Potter T.D., Colman B.R. (co-chief editors). The handbook of weather, climate, and water: dynamics, climate physical meteorology, weather systems, and measurements

Подождите немного. Документ загружается.

mately, by repres enting them in an idealized way in terms of the grid-scale variables

( parameterization). The parameterization may be based on one or a combination of

the following: well-established theory (e.g., for radiative transfer), ﬁeld or laboratory

observations (e.g., for turbulent mixing of heat, moisture and momentum in the

boundary layer), ﬁner scale models (e.g., clouds) and experimentation with the

general circulation models (GCM).

The oceanic component is similar to the atmosphere in resolution, although,

ideally, higher resolution is required to represent mesoscale eddies, which are the

oceanic equivalent of weather, and the parameterizations are generally simpler than

in the atmosphere.

Climate models are numerical models of the atmosphere, ocean, and other

components of the climate system that are used to understand climate and past

climate change and to predict future climate. They range in complexity from

simple globally averaged models that attempt to model changes in the energy

balance of the climate system to complex three-dimensional GCMs. GCMs were

originally developed in parallel with numerical weather prediction models in the

1960s. Indeed, several current GCMs are low-resolution versions of weather predic-

tion models.

Simple climate models are useful in illustrating some aspects of climate change

and can be tuned to mimic some of the global mean changes found in GCMs.

However, only GCMs can provide information on the detailed geographical distribu-

tion of climate change. State-of-the-art GCMs include a full representation of the

atmosphere, ocean, sea ice, and land surface (Fig. 2). The exact formulation of a

climate model will depend on the use to which it is put. For example, accurate

Figure 1 Some factors that affect climate.

156

WHY SHOULD WE BELIEVE PREDICTIONS OF FUTURE CLIMATE?

Figure 2 Basic structure of coupled ocean atmosphere GCM. See ftp site for color image.

157

representation of the radiative effects of greenhouse gases is required if the model is

to be used for studies of anthropogenic climate change. Some new GCMs include

carbon and sulfur models (to understand and predict changes in atmospheric CO

and sulfate aerosol concentrations and their effect on climate) and detailed

atmospheric chemistry (to study ozone depletion and changes in tropospheric

trace gases).

Climate models are used differently from weather prediction models. Weather

prediction models prescribe an initial state of the basic variables from observations,

and the equations are stepped forward in time to give the evolution of the atmo-

sphere. For a few days, it is possible to relate individual features in the forecast to

features that evolve at the corresponding time in the real world. As the forecast

proceeds further, this correspondence disappears as errors due to inexact initial data

and model inadequacies grow because of the nonlinear nature of the equations of

motion. In other words, the chaotic nature of the equations limits the length of time

of deterministic forecasts to about a week or so (depending upon proces ses involved

and the scale of the system being forecast).

A climate model is usually run for long enough that its statistics are independent

of the initial conditions. The experiment is then continued over a number of years,

decades, or even centuries (depending on the availability of computer time and the

application), and the statistics of the simulation over the ﬁnal period are analyzed.

This will include not only the time means, but variability on daily to annual or longer

time scales, storm tracks, extreme events, and so on. If the effect of a particular

change is being investigated (e.g., doubling atmospheric carbon dioxide concentra-

tions, or changing Earth’s orbit al parameters), then the experiment is repeated with

this change in the model, and the statistics of the two experiments are compared.

Both the GCM and observed climate display variations on all time scales due to

Figure 3 Typical climate model grid.

158

WHY SHOULD WE BELIEVE PREDICTIONS OF FUTURE CLIMATE?

internal variability; statistical tests may be needed to demonstrate that the differences

between the control and anomaly simulations are due to the change in the model, and

not to internal variations.

What has been the evolution of climate change in the recent past, and likely

changes in the near future? These questions are addressed using coupled ocean

atmosphere GCMs . Models have been integrated from preindustrial times to the

present using the estimated changes in climate forcing (factors governing climate

due to human and natural causes), assuming that the preindustrial climate was in

quasi-equilibrium, and then extended assuming some future scenario of greenhouse

gas concentrations. As in weather forecasting, errors in initial data and the model

will contaminate these ‘‘hindcasts’’ and forecasts. Hence, the most recent studies use

an ensemble of simulations started from different initial conditions to give a range

of possible future projections taking into account the uncertainty in initial

conditions.

How reliable are GCMs for predi cting future climate change? With climate, this

is argued in several ways. First, they are based on well-established physical laws and

principles based on a wealth of observations. Second, they reproduce many of the

main features of current climate and its interannual variability, including the seasonal

cycle of temperature, the formation and decay of the major monsoons, the seasonal

shifts of the major rain belts, the average daily temperature cycle, and the variations

in outgoing radiation at high elevations in the atmosphere, as measured by satellites.

Similarly many features of the large-scale features in the ocean are reproduced by

current climate models.

Figure 4 Simulations of recent changes in global mean surface temperature (greyband)

compared to observations (black line).

1 INTRODUCTION 159

A model may produce a faithful representation of current climate, yet give unreli-

able predictions of climate change. Hence models have been tested to reproduce both

past climates and recent climatic events. Simulations of temperature and precipita-

tion changes 6000 years ago (due to changes in Earth’s orbital parameters) and the

last glacial maximum (prescribing reconstructed changes in ice sheets, orbital para-

meters, and greenhouse gases) compare tolerably well with reconstructions from

palaeodata. For example, models reproduce the strengthening of the North African

monsoon 6000 years ago and the approximate level of cooling estimated for the last

ice age. The value of these comparisons is limited by the possibility that other factors

not included in the model may have contributed to change in these periods, and

the uncertainty in the reconstructions of temperature and precipitation from the

palaeodata.

Simulations of the last 130 years have been made driven with the observed

increase in greenhouse gases and an estimate of the increase in sulfate particles.

The simulated global mean warming agrees tolerably well with observations. Fore-

casts of the global mean cooling due to the eruption of Mount Pinatubo 1991 were

also very successful. However, it is possible in both these instances that errors in the

estimate of the factors governing climate in these cases (e.g., changes in ozone and

aerosol concentrations) were fortuitously canceled by errors in the model’s sensiti-

vity to those changes. Recent patterns of change including cooling of the upper

atmosphere, a reduction in diurnal temperature range, and a tentative increase in

precipitation in high northern latitudes are in broad qualitative agreement with

available observations.

Various factors limit our conﬁdence in models. The sensitivity of global mean

temperature change with increasing atmospheric carbon dioxide had different

models which varied by a factor of over 2, largely as a result of uncertainty in the

treatment of cloud and related processes. Pronounced differences exist in the regio-

nal changes in model responses to increasing greenhouse gases, although the broad

scale features are quite robust, including enhanced warming in high nor thern lati-

tudes in all seasons but summer, generally greater warming over the land than ocean,

limited warming over the Southern Ocean and norther n North Atlantic, and

increased annual mean runoff in high latitudes. The validation against recent

observed climate change is hampered not only by uncertainty in the factors affecting

climate but also in how much of the recent observed changes are due to natural

variability and naturally forced events (e.g., due to changes in solar intensity).

In summary, climate models are numerical models used to study climate and

climate change. They reproduce many features of observed climate and climate

variability, some of the broad features of past climates. Predictions of future

global mean temperature change vary by a factor of 2 or so for a given scenario

of anthr opogenic greenhouse gas emissions and are reasonably consistent on global

scales, but not on regional scales. The main areas of current research are reducing

model errors by improving the representation of physical process (particularly

clouds), reducing uncertainties in the factors affecting climate, particu larly in the

recent past, and estimating the magnitude of natural climate variability.

160 WHY SHOULD WE BELIEVE PREDICTIONS OF FUTURE CLIMATE?

BIBLIOGRAPHY

Houghton, J. T., Y. Ding, D. J. Griggs, M. Noguer, P. S. van der Linden, X. Dai, K. Maskell, and

C. A. Johnson, (Eds.) (2001). Climate Change 2001: The Scientiﬁc Basis. The Third

Assessment Report of the IPCC: Contribution of Working Group I, Cambridge University

Press, Cambridge, MA.

Trenberth, K. E. (1995). Numerical Modeling of the Atmosphere—Numerical Weather Predic-

tion Haltiner and Martin Climate Modeling—Climate Systems Modeling, Cambridge

University Press, Cambridge, MA.

BIBLIOGRAPHY 161

CHAPTER 14

THE EL NIN

O–SOUTHERN

OSCILLATION (ENSO) SYSTEM

KEVIN TRENBERTH

1 ENSO EVENTS

Every 3 to 7 years or so, a pronounced warming occurs of the surface waters of the

tropical Paciﬁc Ocea n. The warmings take place from the international dateline to

the west coast of South Ame rica and result in changes in the local and regional

ecology and are clearly linked with anomalous global climate patterns. These warm-

ings have come to be known as El Nin˜o events. Historically, El Nin

o referred to the

appearance of unusually warm water off the coast of Peru as an enhancement of the

normal warming about Christmastime (hence Nin

o, Spanish for ‘‘the boy Christ-

child’’) and only more recently has the term come to be regarded as synonymous

with the basinwide phenomenon. The atmospheric component tied to El Nin

ois

termed the Southern Oscillation (SO) whose existence was ﬁrst noted late in the

1800s. Scientists call the interannual variations where the atmosphere and ocean

collaborate together in this way El Nin

o–Southern Oscillation (ENSO).

The ocean and atmospheric conditions in the tropi cal Paciﬁc are seldom close to

average, but instead ﬂuctuate somewhat irregularly between the warm phase of

ENSO, the El Nin

o events, and the cold phase of ENSO consisting of cooling of

the central and eastern tropical Paciﬁc, referred to as La Nin˜a events (La Nin

a is ‘‘the

girl’’ in Spanish). The most intense phase of each event lasts about a year.

This chapter brieﬂy outlines the current understanding of ENSO and the physical

connections between the tropical Paciﬁc and the rest of the world and the issues

involved in exploiting skillful but uncertain predictions.

Handbook of Weather, Climate, and Water: Dynamics, Climate, Physical Meteorology, Weather Systems,

and Measurements, Edited by Thomas D. Potter and Bradley R. Colman.

ISBN 0-471-21490-6 # 2003 John Wiley & Sons, Inc.

163

2 THE TROPICAL PACIFIC OCEAN–ATMOSPHERE SYSTE M

The distinctive pattern of average sea surfa ce temperatures in the Paciﬁc Ocean sets

the stage for ENSO events. Key features are the ‘‘warm pool’’ in the tropical western

Paciﬁc, where the warmest ocean waters in the world reside and extend to depths of

over 150 m, warm waters north of the equator from about 5 to 15



N, much colder

waters in the eastern Paciﬁc, and a cold tongue along the equator that is most

pronounced about October and weakest in March. The warm pool migrates with

the sun back and forth across the equator but the distinctive patterns of sea surface

temperature are brought about mainly by the winds.

The existence of the ENSO phenomenon is dependent on the east–west variations

in sea surface temperatures in the tropical Paciﬁc and the close links with sea level

pressures, and thus surfa ce winds in the tropics, which in turn determine the major

areas of rainfall. The temperature of the surface waters is readily conveyed to the

overlying atmosphere, and, because warm air is less dense, it tends to rise while

cooler air sinks . As air rises into regions where the air is thinner, the air expands,

causing cooling and therefore condensing moisture in the air, which produces rain.

Low sea level pressures are set up over the warmer waters while higher pressures

occur over the cooler regions in the tropics and subtropics, and the moisture-laden

winds tend to blow toward low pressure so that the air converges, resulting in

organized patterns of heavy rainfall. The rain comes from convective cloud systems,

often as thunde rstorms, and perhaps as tropical storms or even hurricanes, which

preferentially occur in the convergence zones. Because the wind is often light or

calm right in these zones, they have previously been referred to as the doldrums.Of

particular note are the Inter-Tropical Convergence Zone (ITCZ) and the South

Paciﬁc Convergence Zone (SPCZ), which are separated by the equatorial dry

zone. These atmospheric climatological features play a key role in ENSO as they

change in character and move when sea surface temperatures change.

There is a strong coincidence between the patterns of sea surface temperatures

and tropical convection throughout the year, although there is interference from

effects of nearby land and monsoonal circulations. The strongest seasonal migra-

tion of rainfall occ urs over the tropical continents, Africa, South America and the

Australian–Southeast Asian–Indonesian maritime region. Over the Paciﬁc and

Atlantic, the ITCZ remains in the Nor thern Hemisphere year round, with conver-

gence of the trade winds favored by the presence of warmer water. In the subtropical

Paciﬁc, the SPCZ also lies over water warmer than about 27



C. The ITCZ is weakest

in January in the Northern Hemisphere when the SPCZ is strongest in the Southern

Hemisphere.

The surface winds drive surface ocean currents, which determine where the

surface waters ﬂow and diverge, and thus where cooler nutrient-rich waters upwell

from below. Because of Earth’s rotation, easterly winds along the equator deﬂect

currents to the right in the Northern Hemis phere and to the left in the Southern

Hemisphere and thus away from the equator, creating upwelling along the equator.

The presence of nutrients and sunlight in the cool surface waters along the equator

and western coasts of the Americas favors development of tiny plant species called

164 THE EL NIN

O–SOUTHERN OSCILLATION (ENSO) SYSTEM

phytoplankton, which are grazed on by microscopic sea animals called zooplankton,

which in turn provide food for ﬁsh.

The winds largely determin e the sea surface temperature distribution along with

differential sea levels and the heat content of the upper ocean. The latter is related to

the conﬁguration of the thermocline, which denotes a region of sharp temperature

gradient within the ocean separating the well-mixed surface layers from the cooler

abyssal ocean waters. Normally the thermocline is deep in the western tropical

Paciﬁc (on the order of 150 m) and sea level is high as waters driven by the easterly

tradewinds pile up. In the eastern Paciﬁc on the equator, the thermocline is very

shallow (on the order of 50 m) and sea level is relatively low. The Paciﬁc sea surface

slopes up by about 60 cm from east to west along the equator.

The tropical Paciﬁc, therefore, is a region where the atmospheric winds are

largely responsible for the tropical sea surface temperature distribution which, in

turn, is very much involved in determining the preci pitation distribution and the

tropical atmospheric circulation. This sets the stage for ENSO to occur.

3 INTERANNUAL VARIATIONS IN CLIMATE

Most of the interannual variability in the tropics and a substantial part of the varia-

bility over the Southern Hemisphere and Northern Hemisphere extratropics is related

and tied together through ENSO. ENSO is a natural phenomenon arising from

coupled interactions between the atmosphere and the ocean in the tropical Paciﬁc

Ocean, and there is good evidence from cores of coral and glacial ice in the Andes

that it has been going on for millennia. The region of the Paciﬁc Ocean most

involved in ENSO is the central equatorial Paciﬁc (Fig. 1), especially the area



Nto5



S, 170



E to 120



W, not the traditional El Nin

o region along the

coast of South America. The evolution of sea surface temperature covering ENSO

events after 1950 is shown in Figure 2.

Inverse variations in pressure anomalies (departures from average) at Darwin

(12.4



S 130.9



E) in northern Australia and Tahiti (17.5



S 149.6



W) in the

South Paciﬁc Ocean characterize the SO. Consequently, the difference in pressure

anomalies, Tahiti minus Darwin, is often used as a Southern Oscillation Index (SOI),

also given in Figure 2. The warm ENSO events clearly identiﬁable since 1950

occurred in 1951, 1953, 1957–1958, 1963, 1965, 1969, 1972–1973, 1976–1977,

1982–1983, 1986–1987, 1990–1995, and 1997–1998. The 1990–1995 event is

sometimes considered as three events as it faltered brieﬂy in late 1992 and early

1994 but reemerged strongly in each case, and the duration is unprecedented in the

past 130 years. Worldwide climate anomalies lasting several seasons have been

identiﬁed with all of these events.

The SO is principally a global-scale seesaw in atmospheric sea level pressure

involving exchanges of air between eastern and western hemispheres (Fig. 3)

centered in tropical and subtropical latitudes with centers of action located over

Indonesia (near Darwin) and the tropical South Paciﬁc Ocean (near Tahiti).

Higher than normal pressures are characteristic of more settled and ﬁne weather,

3 INTERANNUAL VARIATIONS IN CLI MATE 165