufacturing emitted an average of 5.4 billion (5,400 million) tonnes of carbon (in the form of CO2) per year. If the impact of deforestation and other land-use changes is included, the total annual CO2 emission rate is equal to 6-8 billion tonnes of carbon. Monitoring stations indicate, however, that only about 3.4 billion additional tonnes are accumulating in the atmosphere each year. This leaves around 3 or 4 billion tonnes that are somehow being absorbed by the oceans, the land biosphere, or both.One possibility is that most of the man-made CO2 which does not accumulate in the atmosphere is being absorbed by the oceans . . . This view is supported by indirect evidence derived from the atmospheric nuclear bomb tests of the 1950s and 1960s. These explosions produced a very large amount (relative to natural levels) of radioactive carbon-14. By measuring how much of this C-14 has penetrated into the oceans over the years since the tests, and determining the relationship between the absorption rate of C-14 and that of CO2, scientists can estimate the rate at which oceans absorb CO2.but direct measurements of how much CO2 the oceans absorb are very uncertain. The oceans' uptake of CO2 primarily depends on how fast CO2 can be transported downwards from the ocean surface; if "too much" CO2 accumulates in the surface layers at one time, absorption slows down. Global data on the concentrations of CO2 in the oceans are still very sketchy, and many contributing factors, such as ocean biology, are still not well understood. Absorption is also known to depend on local weather conditions. For example, winter storms accelerate CO2 exchange by exposing sea-water to the atmosphere through bubbles and spray. Unfortunately, delicate measurements are hard to carry out on a ship in the middle of a North Atlantic gale. The geographical patterns of atmospheric CO2 increase suggest that there is an unexplained "sink" of CO2 somewhere in the northern hemisphere. Most man-made CO2 is produced in the northern hemisphere, and it is relatively difficult for the atmosphere to "mix" CO2 across the equator. This means that CO2 levels in the northern hemisphere should be significantly higher than those in the southern hemisphere. They are, but the difference is much smaller than models suggest it should be given current emissions. This is known as the "missing carbon" problem. A few scientists have recently suggested that the "CO2-fertilisation effect" may be causing land vegetation to absorb much more of the excess CO2 than was previously thought. The idea is that plants in a CO2-rich atmosphere grow faster, and so absorb more CO2. Such an effect clearly exists: horticulturists today pump CO2 into greenhouses to encourage their tomatoes. Since most land plants are in the northern hemisphere, they are in the right place to provide the missing northern-hemisphere sink. However, many other scientists are sceptical that CO2-fertilisation could be strong enough to account for around 2 billion tonnes of carbon per year.Feedbacks involving different components of the carbon cycle - and climate change itself - will affect how CO2 levels respond to man-made emissions. CO2-fertilisation, for example, is a negative feedback. On the other hand, increased atmospheric CO2 is expected to cause global warming, and warmer ocean waters would release more CO2 into the atmosphere (somewhat as a warm soda-water bubbles more fiercely than a cold one when it is opened). So a small initial increase in CO2 could amplify itself: a positive feedback. Clearly, further research into the carbon cycle will be essential to reduce the level of uncertainty about the climate system's response to CO2 emissions.For further reading: R. T. Watson et al.: "Green-house gases and aero-sols" in Houghton et al., "Climate Change, the IPCC Scientific Assess-ment", Cambridge (1990). P. P. Tans, I. Y. Fung and T. Takahashi, "Science", 247, 1431-1438 (1990).

11 How much will the climate change?

Scientists rely on computerised climate models to make their predictions about climate change.Modelling experiments begin with a computer simulation of the present-day climate. To investigate climate change, scientists increase the concentrations of greenhouse gases (GHGs) in the model. As the model adjusts to these increases, it indicates projected changes in temperature, rainfall, cloudiness, and other climate variables. Climate change cannot be summed up by just a single number. Newspaper reports of climate modelling experiments normally focus on predicted changes in global temperature. But this is a simplification. "Global warming" is only one of many potential consequences of GHG-induced climate change. If model X indicates a smaller warming than model Y, this may be because the two models treat an important climatic process, such as cloud formation, in different ways. This in turn may mean that something other than global temperature - for example, rainfall - has changed much more in X than in Y. A change in local rainfall may affect human society more than a change in global temperature, so we should beware of equating the size of the projected global warming with the potential seriousness of the climate change problem. Global temperature is, nevertheless, a useful and important indicator. Comparing global temperature changes in different models is one way of checking their results. Comparing model predictions of GHG-induced warming with recent natural temperature fluctuations also indicates the potential scale of man-made climate change.Early modelling experiments focused on the total long-term change resulting from a doubling of carbon dioxide (CO2) levels. Doubling CO2 provides a useful benchmark, but it does not represent the full extent of the climate change problem. No matter what policies are adopted now for controlling emissions, levels of "equivalent-CO2" (the combined climatic effect of all man-made GHGs expressed in terms of CO2) are almost certain to double before the middle of the 21st century. Under some emissions scenarios, quivalent-CO2 levels will reach 1,000 ppmv - three times the 1980 CO2 level - by 2100, and continue rising thereafter. Over the next few decades, however, the rate of climate change will matter more than the total long-term change. Doubled-CO2 experiments indicate the total change to which we are "committed", in the long term, after raising GHG levels by this amount. But because of the inertia of the climate system, any warming will take place over many decades. For policy-makers, the speed of climate change over the coming decades matters as much as the total long-term change, since this rate of change will determine whether human societies and natural ecosystems will be able to adapt fast enough to survive.New results indicate a warming rate of about 2.5 C per century over the coming decades (assuming no attempt is made to reduce GHG emissions). These studies assume a gradual increase in CO2 from present-day levels at a rate of about 1% per year. Figure A shows the temperature change under such a "business-as-usual" emissions scenario in a simple climate model. Experiments with coupled ocean-atmosphere general circulation models (which represent the complexity of the climate system much more realistically than this simple model) give similar results. The solid line shows the current "best estimate" of the temperature change; the dotted lines show the range of uncertainty in the climate response to these emissions. Under this scenario, temperatures continue to rise even after the year 2100. This warming comes on top of an estimated 0.45 C warming which may have already occurred due to past GHG emissions. The shaded region in figure A shows, very approximately, the range of recent natural temperature fluctuations on 1,000-year time scales. Any GHG-induced warming to date still lies within this range, but the projected man-made warming over the coming century will be significantly larger and faster than recent natural fluctuations. Even under a minimum-emissions scenario, models indicate a warming of 1.5 C above present-day temperatures by the year 2100. This is almost 2 C above pre-industrial levels. Figure B shows the response of the same simple model to the lowest of the emissions scenarios considered in 1992 by the Intergovernmental Panel on Climate Change (IPCC). This is the only scenario under which the temperature begins to stabilise again by 2100. The key difference from figure A is that much lower population and economic growth rates are assumed in this lower emissions scenario. Higher growth scenarios are equally credible and would lead to even faster warming than shown in figure A.

13 How climate models work

Why is climate modelling different from weather forecasting? Although they have much in common,the two problems are fundamentally different. Weather forecasting is an "initial value" problem: the forecaster needs to know the present state of the weather and how it is evolving so that he can predict its behaviour in the immediate future. Slow-changing factors, such as concentrations of greenhouse gases (GHGs), can be held constant in a weather forecasting model. The climate modeler, on the other hand, is not interested in the system’s behaviour at a particular point in time. Instead, he wants to know how it’s average behaviour will respond to slow changes in precisely those components of the climate system that the weather forecaster specifies as constant.  Climate modelers must consider all five of the climate system’s key components. These components evolve and interact on very different time-scales. The atmosphere changes in hours, and its detailed behaviour is impossible to predict beyond a few days. The upper layers of the oceans change in a matter of weeks, while the deep layers vary over decades to millenia. The biosphere consists of animals and vegetation and normally changes relatively slowly; reduced rainfall might take years to alter vegetation cover significantly, although certain components (such as humankind) are capable of inducing quite rapid local changes (by setting oil-fields ablaze, for example). The cryosphere includes sea-ice, glaciers, ice-sheets, and snow-cover. Snow-cover and sea-ice can change quite rapidly, but glaciers and ice-sheets may require decades to millennia. The geosphere, which is the solid earth itself, changes slowest of all. A single volcano might affect the climate rapidly, but only temporarily. Aspects of the geosphere which affect long-term climate, such as the location of continents and mountain ranges, take millions of years to change. All five of these components are important for climate modelling. The first four are expected to change in response to greenhouse gas emissions, while changes in the geosphere must be understood for the study of past climates. There are many different types of climate models. The simplest models can be run on a personal computer and consist of nothing more than a small number of equations relating key climatic variables. Such models are designed to focus on one particular time-scale, or to investigate a specific phenomenon while making many assumptions concerning the behaviour of the rest of the system. The relevance of any such model to the real climate depends on whether or not the assumptions on which it is based are more-or-less correct. In general, the more complex a model, the less it assumes, and the more easily its individual assumptions can be tested against observations and other models. But all models must assume something: the climate system is far too complex for us to feed the laws of physics into a computer and ask it what the climate will be like in 2030. The most sophisticated climate models are "general circulation models" (GCMs). Consisting of hundreds of inter-related mathematical equations that are processed on super-computers, these models are adapted from those used for weather-forecasting. The main adaptation is that climate-model GCMs have a coarser "grid resolution" that allows them to be run for a large number of model-years with the computers available. The model-atmosphere both receives information (about such things as sea-surface temperature) and produces output (such as winds, temperatures, and rainfall) as sets of numbers on a grid. In a typical climate model, the horizontal spacing between grid-points might be several hundred kilometres, and there might be only ten to twenty vertical levels between the earth’s surface and the outer atmosphere.A climate model cannot "see" any processes that occur on scales smaller than the model’s grid resolution. But small-scale processes such as cloud-formation cannot be ignored, because they do affect the system as a whole (just as the bacteria you can’t see can still make you ill). To get around this problem, small-scale processes are "parameterised". In the case of clouds, for example, this means developing a computer routine that converts large-scale factors affecting clouds (such as moisture levels) into large-scale quantities affected by clouds (such as total rainfall in a grid-box) without ever specifying exactly where the clouds are or what they look like. Parameterisations can be quite sophisticated, but they are often difficult to test, mainly because it is often hard to tell if a parameterisation scheme is doing the right things for the right reasons.On time-scales longer than a few days, interactions between the atmosphere and the oceans become critical. Until recently, "oceans" in climate models were very simple, in effect just large "slabs".They were able to transport heat away from the equator just as the real oceans do, but they were not able to vary much from year to year, apart from simply warming up or cooling down. Recently, modelers have undertaken a small number of studies involving coupled atmosphere-ocean GCMs that attempted to quantify the importance of atmosphere-ocean interactions for climate change. One problem that modelers face is that including even a somewhat realistic model-ocean substantially increases the computing time required for a climate modelling experiment. This is because an ocean model must have a relatively high resolution to distinguish ocean currents. In addition, oceans are slow-moving and so an ocean model tends to take a long time to settle down to a steady state from which a climate-change simulation may begin.All major climate models explicitly calculate snow and sea-ice cover. Details of these calculations differ, but every model must allow the cryosphere to respond to climatic changes because this response plays an important role in determining how much solar radiation is absorbed by the earth. Snow and ice reflect heat very effectively (which is why patches of snow survive long after temperatures rise above freezing), so if warming leads to less snow, then more heat will be absorbed, which warms the planet further. Modelers have begun to experiment with "interactive land biospheres". These experiments are mainly designed to allow for the impact of changing soil-moisture on vegetation. This impact in turn affects how rough the surface will be, how much heat it will absorb, and how fast soil-moisture will evaporate. The diversity of the biosphere makes it is difficult to model, and biosphere-climate interactions remain an area of considerable uncertainty. Depending on what they are used for, models can also include interactive atmospheric chemistry, ocean biology, and other processes. Climate modelling studies seldom attempt to include everything in a single experiment. Although interactions between the different climate components are clearly important, in many areas scientists are still at the stage of studying the different components of the climate individually through specific, focused experiments. See also Fact Sheet 15: "Are Climate Models Reliable?"

16 Are climate models reliable ?

Defining "success" for a climate model is surprisingly difficult. Inevitably, models are good at simulating some aspects of the climate system and less good at others. But since we depend on these models to predict the possible consequences of emitting greenhouse gases, we would clearly like to know how reliable they are overall. The reliability of models can be estimated by comparing model-climates with the present climate, with reconstructions of past climates, and with each other. Comparing models with each other may be instructive, but it cannot be an absolute guide to their accuracy since different models generally rely on similar approximations and work in similar ways. If two models give the same answer, it may be because they both contain the same errors. Consequently, short of waiting until after climate change has occurred, the best guide we have for judging model reliability is to compare model results with observations of present and past climates.Our lack of knowledge about the real climate makes it difficult to verify models. It is hard to say if a model has the answer right if we don’t know what the right answer is. Observational data on many key climatic variables is extremely limited, particularly for sluggish components of the system such as deep ocean currents. Slow components must be observed over long periods of time before they reveal their full range of behaviour, and the systematic records that do exist are all rather recent. Even things like rain- and snowfall rates are not known precisely, particularly in unpopulated regions and the open oceans. There is no simple universal measure of model "accuracy". Because the climate system is much too complicated to be modelled from "first principles" alone, scientists must use observational data to develop climate models. But this means that they could always "force" a model to reproduce a particular set of observations by telling it exactly what those observations are and making it flexible enough to fit them exactly. Obviously, this is not what climate modelling is about. But it shows that defining what it is for a model to be "good" can become quite complicated. We cannot simply ask "how well does model A reproduce the present climate?" Instead - When testing a climate model, we must ask "how much did we have to tell model A about the present climate, and how well did it reproduce the rest?" If scientists drive a modern atmosphere model with observed sea-surface temperatures, for example, it will simulate large-scale patterns of winds, surface pressure, air-temperature and precipitation with considerable skill. Likewise, if they "blow" and "rain" observed winds and precipitation onto an ocean model, it can reproduce the major ocean currents. Clearly, there is much that is correct in these models. But if an atmosphere and an ocean model are coupled together without any adjustment to make sure they "match", the coupled model tends to evolve towards a much less realistic climate. This is to be expected, and does not necessarily mean the models are all that bad, for the following reason:Feedbacks between different climate components amplify some model errors . . . If, for example, the model-atmosphere happens to produce winds that are slightly too weak over a particular part of the ocean, the sea-surface temperature would be affected. This in turn might, under certain circumstances, make the winds weaker still. So errors that would be almost undetectable if either the atmosphere or ocean were modelled independently can become quite serious when the two are coupled together. There are methods of overcoming such feedback loops, but they tend to involve somewhat arbitrary-looking fixes. but other errors are self-correcting. Models are designed to conserve quantities such as energy and water. If it rains too much in a particular grid-box, the model will run out of water there and the rainfall will shut off. Listing all possible sources of error in a climate model would therefore not only be difficult, but might well give the misleading impression that the models are worse than they really are.There are several ways of double-checking whether advances in modelling are really improving model reliability. Increasing the grid-resolution of an atmosphere or ocean model, or introducing more realistic representations of particular processes, generally (but not always) makes the climate which it simulates more realistic. Moreover, changes in models often affect climate simulations in ways that are understandable in physical, real-world terms; increasing an ocean-model’s resolution, for example, makes the simulated Gulf Stream stronger, and thus enhances heat transport to the North Atlantic. The fact that climate simulations improve in a reasonably consistent and comprehensible way suggests that scientists are making successively better approximations to the real climate system. Models are becoming more reliable. New models for simulating the present climate are also simulating year-to-year climate variations and key features of past climates with increasing realism. This is extremely encouraging, since it suggests that models are already capable of, and are becoming better at, capturing the essential features of the long time-scale variations that matter for climate change. Climate models are scientific tools, not crystal balls. Models don’t serve up answers on a plate: the work involved in interpreting the results from a climate simulation using a large climate model is often greater than the work involved in setting up the simulation in the first place. A modern climate model is a remarkable tool, and definitely the best method scientists have for investigating climate change. But there is nothing magic to it. It remains a tool, to be interpreted carefully and responsibly.

17 What happens if we double CO2 in a climate model?

It is much easier to isolate causes and effects in a climate model than in the climate system itself. The advantage of computer-based models of the climate is that, unlike "experiments" on the real thing, experiments on a model climate can be repeated. This makes it possible to work out exactly what is causing what.The first (and most difficult) step in modelling climate change is obtaining a stable and realistic model of the present climate. To be stable, the different processes going on in a climate model must balance each other: for example, the overall rate of evaporation must balance overall rainfall. Crucially, the overall rate at which energy is absorbed from the sun must balance the rate at which energy is emitted into space. There may be temporary imbalances, but they must average out over time.In an "equilibrium-response" experiment, scientists begin by setting up a climate model with concentrations of greenhouse gases (GHGs) at their present real-world levels. They allow this model to settle down to a steady state, in which it is neither warming up nor cooling down in the long term.This becomes the "control" climate. They then double the level of carbon dioxide (CO2) and allow the model to settle down again to a new steady state. This is the "perturbed" climate. If the model is run for long enough (or better still, the experiment is repeated often enough), the effects of chance fluctuations average out. Differences between the control and perturbed model climates can then be attributed to that one change. In the real climate system, of course, CO2 will not double overnight, but will increase gradually. The latest modelling experiments take this into account, but it is easier to understand causes and effects in an equilibrium-response experiment.The first thing that happens when CO2 is doubled is that less energy in the form of radiation escapes to space. Before doubling CO2, the model climate system absorbs energy from the sun at exactly the same rate as it emits energy to space. This is indicated in Figure A, where the incoming (left) and outgoing (right) arrows are the same width. Hence the control climate is in a steady state. Immediately after doubling CO2 (Figure B), the model climate is absorbing the same amount of energy from the sun as before (the incoming arrow in B is the same width as in A), but less energy is escaping to space (the outgoing arrow is narrower). The global energy budget is unbalanced. Energy is accumulating at the surface and in the lower atmosphere, so the model climate is no longer in a steady state. The model climate must adjust to restore the balance between incoming and outgoing energy.Energy cannot just accumulate indefinitely. The climate must change to get rid of the excess. The simplest such change is global warming itself. By becoming warmer, the earth's surface and the lower atmosphere shed more energy to space, acting to balance the energy budget. If global temperatures were the only variable to change, a warming of only 1.1 C would be enough to offset the effect of doubling CO2. Figure C shows this scenario: after doubling CO2, nothing else has been allowed to change in the model climate apart from temperature. Incoming energy (the left-hand arrow) is unchanged from Figure B, while outgoing energy increases. The thicker right-hand arrow shows the energy emitted after a 1.1 C warming. Since the incoming and outgoing arrows now equal each other, this model would be stable from the point of view of the global energy budget. It would, however, not be stable internally. No realistic model climate can warm up by 1.1 C without other aspects of the climate also changing.Other variables besides temperature must change. As the surface and atmosphere warm, the model atmosphere becomes moister, which in turn causes more warming - a positive feedback effect. Here's how: because water vapour is itself a greenhouse gas, the extra moisture traps more energy (D). So yet more warming is required (up to 1.8 C)1 to get rid of this additional energy and bring the budget back into balance (E). Snow and sea-ice melt, which enhances warming near the poles. Snow and sea-ice reflect the sun's energy very effectively so, as they melt, more energy is absorbed at the earth's surface. Still more warming (up to 2.2 C in this model) is required to balance this effect. Introducing the snow/ice feedback also affects the amount of energy trapped by water vapour. Feedbacks feed back upon feedbacks, which is why climate modelling is so difficult. In most models cloud cover increases in a warmer climate. This affects the energy budget in two opposing ways. Clouds reflect sunlight, reducing the amount of energy reaching the surface. They also act as a "blanket", reducing the earth's energy losses to space. As the total cloud cover increases, the first effect acts to reduce the warming (a negative feedback) while the second effect acts to increase it (positive feedback). Clouds are a major source of uncertainty. If clouds are allowed to change (and changes in sea-ice are suppressed), different climate models give answers ranging from 1.5 to 4.5 C for the warming due to doubling CO2. If the effects of cloud feedbacks are eliminated, this range is reduced to 1.7-2.3 C.2 Many other feedbacks, particularly those involving chemistry and biology, may also be important. The most important chemical feedbacks are interactions among greenhouse gases in the atmosphere. Biological feedbacks can, like clouds, work both ways. Plants (both on land and in the sea) may absorb more CO2 in a warmer, CO2-rich world, or they may absorb less CO2 as the world becomes cloudier. Other important feedbacks include  ocean-atmosphere interactions and the possibility that methane will be released from melting tundra. Improving the representation of feedbacks in climate models,and checking them against observations, is probably the most important area of climate modelling research at present. Notes:1 J. F. B. Mitchell, "The Greenhouse Effect and Climate Change", Reviews of Geophysics, Vol. 27, pp.115-139, (1989).2 R. D. Cess et al, "Interpretation of cloud-climate feedback as produced by 14 atmospheric general circulation models", Science, Vol. 245, pp. 513-516, (1989). Figures derived from full and clear-sky sensitivity parameters (table 2), assuming radiative forcing due to doubling CO2 of 4 W/m2.

18 Natural climate variablility vs. man-made climate change

Understanding natural climate variability is essential for understanding man-made climate change. When scientists warn that emissions of carbon dioxide and other greenhouse gases (GHGs) may cause a global warming of 1.5-4.5 C over the next 100 years, one possible reaction is: "So what? Temperatures where I live vary by much more than that in the course of a day." To understand the true significance of this temperature change, we must distinguish between natural weather cycles (such as the changing seasons), transitory climate variations (such as a temporary drought), and long-term climatic change.The earth's climate varies naturally for many reasons. Put simply, a "climate variation" is a change in the average weather for a particular time of year; for example, winters becoming warmer. Such variations can affect a small region or the entire planet. Their causes range from completely unpredictable events like volcanic eruptions (which have mainly local effects) to more regular phenomena such as "El Niño" (a warming of the surface waters of the tropical Pacific that occurs every three to five years, temporarily affecting weather world-wide). Natural climate variability hinders the detection of any man-made warming trend. Figures A and B show past variations in the global mean temperature inferred from direct measurements (A) and from the analysis of ice-cores (B). While the state of the climate clearly involves much more than just global temperature, changes in global temperatures do indicate the scale of different climatic events, both natural and man-made. The range of natural year-to-year temperature variations is quite similar to the size of the warming that appears to have occurred over the past century (0.3-0.6 C). Moreover, the 16th to 18th centuries appear to have been unusually cold, and the climate may still be recovering from that time. So scientists cannot yet claim to have found an unambiguous temperature-related "greenhouse signal". If the models are right, however... The projected warming of 1.5-4.5 C over the coming century would be larger than any natural climate variation since the dawn of human civilisation. The figures also show the current "best guess" of the man-made warming to the year 2100, assuming no action is taken to reduce greenhouse gas emissions. It dwarfs recent century-time-scale variations. Moreover, unless GHG emissions are drastically reduced this warming may last indefinitely. Such a one-way transition could be much more difficult to deal with than a natural fluctuation such as a one- or two-year drought. The climate of the past 9,000 years appears to have been exceptionally stable. For two million years, the earth's climate has been dominated by periodic ice-ages, each lasting tens of thousands of years. The ice-ages are separated by warmer "interglacial" periods, such as the one we are in now. Large climate fluctuations on time-scales of decades to centuries have long been known to occur during ice-ages, presumably due to sudden surges or collapses of the massive ice sheets. Recent evidence from ice-core drilling in Greenland indicates that similar fluctuations also occurred during the previous interglacial period, possibly due to rapid changes in ocean circulation. No one knows why these fluctuations have not occurred during the current interglacial period, allowing a (possibly essential) "window" of climatic stability for the development of human civilisation. Nor do we know how man-made GHG emissions might affect this stability. The projected man-made warming represents a significant change even on the longest geological time-scales. One hundred million years ago, in the time of the dinosaurs, the earth was 5-15 C warmer than it is today, probably because the continents were arranged differently. Figure C shows, very schematically, how the earth has cooled since then. The projected man-made warming over the next 100 years is clearly visible on a plot of global temperature over the past 100 million years. Abrupt climatic variations in the past appear to have been traumatic for life on earth. Because of the speed of man-made climate change, it is difficult to compare it directly with past climatic events. The closest analogies are with climatic shifts associated with abrupt changes in ocean circulation, or the sudden global cooling at the end of the dinosaur era that may have been caused by a large asteroid colliding with the earth. These abrupt changes seem, on several occasions, to have coincided with mass extinctions which wiped out a significant fraction of the planet's animal and plant species. This supports evidence from climate impact models indicating that climate change over the next 100 years may place a considerable strain on natural eco-systems and climate-dependent human communities.

19 How researchers develop regional scenarios of climate change

It is now widely accepted among scientists that mankind's greenhouse gas emissions will lead to global warming. However, researchers have less confidence about how temperatures may change at the regional level. They are also uncertain about how rainfall and other climate variables will be affected. It seems likely that some areas of the globe will become wetter while others will become drier. The distribution and severity of extreme events may also change: storms may become more frequent and droughts more intense. In everyday life, the effects of global warming will be felt at the regional and local level. Researchers are therefore exploring how to develop scenarios of regional climate change. One recent study1 on the Mediterranean basin illustrates a possible approach. The researchers produced annual and seasonal scenarios of the changes in regional mean temperature and precipitation levels that might be expected from a 1 C change in the global mean temperature. (Scenarios are not forecasts: they are internally-consistent pictures of a plausible future.) General Circulation Models (GCMs) are the best source of data available to researchers for developing regional scenarios. GCMs are three-dimensional mathematical models of the atmosphere's circulation that are run on super-computers. They receive inputs in the form of information on climate variables and processes, make calculations over a given number of "model-years", and then produce temperatures and other data as sets of numbers on a grid. Because of the complexity of these models and restrictions imposed by limited computing speed and power, various simplifying assumptions must be made. GCMs can simulate the effect that an increase in atmospheric concentrations of greenhouse gases will have on global temperatures. For an equilibrium model, the first step is to make a control run using nominal pre-industrial atmospheric CO2 concentrations. Then a perturbed run is made, usually based on a doubling of CO2 concentrations. In each run, the model is allowed to operate until it reaches an equilibrium, and only then are the results recorded and compared. In the real climate system, at any particular time the actual change in climate would lag behind the corresponding equilibrium change for any given CO2 level, largely because of the thermal inertia of the oceans. However, the pattern of actual change should be similar enough to the equilibrium pattern, except in regions such as the North Atlantic and high southern latitudes where the ocean's thermal inertia is large. Because each GCM has a different climate sensitivity, the global warming which occurs due to a doubling of CO2 varies from model to model. The performance of a particular GCM can be judged by comparing the results of the control run with the actual present-day climate. Since none of the four models considered by the Mediterranean study consistently out-performed the others in simulating present-day climate, the researchers combined their results to produce a single composite scenario for each climate variable.2 To prevent the GCM with the greatest sensitivity from dominating the scenarios, they first had to standardise the model results. This was done by calculating the climate change occurring in each model as a result of a 1 C increase in global mean temperature.The output from GCMs can be used directly to construct regional scenarios. Figure A illustrates how 1 C of global warming might affect the annual mean temperature over the Mediterranean Basin. This scenario indicates a change in annual mean temperature of between 0.9 C and 1.2 C over the land areas close to the Mediterranean Sea. Unfortunately, the figure also confirms that the spatial resolution of theoutput from the GCMs used in the Mediterranean study is too coarse for constructing detailed regional scenarios.To develop more detailed regional scenarios, modelers can combine the GCM results with output from statistical models.3 This is done by constructing a statistical model to explain the observed temperature or precipitation at a meteorological station in terms of a range of regionally-averaged climate variables. Then, by substituting GCM results for the regionally-averaged variables, the change in temperature at the meteorological station due to global warming is estimated. When this procedure is repeated for a network of sites, the resolution of the resulting regional scenario is limited only by the density of sites. It will reflect, for example, small climatic variations due to altitude or proximity to the sea. Figure B shows the scenario for the change in annual mean temperature per 1 C global warming using this method for 248 meteorological stations. It has a much finer resolution than does Figure A.For further reading:Houghton, J. T. et al., eds, "Climate Change: the Scientific Assessment", Cambridge University Press,Cambridge, 1990. Houghton, J. T. et al., eds., "Climate Change 1992: the  supplementary Report to the IPCC Scientific Assessment", Cambridge University Press, Cambridge, 1992.Notes:1 J.P. Palutikof et al., 'Regional Scenarios in Climate in the Medi-terranean Basin due to Global Greenhouse Gas Warming. MAP Technical Report Series No. 66', UNEP, Athens, 1992. The study was commissioned by UNEP and carried out by the Climate Research Unit of the University of East Anglia.2 Following the method of B. D. Santer et al., 'MP1 Report No. 47', Max Planck Institute for Meteorology,Hamburg. The model runs used are described in the following papers: UKMO-C. A. Wilson and J. F. B.Mitchell, 'J. Geophys. Res.', 92D11, 13315-13343; GISS-J. E. Hansen et al., in 'Climate Processes and Climate Sensitivity' J. E. Hansen and T. Taka-hashi, eds., American Geophysical Union, Washington, 1884, 130-163; GFDL-R.T. Wether-ald and S. Manabe, 'Climate Change', 8, 5-23; M.E. Schlesinger and Z.-C. Zhao, 'J. Climate', 2, 459-495.3 Following the method developed by J. W. Kim et al., 'Mon. Wea. Rev.', 112, 2069-2077 and T. M. L.Wigley et al., 'J. Geophys. Res.', 95D2, 1943-1953.

20 Oceans and the carbon cycle

The oceans influence the climate by absorbing and storing carbon dioxide. Climate change is caused by the accumulation of man-made carbon dioxide (CO2) and other greenhouse gases in the atmosphere. The rate of accumulation depends on how much CO2 mankind emits and how much of this excess CO2 is absorbed by plants and soil or is transported down into the ocean depths by plankton (microscopic plants and animals). Scientists believe that the oceans currently absorb 30-50% of the CO2 produced by the burning of fossil fuel. If they did not soak up any CO2, atmospheric CO2 levels would be much higher than the current level of 355 parts per million by volume (ppmv) - probably around 500-600 ppmv.Plankton influence the exchange of gases between the atmosphere and the sea. In any given region, the relative amounts of CO2 contained in the atmosphere and dissolved in the ocean's surface layer determine whether the ocean-water emits or absorbs gas. The amount of gas dissolved in the water is in turn influenced by the amount of phytoplankton (microscopic plants, particularly algae), which consume CO2 during photosynthesis. Phytoplankton activity occurs mostly within the first 50 metres of the surface and, although oceanographers don't fully understand why, varies widely according to the season and location. Some areas of the ocean do not receive enough light or are too cold. Other areas appear to lack the nutrients or trace minerals required for life, or zooplankton (microscopic animals) that feed on phytoplankton so limit the population growth of the latter that not all of the available nutrients are consumed. Rather like a pump, plankton transport gases and nutrients from the ocean surface to the deep. Their role in the carbon cycle is quite different from that of trees and other land plants, which actually absorb CO2 and serve as a storehouse, or "sink", of carbon. Instead, ocean life absorbs CO2 during photosynthesis and, while most of the gas escapes within about a year, some of it is transported down into the deep ocean via dead plants, body parts, faeces, and other sinking materials. The CO2 is then released into the water as the materials decay, and most of it becomes absorbed in the sea-water by combining chemically with water molecules (H2O). Although a small but possibly significant percentage of the sinking organic material becomes buried in the ocean sediment, most of the dissolved carbon dioxide is eventually returned to the surface via ocean currents - but this can take centuries or millennia. Measuring the level of plankton activity in the ocean is difficult. The rate at which plankton consume carbon dioxide and convert it into sugars for producing tissue and energy varies enormously. This makes it difficult to sample and estimate their annual consumption of CO2. The enormous expanse and remoteness of the oceans (few oceanographers want to go to Antarctica in the middle of winter) also hampers sampling. Satellite pictures of chlorophyll (cell pigment that converts sunlight into energy) give a general idea of the amount of phytoplankton present, and oceanographers hope that future satellite measurements will further clarify the picture. Climate change will affect plankton, and vice versa. Warmer temperatures may benefit some species and hurt others. Changes in carbon dioxide levels may not have a direct impact, but related "feedback

21 How the oceans influence climate

The oceans influence climate over long and short time-scales. On the longest time-scale of geologic time, the shape and location of the continents helps to determine the oceans' circulation patterns. Since continental plates drift at about 5 cm per year and mountain ranges rise by about 1 mm, it usually takes millions of years for new land formations to change the oceans. Patterns of ocean circulation and up-welling can also change much more rapidly, resulting in climate variations and fluctuations on a human time-scale. Records of global and, in particular, regional climate show periods lasting from years to centuries during which the climate was systematically different from earlier and later periods. Scientists believe that this behaviour is related to changes in the way the oceans store and transport heat, although the precise causes of these changes are not always clear. The oceans and the atmosphere are tightly linked and together form the most dynamic component of the climate system. Changes in external factors such the sun's energy, the distribution of various plant species, or the emission of greenhouse gases into the atmosphere can alter the temperature and circulation patterns of the atmosphere-ocean system. Because the atmosphere and oceans are turbulent, they can also generate their own internal fluctuations. Short-term fluctuations in wind or temperature (that is, weather) can directly influence the currents and temperature of the underlying ocean, while oceanic fluctuations can magnify, diminish, or modify atmospheric fluctuations. The oceans play a critical role in storing heat and carbon). When the earth's surface cools or is heated by the sun, the temperature change is greater - and faster - over the land than over the oceans. Because it is a fluid, the ocean diffuses the effects of a temperature change for great distances via vertical mixing and convective movements. The solid land cannot, so the sun's heat penetrates only the thin, upper crust. One consequence of the ocean's ability to absorb more heat is that when an area of ocean becomes warmer or cooler than usual, it takes much longer for that area to revert to "normal" than it would for a land area. This also explains why "maritime" climates tend to be less extreme than "continental" ones, with smaller day-night and winter-summer differences. The ocean's waters are constantly being moved about by powerful currents. Surface currents are largely wind-driven, although the rotation of the earth, the presence of continents, and the oceans' internal dynamics also have a strong influence. Deep-ocean flow (and, to a lesser extent, surface flow) is driven by density differences produced by heating and cooling and by precipitation and evaporation (cool saline water is denser than warm fresh-water). The behaviour of the atmosphere strongly affects these density differences. For example, clouds can cool the sea by blocking the warming rays of the sun or reduce surface salinity by bringing rain. The wind can influence evaporation rates by blowing more strongly or more weakly. These currents influence the climate by transporting heat. Horizontal currents, particularly those moving north or south, can carry warmed or cooled water as far as several thousand kilometres. The displaced water can then warm or cool the air and, indirectly, the land over which this air blows. For example, water from the tropical and subtropical Atlantic (including some from the Gulf of Mexico) moves north through the Atlantic in a current popularly (if misleadingly) called the "Gulf Stream". There it bathes the shores of Western Europe, producing a climate that is surprisingly mild for that latitude. In addition to currents, up-wellings of cold water in places where the wind blows surface water away can also affect climate. Thus San Francisco, influenced by coastal up-welling, is hardly warmer than Dublin, which is influenced by the Gulf Stream, despite being over 1,600 km further south. Currents involved in "deep-water formation" are particularly important for climate. In winter,surface cooling causes water to become more dense. (While fresh-water that is cooled starts to expand at temperatures below 4 C, salt-water continues to compress all the way down to its freezing point of -2 C.) In areas where evaporation exceeds precipitation, the resulting rise in salinity also increases density. When the surface water becomes denser than the underlying water, "convective overturning" occurs and the dense surface water mixes downwards. In certain places this downward mixing can occasionally extend all the way to the bottom, even in deep oceans. The dense, deep water thus formed spreads out over the whole ocean. As a result, when downward mixing takes place at high latitudes it creates a circulation pattern in which warm water from tropical and subtropical regions moves poleward, surrenders heat to the atmosphere, cools and sinks, and flows back towards the equator. The net result is a transport of heat poleward. An apparently small change in just one aspect of the ocean's behaviour can produce major climate variations over large areas of the earth. The areas of cold-water formation are one known example of this possibly wide-spread phenomenon. Although more research is needed, there is some agreement among oceanographers that, for the entire area north of 30 N latitude, the ocean's poleward transport of heat is the equivalent of about 15 watts per square metre of the earth's surface (W/m2). This can be compared with some 200 W/m2 from direct sunshine, and about 6 W/m2 for what climate change models predict will happen if the atmospheric concentration of carbon dioxide doubles. Recent observations, ocean core records, and some modelling results indicate that North Atlantic deep-water formation and its associated ocean heat flow fluctuate substantially over time-scales ranging from years to millennia. The system is vulnerable because even a relatively small decrease in surface salinity prevents water - no matter how cold it is - from sinking. This could occur if there is a flood of fresh-water run-off from the Arctic due to global warming.