22.2.1 Review of basic greenhouse physics
The basic physics of the greenhouse effect was described in Chapter 14. As the amount of infrared-absorbing gases is increased, Earth’s atmosphere becomes more opaque to infrared photons. The altitude above the surface at which such photons are finally free to escape (themean radiating level) therefore moves upward, toward lower air density, as greenhouse gas con- centration increases. However, as Figure 22.3 shows, because
the temperature falls with altitude in the troposphere, the new mean radiating level is colder than the old one, and hence less efficient at removing energy. Its temperature must increase, rais- ing the entire temperature profile of the troposphere. In this way, increasing greenhouse gas concentrations raise the mean surface temperature of Earth.
This effect can be expressed in terms of the “mean radiative forcing” of greenhouse gases, which is the change due to green- house gases of the net irradiance (solar photons minus infrared photons) the atmosphere radiates as measured at tropopause.
While this sounds complicated, it effectively means the added power associated with more greenhouse gases, which cause the atmosphere to absorb more of the infrared energy mov- ing upward from the surface. The radiative forcing thanks to all greenhouse gases has risen by 2.8 watts per square meter rela- tive to what is calculated for 1750, and by 1.1 watts per square meter relative to what was measured in 1980. While this seems small relative to the total solar input of roughly 1,300 watts per square meter, remember that the atmospheric temperatures are the result of a balance between incoming and outgoing solar radiation, and thus small changes in the amount of infrared radi- ation the atmosphere absorbs have a big effect. Further, note that 40% of the increase in the mean radiative forcing since 1750 has occurred in just the last 30 years (1980–2010).
The basic physics of the greenhouse process described above is straightforward enough that there is little argument about its validity. We know, for example, that Earth’s neighboring planet, Venus, receives less sunlight at and near its surface than does Earth because of a layer of bright, reflecting clouds. However, the surface temperature of Venus is over twice that of Earth’s, and the atmosphere is possessed of a carbon dioxide pressure of 90 bars, 300,000 times the amount of CO2in our atmosphere.
It is not too great a leap to infer that the Venusian atmosphere is in a state in which the enormous amounts of carbon dioxide create a greenhouse effect much larger than Earth’s, and models show that the surface temperature and CO2abundance are indeed consistent with each other.
22.2.2 Some complications
As simple as the basic physical concept is, it does not fully describe the actual situation. The most fundamental complica- tion is that water vapor is also a greenhouse gas, but its abun- dance in the atmosphere depends on the global mean surface temperature through evaporation from the ocean and rainout in precipitation events on land, ice, and sea. Cloud forma- tion (see below) complicates any direct relationship between increases in surface temperature and consequent increase in water abundance. However, it appears that as other greenhouse gases increase the global average temperature, a slight positive feedback occurs through an increase in the amount of atmo- spheric water vapor.
Another complication is that radiation (transport of photons) is not the sole means of the movement of heat energy outward.
Particularly in the lower part of the atmosphere, the temperature profile becomes so steep (decreases so sharply with altitude) that bulk air movement (that is, convection) plays an important role.
Dry convectioninvolves bulk movement of air without conden- sation of water to form clouds; it occurs in the lowermost part
of the atmosphere and particularly in dry regions.Moist convec- tionincludes the effects of cloud condensation and evaporation, which add and delete heat from the surrounding air. Cloud for- mation most often is a result of air containing water vapor rising, expanding as the surrounding pressure drops with altitude, and then cooling until the air can no longer hold the water as a gas.
Thedew pointis thus reached, and water condenses out to form small liquid drops or solid ice particles.
Moist convection is a sufficiently energetic process that it alters the environment around it and the consequent transport of energy. Large amounts of water in an atmosphere initially unsta- ble (tending toward bulk air motions to remove heat) can create large thunderstorm complexes, in which updrafts and down- drafts may reach all the way up to and beyond the tropopause (defined in Chapter 15). This is particularly the case in the trop- ics, but large storm complexes also dominate weather in mid- latitude continental regions. The convective transport of heat, particularly involving moist convection, alters the relationship between greenhouse gas increase and the temperature response of the atmosphere; by how much (and even in what direction) remains a matter of dispute.
Formation of clouds also alters the radiative balance of the atmosphere, aside from the effects of moist convection. Clouds can reflect, scatter, and even absorb incoming solar visible radi- ation; they also may absorb infrared radiation moving upward from the deeper atmosphere. The overall effect of clouds on global climate is complicated. The difficulty arises from the wide range in shapes of clouds, size of the cloud droplets or ice particles, the breadth of altitudes over which clouds form and extend, and conditions under which precipitation (rain, hail, sleet, or snow) forms. Some cloud types may lead to a net warming of the atmosphere, whereas others will cool it. Hence, if global temperatures increase because of enhanced greenhouse gases, and the resulting increased moisture (from more vigor- ous evaporation of ocean water) creates more cloudiness, the net effect of that cloudiness depends largely on the types of clouds and their mean altitude. Recent satellite and aircraft measure- ments of the amount of visible and infrared radiation coming out of, and moved around within, clouds are beginning to untangle these very complicated effects.
Snow, continental ice sheets, and sea ice provide very highly reflective surfaces that prevent much sunlight from being absorbed at high latitudes on Earth’s surface. As global tem- peratures increase, the amounts of land and sea ice and snow will decrease, causing more sunlight to be absorbed and ampli- fying the greenhouse warming. How much of an amplification will occur depends on the details of the response of the ice and snow to warming. Increased precipitation at high latitude, another likely result of warming, could actually increase snow and ice cover in winter at high latitudes and/or altitudes, provid- ing a moderating effect to the amplification.
Variability in the output of the Sun affects the amount of energy the atmosphere must transport back out, and has the potential to obscure the signature of human-induced global warming. Measurements of the Sun’s luminosity taken over the past couple of decades show that it has varied only by plus or minus 0.02%. Compared to the effects of increased CO2 over the same period, this number is quite small and, although some climate amplifications of the solar variability are possible, they
aerosols H2O, N2, O2, CO2, O3, etc.
changes in solar inputs
atmosphere
changes in the atmosphere:
composition, circulation
changes in the hydrological cycle
clouds
air–biomass coupling
biomass land–biomass
coupling rivers
lakes
changes in/on the land surface:
orography, land use, vegetation, ecosystems land
changes in ocean:
circulation, biogeochemistry air–ice
coupling heat
exchange wind stress
precipitation- evaporation
terrestrial radiation
human influences
ice-ocean coupling
ocean sea–ice
Figure 22.4Processes affecting the nature of climate today, with an emphasis on the changes that might result from human influences.Wind stress is the movement of ocean water caused by the action of wind;biomassrefers to biological organisms both living and dead, that interact chemically with the atmosphere, land, and oceans. From Trenberthet al.(1996).
are unlikely to reverse or dominate global warming associated with increasing carbon dioxide. On longer timescales, the Sun’s luminosity varies more significantly (Chapter 14), but projec- tions of human-induced global warming are concerned primarily with the next half-century, a time not much longer than that over which detailed solar measurements have been made.
Perhaps the most important uncertainty lies in the role of the oceans. A thorough discussion of this is deferred to section 22.5, because of its complexity. Figure 22.4 illustrates graphically how the processes discussed above fit together and emphasizes that climate is not simply a matter of the vertical structure of the atmosphere, but also of what is happening from place to place on Earth’s surface and in its oceans. We know from our expe- rience with weather patterns that the three-dimensional nature of the planet is important. To capture this aspect of the prob- lem requires rather involved computer models, to which we now turn.
22.2.3 General circulation models
One-dimensional climate models simulate the transfer of energy and matter only in one direction, namely, up and down. How- ever, on a planet, energy and matter also move sideways in the atmosphere and on the surface. It is useful to define the
sideways direction parallel to a line of latitude aszonal, and parallel to a line of longitude asmeridional. Because Earth is roughly spherical, different latitudes receive varying amounts of sunlight; even though Earth’s axis is tilted, the equator receives the largest amount of heat averaged over the year. As a con- sequence, heat tends to be redistributed by the oceans and the atmosphere in a meridional direction, that is, from the equator to pole. Warm tropical air rises, moves away from the equator, and sinks; this cycle is repeated at higher latitudes.
The Earth also spins on its axis, and this spinning motion modulates the transport of heat from equator to pole. Sinking air in the northern hemisphere is forced to spin clockwise, and in the southern hemisphere counterclockwise. Regions in which air is drawn inward by low pressure, forced to rise and form clouds and precipitation, will rotate counterclockwise in the north and clockwise in the south.
These systems of high and low pressure produce much of the weather with which we are familiar at middle and low latitudes.
Their sense of rotation, induced by Earth’s spin, interacts with the distribution and shape of continents to produce complex patterns. Low pressure spiraling counterclockwise as it moves eastward across the central United States draws moisture off the Gulf of Mexico to produce the well-known severe thunderstorms that often plague Texas, Oklahoma, Arkansas, and other midland
states. The positions of high and low pressure systems in the Pacific and Indian Oceans determine each summer the strength of the south Asian monsoon rains, critical to food production cycles for billions of human beings.
To simulate such complex weather patterns, computer models must do more than calculate how photons are absorbed and re- emitted on their way out of the Earth’s atmosphere. They must also keep track of how energy (heat), moisture, and bulk air flow from one region to another. Models that do this are called general circulation models, or GCMs. The strategy is to divide Earth into a checkerboard in which each square, or grid point, is as small as possible; smaller gridding requires faster computers to handle the more numerous grid points. Newton’s laws of motion, along with the laws of thermodynamics, are applied to the air, water, and heat in each grid, and both matter and energy are allowed to flow from one grid point to another. Sunlight shines according to latitude, season, time of day, and amount of cloudiness. In this way the flow of moisture, winds, and heat around Earth can be simulated on large, fast computers.
Such GCMs, relying on detailed temperature, wind, pressure, and moisture information from thousands of weather stations worldwide, are used to predict weather several days or more in advance. General circulation models have also been adapted to predict atmospheric circulation patterns on other planets, as well as the nature of the climate at earlier times in Earth’s history. They are the basic computational tool for evaluating climate change caused by increasing abundance of greenhouse gases.
As carefully constructed as they are, GCMs have limitations.
The first of these is intrinsic to the nature of climate itself. The ocean, atmosphere, and land form a coupled,nonlinearphys- ical system. In recent decades the properties of such systems have been investigated and found to exhibit chaos. Chaos does not imply complete randomness (Chapter 19). However, such systems can evolve into many different states, unlike simpler systems. A simple system, started out in two slightly differ- ent configurations, will diverge rather slowly in appearance. A chaotic system, started out in two different states, willexponen- tiallydiverge in its characteristics – an almost explosive parting of the ways between the two slightly different starting states.
Insight into the difference between a simple and a chaotic sys- tem is not easy to gain, but Figure 3.3 may be of help: it shows that an exponential function of a parameter always grows more rapidly than a power-law function of the same parameter. The general nature of a chaotic system can be described from a probabilistic point of view, but not its details. Climate has this nature. Thus, although GCMs are very good at using data to predict trends in climate over various periods of time (months, years, decades), they cannot completely capture the details of climate fluctuations (which we perceive as “weather”), and may fail to identify when Earth’s climate could shift into a drastically different state.
The second limitation has to do with grid size. Most GCMs today are limited to grid sizes of a hundred kilometers in each direction. Smaller grid sizes come at the cost of vastly increased computing time to obtain a result. However, weather is affected by processes on much smaller scales; mountains, shapes of coastlines, and changing surface characteristics may occur on scales of ten kilometers or less. Moist convection cloud and rain formation must be characterized on kilometer and smaller
scales. Thesesubgrid processes play key roles in determining the movement of air, moisture, and heat around Earth, yet they cannot be explicitly computed in GCM models. The strategy is to try to approximately characterize such processes com- putationally so that, on the scale of a grid point, they pro- duce the same effects that the real processes do. Studies of how well GCMs account for the effects of moist convection, for example, suggest that, as yet, this strategy is only partly successful.
The third limitation of general circulation models lies in the coupling of atmospheric and oceanic processes. Because the nature and causes of ocean circulation patterns are only incom- pletely understood, no model exists today that fully characterizes how the atmosphere and the ocean interact. General circulation models may be particularly sensitive to this limitation because of their large demand for computing power and the difficulty of handling simultaneously the short timescales of the atmosphere (days) and the long timescales associated with ocean mixing (centuries). However, much effort over the past decade has been put into improvements in the accuracy of the air–ocean interac- tion in such models, based on better understanding of oceanic circulation, the detailed physics of the exchange of material between ocean and air at the sea surface, and increased com- puting power. The most recent GCMs, to emphasize their more sophisticated incorporation of coupled ocean and atmosphere processes, are sometimes called atmosphere–ocean global cir- culation models (AOGCMs).
AOGCMs represent the most detailed and accurate simula- tions of Earth’s climate that is available with present-day com- puting power. As computers continue to improve in speed and memory, the challenge will be to incorporate physical processes with greater fidelity. It is part of the nature of scientific research to test models of physical systems against their real behavior, based on observational data. With expanded means of collecting data on the current Earth environment, as well as on those of other planets and the Earth in its past, the reasonable expectation is that AOGCMs will continue to improve in their capability to elucidate the behavior of Earth’s climate and make forecasts of future climate changes. By way of example, Figures 22.5 and 22.6 illustrate climate-change predictions of some state-of-the- art GCMs.