Below is the uncorrected machine-read text of this chapter, intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text of each book. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.
3 PHYSICAL BASIS OF CLIMATE AND CLIMATIC CHANGE CLIMATIC SYSTEM The term climate usually brings to mind an average regime of weather. The climatic system, however, consists of those properties and processes that are responsible for the climate and its variations and are illustrated in Figure 3.1. The properties of the climatic system may be broadly classified as thermal properties, which include the temperature of the air, water, ice, and land; kinetic properties, which include the wind and ocean currents, together with the associated vertical motions, and the motion of ice masses; aqueous properties, which include the airâs moisture or humidity, the cloudiness and cloud water content, ground- water, lake levels, and the water content of snow and of land and sea ice; and static properties, which include the pressure and density of the atmosphere and ocean, the composition of the (dry) air, the oceanic salinity, and the geometric boundaries and physical constants of the system. These variables are interconnected by the various physical processes occurring in the system, such as precipitation and evaporation, radiation, and the transfer of heat and momentum by advection, con- vection, and turbulence. Components of the System In general terms the complete climatic system consists of five physical componentsâthe atmosphere, hydrosphere, cryosphere, lithosphere, and biosphere, as follows: 13
âeBueYyD DQeUID Uy Sessec01d j;BUIAaQU} 30 sojdwiexe que (<=) smouie uado at pues âsessed0id jeuseyxe jo sojduiexe aie (-â) SMOLIN JIMNY SUL âWIYSKhsS BNSUYD YWesâ77} ud 3@ pajdno>d ey JO SWUsUOdWOD ey JO LORENSNI WMewWeEYIZS T°⬠JUNDIA "03a âAjiuNeg âadeug ulseg UBZDCO 4O saBueYy "238 âopeqiy âuoneja6a~ âAydesb0109 HLÂ¥V3s | âsaunyeay pue 7 jo sabueysy v A, uea9CQ-39} Buyrdnog [uonssodwoy sueydsour yo sebueuy NV390 Buiyjdnog uesdQ-seydsouny A Vv, a hn (BLD LLLLLL LILLIA Le 7 r A) Ny, i ssag PUIM aBueyoxy 182} ', 39 14 uonerodeag Buydnod adi-siy âuoneyIdioedg saponued snp â2a ââ¬Q â%Q9 âtO âSH 3YU3HdSONLV + |" -39was UOReIPeY 10S | jo sabueus
PHYSICAL BASIS OF CLIMATE AND CLIMATIC CHANGE 15 The atmosphere, which comprises the earthâs gaseous envelope, is the most variable part of the system and has a characteristic response or thermal adjustment time of the order of a month. By this we mean that the atmosphere, by transferring heat vertically and horizontally, will adjust itself to an imposed temperature change in about a monthâs time. This is also approximately the time it would take for the atmo- sphereâs kinetic energy to be dissipated by friction, if there were no processes acting to replenish this energy. The hydrosphere, which comprises the liquid water distributed over the surface of the earth, includes the oceans, lakes, rivers, and the water beneath the earthâs surface, such as groundwater and sub- terranean water. Of these, the worldâs oceans are the most important for climatic variations. The ocean absorbs most of the solar radiation that reaches the earthâs surface, and the oceanic temperature structure repre- sents an enormous reservoir of energy due to the relatively large mass and specific heat of the oceanâs water. The upper layers of the ocean interact with the overlying atmosphere on time scales of months to years, while the deeper ocean waters have thermal adjustment times of the order of centuries. The cryosphere, which comprises the worldâs ice masses and snow deposits, includes the continental ice sheets, mountain glaciers, sea ice, surface snow cover, and lake and river ice. The changes of snow cover on the land are mainly seasonal and are closely tied to the atmospheric circulation. The glaciers and ice sheets (which represent the bulk of the worldâs freshwater storage) respond much more slowly. Because of their great mass, these systems develop a dynamics of their own, and they show significant changes in volume and extent over periods ranging from hundreds to millions of years. Such variations are, of course, closely related to the global hydrologic balance and to varia- tions of sea level (see Appendix A). The lithosphere, which consists of the land masses over the surface of the earth, includes the mountains and ocean basins, together with the surface rock, sediments, and soil. These features change over the longest time scales of all the components of the climatic system, ranging up to the age of the earth itself. The processes of continental drift and sea-floor spreading, which have resulted in mountain building and in changes in the shapes and depths of the oceans, occur over tens and hundreds of millions of years. These events are not generally regarded as representing the same kind of interaction with other components of the system as the variations described above. We note, however, that there may be a significant relationship between the occurrence of major glacial periods and the times when continental land masses occupied
16 UNDERSTANDING CLIMATIC CHANGE positions near the rotational poles of the earth (see Appendix A). The processes of isostatic adjustment and the accumulation of deep- ocean sediments also represent significant changes of the lithosphere, and as such may be viewed as earthâiceâocean interactions. The intro- duction of volcanic debris into the atmosphere and its subsequent dis- persal may also be cited as an example of earthâair interaction. The biosphere includes the plant cover on land and in the ocean and the animals of the air, sea, and land, including man himself. Although their response characteristics differ widely, these biological elements are sensitive to climate and, in turn, may influence climatic changes. It is from the biosphere that we obtain most of the data on paleoclimates (see Appendix A). Natural changes in surface vegetation occur over periods ranging from decades to thousands of years in response to changes in temperature and precipitation and, in turn, alter the surface albedo and roughness, evaporation, and ground hydrology. Changes in animal populations also reflect climatic variations through the avail- ability of suitable food and habitat. The anthropogenic changes due to agriculture and animal husbandry are not known but may well be appreciable in altering at least regional climates. Physical Processes of Climate The climate at any particular time represents in some sense the average of the various elements of weather, along with the state of the other components of the system. The physical processes responsible for climate (as distinct from climatic change) are therefore basically the same as those responsible for weather. These processes are expressed in quantitative fashion by the dynamical equation of motion, the thermo- dynamic energy equation, and the equations of mass and water substance continuity, as applied to the atmosphere and ocean (see Appendix B). A process of primary importance for the circulation of the atmosphere and ocean is the rate at which heat is added to the system, the ultimate source of which is the sunâs radiation. The atmosphere and ocean re- spond to this heating by developing winds and currents, which serve to transport heat from regions where it is received in abundance, such as in the equatorial and tropical areas, to regions where relatively little radiation is received, such as the polar regions of the earth. In this way, the atmosphere and ocean maintain the overall global balance of heat. A great deal of this heat is transported by the disturbances re- sponsible for much of our weather in middle and high latitudes, and similar disturbances may occur in the ocean. These eddies of the general circulation also participate in the transports necessary to maintain the
PHYSICAL BASIS OF CLIMATE AND CLIMATIC CHANGE 17 global balances of momentum, mass, and the total quantity of water substance. While this simple view is a fair summary of our basic understanding of the general circulation, it is not without shortcomings. For example, it does not consider the basically different circulation regime in the low latitudes or the role of convective phenomena, and it does not consider the important variations of the circulation with height. It might also be noted that for other combinations of the planetary size and rotation rate, atmospheric composition, and meridional heating gradient, such as occur on other planets, an altogether different circula- tion regimeâand hence climateâcould result. Although the equations referred to above are fundamental in that they form the basis of our ability to simulate numerically the climate with dynamical models, they are not in themselves particularly reveal- ing as far as the more subtle physical processes of climate are con- cerned, to say nothing of the processes of climatic change. The heating rate is itself highly dependent on the distribution of the temperature and moisture in the atmosphere and owes much to the release of the latent heat of condensation during the formation of clouds and to the subse- quent influence of the clouds on the solar and terrestrial radiation. These processes, together with others that contribute to the overall heat balance of the atmosphere, are shown in Figure 3.2, in which data derived from recent satellite observations have been incorporated (see, for example, Vonder Haar and Suomi, 1971). Here the presence of clouds, water vapor, and CO. is seen to account for over 90 percent of the long-wave radiation leaving the earth-oceanâatmosphere system. This effective blocking of the radiation emitted by the earthâs surface, commonly referred to as the greenhouse effect, permits a somewhat higher surface temperature than would otherwise be the case. It is interesting that this important effect is achieved by gases in the at- mosphere that exist in near trace amounts. We see from Figure 3.2 that the role played by clouds is an important one: the reflection and emission from clouds accounts for about 46 per- cent of the total radiation leaving the atmosphere; and in terms of the shortwave radiation alone, clouds account for two thirds of the planetary albedo. The largest single heat source for the atmosphere is that supplied by the release of the latent heat of condensation, and this is particularly important in the lower latitudes. There is also an ap- preciable supply of sensible heat from the oceans, especially in the middle and higher latitudes. It is therefore clear that water substance, in either vapor or droplet form, plays a dominant role in the atmospheric heat balance. And when we recall that the oceans themselves absorb
âSUOIJBAIASGO jEUO}JUZAUOD PUL SJUsWaINSeOW 921]j9]8S UO peseq âUONEIPSs JBjOS Bujwosuy jo SPUN OOT 0} 2AQejes âasaydsowje ay} JO soUEIBG JeOY PUB UONSIPes jenuue UBeW 241 ZâE AYNNDIS £2 L | iz LS GNV1 âNV390 | UONneIpey aAem-Bu07} I xnjy 20H yo uolssiw3 a1qIsues SOSH YON ae XNj4 2eeH us} / paenen Ce7 sprog Aq ton â10de A a1 / pequosqy sayena Aq uondsosqy WH 18 spnojs Aq EG âsng âs0deA Ip Aq 91 seyena Aq uolssiw3 / pequosay â t0OO 4) âs0deA A 13838 488M hq uorsstusg 10N | pessneos peg 3YSHdSOWLV Y J | y a _ 92 se J v 02 9 Oot aaem-6u0} SABM-LIOUS uoneipey uoneipey Bul063nG wis Buywoou| 39vds
PHYSICAL BASIS OF CLIMATE AND CLIMATIC CHANGE 19 most of the solar radiation reaching the surface, and that the presence of ice and snow also affect the heat balance, the climatic dominance of global water substance becomes overwhelming, even if ice is not taken into account. Definitions It is useful at this point to introduce a number of definitions related to climate and climatic change. In what may be called the âcommonâ definition, climate is the average of the various weather elements, usually taken over a particular 30-year period. A more useful definition is what we shall call the âpracticalâ definition, which introduces the con- cept of a climatic state. This and related definitions are as follows: Climatic state. This is defined as the average (together with the variability and other statistics) of the complete set of atmospheric, hydrospheric, and cryospheric variables over a specified period of time in a specified domain of the earthâatmosphere system. The time interval is understood to be considerably longer than the life span of individual synoptic weather systems (of the order of several days) and longer than the theoretical time limit over which the behavior of the atmosphere can be locally predicted (of the order of several weeks). We may thus speak, for example, of monthly, seasonal, yearly, or decadal climatic states. Climatic variation. This is defined as the difference between climatic states of the same kind, as between two Januaries or between two decades. We may thus speak, for example, of monthly, seasonal, yearly, or decadal climatic variations in a precise way. The phrase âclimatic changeââ is used in a more general fashion but is generally synonymous with this definition. Climatic anomaly. This we define as the deviation of a particular climatic state from the average of a (relatively) large number of climatic states of the same kind. We may thus speak, for example, of the climatic anomaly represented by a particular January or by a par- ticular year. Climatic variability. This we define as the variance among a number of climatic states of the same kind. We may thus speak, for example, of monthly, seasonal, yearly, or decadal climatic variability. Although it may be confusing, this definition of climatic variability includes the variance of the variability of the individual climatic states. The foregoing definitions are useful for two reasons. First, the con- cept of climatic state preserves the essence of what is usually connoted by climate, while circumventing troublesome problems of statistical
20 UNDERSTANDING CLIMATIC CHANGE stability. Second, climatic states represent definite realizations or samples of climate (rather than the climate per se) and are comparable with the climates simulated by numerical general circulation experi- ments. There are many other definitions in existence to distinguish particular statistical characteristics of climate and climatic change (such as climatic fluctuations, oscillations, periods, cycles, trends, and rhythms). The above definitions are generally adequate for our pur- poses, although we shall later consider another definition of climate related to the climatic system. We shall also subsequently introduce the concepts of climatic noise and climatic predictability. Except when otherwise indicated, the use of the word âclimateâ in this report is to be considered an abbreviation for climatic state. It should be noted that we have included the oceans in the definition of a climatic state, as well as information on other aspects of the physical environment. The ensemble of statistics required to completely describe a climatic state is presently available for only a few regions and for limited periods of time. The climatic data-analysis and monitoring programs recommended in Chapter 6 are intended to fill in as much of the gap as possible with available data and to ensure that at least certain critical data are systematically gathered for an extended period of time in the future. CAUSES OF CLIMATIC CHANGE While the above discussion may describe the processes responsible for the maintenance of climate, it is an inadequate description of the processes involved in climatic change. Here we are on less secure ground and must consider a wide range of possible interactions among the elements of the climatic system. It is these interactions that are responsible for the complexity of climatic variation. Climatic Boundary Conditions If we view the gaseous, liquid, and ice envelopes surrounding the earth as the internal climatic system, we may regard the underlying ground and the space surrounding the earth as the external system. The boundary conditions then consist of the configuration of the earthâs crust and the state of the sun itself. Changes in these conditions can obviously alter the state of the climatic system, i.e., they can be causes of climatic variation. Each of the external processes illustrated in Figure 3.1 may be used to develop a climatic theory, on which basis one may attempt to explain
PHYSICAL BASIS OF CLIMATE AND CLIMATIC CHANGE 21 certain features of the observed climatic changes. For example, changes of the distribution of solar radiation have been used since the time of Milankovitch (1930) to explain the major glacialâinterglacial cycles of the order of 10â to 10° years. Aside from the question of variations of the sunâs radiative output, variations of the earthâs orbital parameters produce changes in the intensity and geographical pattern of the seasonal and annual radiation received at the top of the atmosphere and in the length of the radiational seasons in each hemisphere. These effects, which are known with considerable accuracy, have resulted in occasional variations of the seasonal insolation regime several times larger than those now experienced. These orbital elements (eccentricity, obliquity, and precession) vary with periods averaging about 96,000 years, 41,000 years, and 21,000 years, respectively. Because the seasons themselves represent substantial climatic variations, such astronomical theories of climatic change must be given careful consideration. The separate question of the climatic effects of possible changes in the sunâs radiation (1.e., changes of the so-called solar constant) has a much less firm physical basis. Not only are the measured short-period variations of solar output quite small, but the repeated search for climatic periodicities linked with the 11-year and 80-year sunspot cycles has not yielded statistically conclusive results. The question of still longer-period solar variations cannot be adequately examined with present data, although over periods of the order of hundreds of millions of years the sunâs radiation seems likely to have changed. The time range of this and other possible causative factors of climatic change is shown in Figure 3.3. On time scales of tens of millions of years there are changes in the shapes of the ocean basins and the distribution of continents as a result of sea-floor spreading and continental drift (see Figure 3.3). Over geological time, these processes must have resulted in substantial changes of global climate. Just how much of the recorded paleoclimatic variations may eventually be accounted for by such effects, however, is not known, and applying climatic models to the systematic reconstruc- tion of the earthâs climatic history prior to about 10 million years ago is an important component of the research program recommended in this report (see Chapter 6). In such climatic reconstructions, the oceans must be simulated along with the atmosphere, and eventually the ice masses must also be reproduced. Accompanying the migration of the land masses are the processes of mountain building, epeirogeny, iso- static adjustment, and sea-level changes, all of which must also be taken into account. Yet another external cause of climatic variation is the changes in
â@BusYD DjYSUD JeEqOojS yO S10}9De3 SANESNeD ajqissod pue siaydsojq pue âssdydsoyyw âessyds -of12 âquoydsoupAy âesoydsowje ay} ul Ssassadoid pues SJUaAD BYBUII}D BQSHa}IBIeYD Ef JYNDIG sse0A Ul OU) 1.08 t OL 20l wl pol sol ~ol .ot ,ol sol , 4 | oT | | oT 1 { qT I dT | asaydsowry 40 â_ MH Uonewercgny | « auaydsowi\y-Ubs2O | yO UORQUEAOINY A âw lee. 85) âsqUBINYJOgââw <= 10A03 [81080 OVID WAU3LNI l Widv15 <â_â_âââ- a7) puke) $,ueyy ââ__â__» INVNIWOGId INIIOLSIFI YOrVN | 40A87] U8390 JOLEM HLYV3 ââ*ââ_ nepng - ⢠eh - 40 39V ! UOREUSUIPSS | 22 âUOINSOGWUOZ JI1UGIO | <ââ_â_â_âââââ JUSWSNIpy_ 31383SOS); âJ9A9] 9B] â|9A9)-20S _~! ~aâââ 99} 895 â<$<ââ ea ââââ_ 820045 280) [82U9UIIUCD ââ__»> | | Pe $49198|5) i y moug ulejunOW â=â BGuipjing uejUuNnOW âs i | SIZ OWEIEY â1 jG aw Burpeaids J00| 4-895 sue; âYug jeUuEeUnUOD ! â_â ISNG DUCDOA âUONBOdWOD âsseyy DeYdsOUNY ~ NOLLIIGaud YSHLVIM âucamout. tues ,. : Yu S,yueS 1IW9071 40 LIA â AyipiqeueA 18}0S =~ | a i
PHYSICAL BASIS OF CLIMATE AND CLIMATIC CHANGE 23 the composition of the atmosphere resulting from the natural chemical evolution of the nitrogen, oxygen, and carbon dioxide content in re- sponse to geological and biological processes, as well as from the efflu- ents of volcanic eruptions. On shorter time scales, however, it is prob- ably the injection of dust particles into the atmosphere by volcanoes that has produced a more significant climatic effect by modifying the at- mospheric radiation balance (see Figure 3.2). The progressive enrich- ment of the atmospheric CO, content, which has occurred during this century as a result of manâs combustion of fossil fuels (amounting to an increase of order 10 percent since the 1880âs), must also be con- sidered an external cause of climatic variation. These considerations lead to the âphysicalâ definition of climate as the equilibrium statistical state reached by the elements of the at- mosphere, hydrosphere, and cryosphere under a set of given and fixed external boundary conditions. There is, of course, the possibility that a true equilibrium may not be reached in a finite time due to the disparity of the response times of the systemâs components, but this is neverthe- less a useful definition. By progressively reducing the internal climatic system to include only the atmosphere and ocean (in equilibrium with the land and ice distribution), and then to include only the atmosphere itself (in equilibrium with the ocean, ice, and land), a hierarchy of climates may be defined which is useful for the analysis of questions of climatic determinism. Climatic Change Processes and Feedback Mechanisms Important as the above processes may be for the longer-period varia- tions of climate, there are other factors that may also produce climatic change. These involve changes in the large-scale distribution of the effective internal driving mechanisms for the atmosphere and ocean. Variations of the global ice distribution, for example, have a sig- nificant effect on the net heating of the atmosphere (by virtue of the iceâs effective control of the surface heat budget), and thereby may change the meridional heating gradient that drives the atmospheric (and oceanic) circulation. An equally significant change (for the oceans, at least) may be introduced by widespread salinity variations, as caused, for example, by the melting of ice. The salinity of the ocean surface water is in turn closely related to the formation of relatively dense bottom water, which by sinking and spreading fills the bulk of the worldâs ocean basins. Such processes may act as internal controls of the climatic system,
24 UNDERSTANDING CLIMATIC CHANGE with time scales extending from fractions of a year to hundreds and even thousands of years (see Figure 3.3). Some of these processes dis- play a coupling or mutual compensation among two (or more) elements of the internal climatic system. Such interactions or feedback mecha- nisms may act either to amplify the value or anomaly of one of the interacting elements (positive feedback) or to damp it (negative feed- back). Because of the large number of degrees of freedom of the oceanâatmosphere system (for the moment considering the ice distribu- tion to be fixed), there are a large number of possible feedback mechanisms within the ocean, within the atmosphere, and between the ocean and the atmosphere. These same degrees of freedom, however, invite a high risk of error in any qualitative analysis, and in some cases equally plausible arguments of this sort lead to opposite conclusions. Some of the more prominent feedback effects operate among the shorter-period processes of climatic change, especially those concerning the radiation balance over land and the energy balance over the ocean. For example, a perturbation of the ocean-surface temperature may modify the transfer of sensible heat to the overlying atmosphere, and thereby affect the atmospheric circulation and cloudiness. These changes may in turn affect the ocean-surface temperatures through changes in radiation, wind-induced mixing, advection, and convergence (and may subsequently affect the deep-ocean temperatures through geostrophic adjustment to the convergence in the boundary layer). These processes may result either in the enhancement or reduction of the initial anomaly of sea-surface temperature. A number of studies have shown positive feedback of this sort for several yearsâ time in the North Pacific Ocean. The greenhouse effect (in which the absorption of long-wave radiation by water vapor produces a higher surface temperature), is probably the best known example of a semipermanent positive feedback process, al- though other positive feedbacks of climatic importance may be noted. One of these is the snow coverâalbedoâtemperature feedback, in which an increase of snow extent increases the surface albedo and thereby lowers the surface temperature. This in turn (all else being equal) further increases the extent of the snow cover. An example of negative feedback is the coupling between cloudiness and surface temperature noted earlier. In this scheme, an initial increase of surface temperature serves to increase the evaporation, which is followed by an increase of cloudiness. This in turn reduces the solar radiation reaching the surface and thereby lowers the initial temperature anomaly. Here we have ignored the effects of long-wave radiation and of advective pro- cesses in both ocean and atmosphere, but these examples serve to illus- trate the uncertainty that must be attached to such arguments.
PHYSICAL BASIS OF CLIMATE AND CLIMATIC CHANGE 25 While there is much evidence to support the existence of feedback processes, the key phrase in their qualitative use is ââall else being equal.â In a system as complex as climate, this is usually not the case, and an anomaly in one part of the system may be expected to set off a whole series of adjustments, depending on the type, location, and magnitude of the disturbance. Any positive feedback must, in any event, be checked at some level by the intervention of other internal adjustment processes, or the climate would exhibit a runaway behavior. We do not adequately understand these adjustment mechanisms, and their system- atic quantitative exploration by numerical climate models is an important task for the future (see Chapter 6). In that research it will be essential to use coupled models of atmosphere and ocean, and these must be calibrated with great care so as not to distort the feedback mechanisms themselves. Climatic Noise Climatic states have been defined in terms of finite time averages and as such are subject to fluctuations of statistical origin in addition to the changes of a physical nature already discussed. Since these statistical fluctuations arise from the day-to-day fluctuations in weather (the autovariation of the atmosphere identified in Figure 3.3), they are un- predictable over time scales of climatological interest and are therefore appropriately defined as âclimatic noise.â The amplitude of this noise decreases approximately as the square root of the length of the time- averaging interval, but some remains at any finite time scale (Leith, 1973; Chervin et al., 1974). A key problem of climatic variation on any time scale is therefore the determination of the âclimatic predictability,â which we may define as the ratio of the magnitude of the potentially predictable climatic change of physical origin to the magnitude of this unpredictable climatic noise. ROLE OF THE OCEANS IN CLIMATIC CHANGE It has been noted that the oceans play a prominent role in the determina- tion of climate through the processes at the air-sea interface that govern the exchanges of heat, moisture, and momentum. While these condi- tions are actually determined mutually by the atmosphere and the ocean, they are likely dominated by the ocean on at least the longer climatic time scales. It is the high thermal and mechanical oceanic inertia that requires that special consideration be given to the role of the ocean in climatic change.
26 UNDERSTANDING CLIMATIC CHANGE Physical Processes in the Ocean Over half of the solar radiation reaching the earthâs surface is absorbed by the sea. This solar radiation, along with the surface wind stress, is the ultimate energy source for a variety of physical processes in the ocean whose climatic importance is essentially a function of their time scales. The absorption of solar radiation is primarily responsible for the exist- ence of a warm surface mixed layer of order 10? m deep found over most of the worldâs oceans. This warm surface layer represents a large reservoir of heat and acts as a significant thermodynamic constraint on the atmospheric circulation. The exchange of the oceanâs heat with the atmosphere occurs over a wide range of time scales and largely determines the relative importance of other physical processes in the ocean for climatic change. Some of this heat is used for surface evaporation, some is stored in the surface layer, and some is moved downward into deeper water by various dynamical and thermodynamical processes. The fluxes of latent and sensible heat into the atmosphere are commonly parameterized in atmospheric models as functions of the large-scale surface wind speed and the vertical gradients of humidity and temperature in the air over the ocean surface. These fluxes are actually accomplished by small-scale turbulent proc- esses in the surface boundary layer whose behavior is not adequately understood. Physical processes in the ocean such as vertical convective motions (depending on the local vertical stratification of temperature and salinity) and wind-induced stirring also affect the depth and struc- ture of the mixed layer, as shown, for example, by the simulations of daily variations of local mixed layer depth by Denman and Miyake (1973). Other small-scale processes such as salt fingering and internal waves also produce transports that may contribute significantly to the overall vertical mixing in the ocean. Therefore, the dynamics of the oceanâs surface layer must be taken into account in even the simplest of climate models. It is becoming apparent that the most energetic motion scale in the oceans is that of the mesoscale eddy, whose period is of the order of a few months and whose horizontal wavelength is of the order of several hundred kilometers. The kinetic energy of these motions, which is pre- dominantly in the barotropic and first baroclinic vertical mode, may be one or two orders of magnitude greater than that of the time-averaged motions themselves. In a general sense, these slowly evolving eddies are the counterpart of the larger-scale transient cyclones and anticyclones in the atmosphere. An understanding of the physical processes responsible for the origin and behavior of these eddies and their role in the oceanic
PHYSICAL BASIS OF CLIMATE AND CLIMATIC CHANGE 27 general circulation is essential for further insight into the dynamics of the vast open ocean regions. In addition to the surface interactions, vertical mixing processes, and mesoscale motions, the study of the longer-period variations of climate clearly requires consideration of the large-scale dynamics of the com- plete oceanic circulation. This includes the large-scale pattern of wind- driven and thermohaline currents and their associated horizontal and vertical transports of heat, momentum, and salt. Of particular im- portance here is the study of the local dynamics of the intense bound- ary and equatorial currents and the relative roles of inertial and topo- graphic influences. The characteristic variations of these large-scale processes are on time scales of the order of seasons and years in the near-surface waters but may occur in progressively longer time scales in deeper water. The longest oceanic adjustment time associated with the âpermanentâ ocean circulation is of the order 10* years (see Figure 3.3). For climatic variations on these time scales, therefore, the entire water mass of the global ocean must be taken into account. Modeling the Oceanic Circulation The systematic examination of the various mechanisms and feedbacks by which the oceanic thermal structure and circulation are maintained on various time scales is largely a task for the future. In this research, it will be necessary to conduct intensive observational programs in order to gain greater understanding of the various oceanic physical processes themselves and to construct numerical models of the oceanic circulation in which these processes are correctly represented. For climatic studies, it is important that the heat and energy balances of the ocean be modeled correctly over the time and space scales of interest, and this cannot now be said to have been achieved. The classical ocean circulation models, which were initiated in the late 1940âs and further developed in the following decades, do account for the gross features of the ocean circulation, such as the major current systems and the large-scale oceanic thermal structure (see Appendix B). But even these features are physically and geometrically distorted by the con- sideration of only the larger-scale, relatively viscous motions. The commonly used vertical thermal eddy diffusivity in such models is also questionable and may be an order of magnitude too high, as in- dicated, for example, by recent studies on oceanic tritium concentra- tions. This alone will produce a distortion of the processes responsible for deep-water formation in such models. But perhaps more important is the fact that numerical ocean models
28 UNDERSTANDING CLIMATIC CHANGE have not had a sufficiently fine horizontal resolution to portray the mesoscale eddies, either in the open ocean or in the restricted regions of concentrated currents. The accuracy with which the meandering and vortex shedding of boundary currents such as the Gulf Stream or Kuroshio must be modeled, or the resolution required for the transient behavior of the equatorial and Antarctic current systems, depends on the extent to which these features are coupled to the semipermanent or primary current systems themselves and on the time scales under con- sideration. It is unlikely, however, that these features, or the mesoscale eddies, can be successfully modeled with constant eddy diffusion coefficients. To study the role of the oceans in climatic change, it is necessary to construct dynamically and energetically correct oceanic general cir- culation models and to couple them, in appropriate versions, to similarly accurate and compatible atmospheric models. Some experi- ence with simplified coupled models of coarse resolution has already been gained, as discussed in Appendix B. Further tests of coupled models are necessary in which the oceanic mesoscale eddies are re- solved, in order that we may understand their role in the oceanic heat balance and their relationship to the climatically important changes of sea-surface temperature. Since computational limitations will likely preclude the resolution of these eddies throughout the world ocean, their successful parameterization will become an important problem for future research. Of particular importance for climate studies is the construction of an accurate model of the oceanic-surface mixed layer, since all the physical processes in the ocean ultimately exert their influence on the atmosphere through the surface of the sea. Until the dynamics of this oceanic boundary layer are better understood, our ability to model climatic variations on any time scale will remain seriously limited. SIMULATION AND PREDICTABILITY OF CLIMATIC VARIATION Climate Modeling Problem From the above remarks it is clear that the problem of modeling climatic variation is fundamentally one of constructing a hierarchy of coupled atmosphereâocean models, each suited to the physical processes domi- nant on a particular time scale. The attack on this problem is now in its infancy. Whether we consider changes of the external boundary con- ditions or changes of the internally controlled physical processes and feedback mechanisms, we note from Figure 3.3 the wide range of time
PHYSICAL BASIS OF CLIMATE AND CLIMATIC CHANGE 29 intervals over which characteristic climatic events occur and that many of these involve interactions among the atmosphere, oceans, ice, and land. Because of the systemâs nonlinearity, we may expect a broad range of response in both space and time in the individual climatic variables. This is just what the climatic record shows. To study the relative contribution of individual physical processes to the overall âequilibriumâ climatic state, one approach is to test the sensitivity of the statistics generated by a climate model to perturbations in the parameters that influence that particular physical process. In such a modeling program, the effects of changes can first be tested in isolation from other interacting components of the system and then in concert with all known processes in a complete climatic model. In this research, we should not rely exclusively on the general circulation models (GCMâs) but should employ a variety of modeling approaches. We note, however, that not only are the GCMâs (and the coupled Gcmâs in particular) useful in the calibration of the simpler models, but they are essential to the detailed diagnosis of the shorter-period climatic states that are in approximate statistical equilibrium with slowly changing boundary conditions. A fundamental approach to the problem of modeling climate and climatic variation must proceed through the consideration of dynamical models of the coupled components of the climatic system. In minimum practical terms, this means the joint atmosphereâocean system, although for some purposes (such as the behavior of ice sheets and glaciers) the cryosphere must be included as well. Efforts to assemble such models are just getting under way, and their further development is given high priority in the research program recommended in Chapter 6. Predictability and the Question of Transitivity It is possible to regard climatic change as a conventional initial/ boundary-value problem in fluid dynamics, if we define the climatic sys- tem as consisting of the atmosphere, hydrosphere, and cryosphere. In this deterministic view the behavior of the system is governed by the changes of the external boundary conditions (see Figure 3.1). Over relatively short periods, it is even possible to regard the land ice masses as part of the external conditions as well. It is probably not possible, however, to remove the hydrosphere from the internal system and still talk meaningfully about climatic variation, as the surface layers of the ocean interact with the atmosphere on the shortest time scales associated with climate (see Figure 3.3). Decoupling of the ocean, however, is exactly what has so far been done in conventional atmospheric and
30 UNDERSTANDING CLIMATIC CHANGE Oceanic general circulation models, although preliminary efforts to consider the coupled system have been made (see Appendix B). Even with the atmosphere (together with certain surface effects) regarded as the sole component of the climatic system, and with all ex- ternal boundary conditions held fixed, there is, in spite of our physical expectations, no assurance that there will be a climate in the sense that time series generated by the atmospheric changes will settle into a statistically steady state; and no assurance that the climate, if it exists, is unique in the sense that the statistics are independent of the initial state. It is therefore useful to define a random time series (or the system generating such a series) as transitive if its statistics (and hence its climatic states) are stable and independent of the initial conditions and as intransitive if not. As shown by Lorenz (1968), nonlinear systems, which are far simpler than the atmosphere, sometimes display a ten- dency to fluctuate in an irregular manner between two (or more) in- ternal states, while the external boundary conditions remain completely unchanged. This behavior is related to the systemâs transitivity and is illustrated in Figure 3.4. Let us assume that two different states of a climatic system are possible at a time t=0, such as A and B in Figure 3.4, and let us con- sider that A is the climatic state that would normally be âexpectedâ under the given constant boundary condition. In a completely transitive system, the climatic state B would approach the state A with the passage of time and eventually become indistinguishable from it. This would correspond to a unique solution for the climate under fixed boundary conditions. In a completely intransitive system, on the other hand, the climatic state B would remain unchanged, and two possible solutions would exist. There would in this case, moreover, be no way in which we could continue to identify the state A as the ânormalâ or correct solution, as state B would presumably furnish an equally acceptable set of climatic statistics. A third behavior, however, is perhaps the most interesting of all, and is displayed by an almost-intransitive system. In this case, the system in state B may behave for a while as though it were intransitive, and then at time ¢, shift toward an alternate climatic state A, where it might re- main for a further period of time. At time ¢, the system might then return to the original climatic state B, where it could remain or enter into further excursions. The climate exhibited by such a system would thus consist of two (or more) quasi-stable states, together with periods of transition between them. For longer periods of time the system might have stable statistics, but for shorter periods of time it would appear to be intransitive.
âSUORIPUCD AIEPUNOG BWES 3Y} JOPUN 9283S JIqGEYSIq SANBUIIYS UE SI Y 3389S DEUNID GyUL âG B2eIS DQeUINID je;QWIUI Ue 0) yDadsa1 YUM sUaysAs I1QBUNID (â) SANIsUBIQU]-YsOUNe PUB (.. .) SAQISUBIQU! â(ââ) DAIWISUBIy Jo 1O;AeYaq |YyY JO UOREeYSNI PE 3UNDIS auiy y Ny 0 WI1SAS aANIsUeIUY Wa}SAS 3AIZISULI2U)-JSOWYy 31 Wa}SAS BAI}ISUBL) â suonipuoy Asepunog juesU0D =
32 UNDERSTANDING CLIMATIC CHANGE Because the atmosphere is constantly subject to disturbances, such as those arising from flow over rough terrain or from the occurrence of baroclinic instability, one might think that it could not be an almost- intransitive system and fail to show greater excursions of annual and decadal climatic states than it does. This depends, of course, on the level of variability associated with individual climatic states and hence on the time interval we select to define the climatic state itself (and on how close neighboring quasi-stable states might be). What may appear to be a climatic transition on one time scale may become the natural noise level of a climatic state defined over a longer interval. This is consistent with many of the climatic records presented in Appendix A. Even so, it might still be possible for the coupled oceanâatmosphere system or for the coupled oceanâatmosphereâice system to be almost intransitive. One cannot help but be struck by the appearance of those proxy records that display repeated transitions between two states (see Figures A.13 and A.14 in particular). This evidence suggests that the glacialâinterglacial oscillations that have characterized the past million years of the earthâs climatic history may be the climatic transitions of an almost-intransitive system. Another possible example of this phenomenon is the irregular and relatively sudden reversal of the earthâs magnetic poles. The search for further evidence of this sort in both the paleoclimatic record and in the climatic history generated by numerical models is an important task for future research. As though the specter of almost-intransitivity were not enough, on the longer-time scales of climatic variation it is equally important to recognize another, potentially serious complication. If it turns out that climatic evolution is influenced to a significant degree by environmental impacts originating outside the atmosphereâoceanâcryosphere system, then the predictability of climate will be additionally constrained by the predictability of the environment in a larger sense. This, in turn, could turn out to be the greatest stumbling block of all, as illustrated, for example, by the difficulty of predicting the timing and intensity of vol- canic eruptions (which inject radiation-attenuating layers of dust into the upper atmosphere) and by the difficulty of predicting the behavior of the sun itself, which is the ultimate source of the energy driving the climatic system. As noted earlier, the predictability of climatic variation is con- strained by an inherent limitation in the detailed predicability of the atmosphere and ocean. Climatic noise as previously defined thus arises from the unavoidable uncertainty in the determination of the initial state and from the nonlinear nature of the relevant dynamics, as shown, for example, by Lorenz (1969). Fluctuations in the weather for periods
PHYSICAL BASIS OF CLIMATE AND CLIMATIC CHANGE 33 beyond a few weeks may therefore be treated in large part as though they were generated by an unpredictable random process. The observed time series of many meteorological variables may be reasonably well modeled by a first-order Markov process with a time (r)-lagged cor- relation given by R(7) =exp(â»|7|), with a constant decay rate v of the order of 0.3 day-'. The corresponding power spectrum as a function of frequency w is given by P(w)=A/(»*+?), where A is a constant. As wâ0 for such a spectrum, we have P(w)â»A/v*, which is a constant, and the very-low-frequency end of the spectrum therefore appears âwhite.â There is thus some contribution to climatic variations on all time scales, no matter how long, arising from the fluctuations of the weather. While these considerations do not directly address the physical basis of climatic change, they are nevertheless basic to our view of the pre- dictability of climatic change. What parts of climatic variations on various time scales are potentially predictable, and what parts are just climatic noise? In the power spectrum is there potentially predictable variability above the âwhite,â low-frequency end of the daily weather fluctuations, or is it possible that some of the long-term compensation processes, such as those shown in Figure 3.3, might depress the spec- trum below its white extension to »=0? The 250-year record of monthly mean temperatures in central England compiled by Manley (1959) shows small lagged correlations significantly above that of weather out to about 6 months, small but perhaps significant lagged correlations at 2 and 4 years, and a generally white spectrum with some evidence of extra variability for periods of a few decades and longer (C. E. Leith, NCAR, Boulder, Colorado, un- published results). The 6-month lagged correlation may well be a re- flection of the role of North Atlantic sea-surface temperature anomalies on the English climate and illustrates the somewhat longer periods of the autovariation of the coupled oceanâatmosphere system over those of the atmosphere alone, as indicated in Figure 3.3. Additional evidence of even longer-period variability is found in the historical and paleo- climatic records (Kutzbach and Bryson, 1974; see Figure A.5). Further studies of this kind should be made with statistical tests not only of the pessimistic null hypothesis that nothing is predictable but also of hypotheses that are framed more optimistically. Long-Range or Climatic Forecasting As our understanding of the physical basis of climatic variation grows, we hope to be able to discern the predictable climatic change signal
34 UNDERSTANDING CLIMATIC CHANGE from the unpredictable climatic noise and to describe with some con- fidence the character of both past and likely future climates. In view of the questions posed by limited predictability, however, this discern- ment may be limited to those circumstances in which there is a relatively large change in the processes or boundary conditions of the climatic system. The related problem of forecasting specific seasonal and annual climatic variations rests upon the same physical basis and may prove more difficult to solve. To reach these goals will require the coordinated use of all our research tools, whether they be observational, numerical, or theoretical. The capstone of these efforts will be the emergence of an increasingly well-defined and tested theory of climatic variation. Whether the predictability of climatic change turns out to be lower than many would like to believe or to be limited to a finite range as in the problem of weather forecasting, the quest for understanding must be made. Our recommendations for the research that we believe to be a necessary part of this effort are presented in detail in Chapter 6.