James I.N. Introduction to Circulating Atmospheres

Подождите немного. Документ загружается.

8.4 Intraseasonal oscillation 279

flow responds to such forcing. But this can be very misleading. For example,

suppose the wind field associated with an equatorial Kelvin wave is split into

its rotational and divergent parts, which may be represented by a stream-

function and velocity potential respectively. Both fields will be global. Yet

the Kelvin wave is essentially a trapped equatorial disturbance which dies

away rapidly at latitudes greater than about 20°. In fact, the rotational and

divergent wind vectors for a pure Kelvin wave cancel out in the midlatitudes,

but reinforce one another in the tropics. The Helmholtz partitioning of the

wind field, while mathematically perfectly valid, is physically misleading in

this case. The same is true, at least to a certain extent, of the flow associated

with the intraseasonal oscillation.

In the longitude-height plane, the oscillation of the zonal wind is strongest

in the upper troposphere, between 15 and 20kPa, and it has the familiar

baroclinic structure of tropical disturbances, with low level easterly anomalies

beneath upper level westerly anomalies, and vice versa. The relationship

between the wind, surface pressure fields and the distribution of regions

of enhanced convection is shown schematically in Fig. 8.13. The cycle

is dominated by enhanced convection above an anomalously large low

convergence. This appears over Indonesia and moves eastwards across the

Pacific before dying away near 180

°W.

The anomalous convection is part of

a zonal convection cell, or 'Walker circulation', very similar to the Walker

circulations of the time mean flow described in Section 8.1.

This structure, together with the eastward phase speed, has suggested

that the intraseasonal oscillation may be interpreted in terms of a propag-

ating Kelvin wave. The difficulty with this interpretation is that the observed

phase speed is slower than the phase speed of a simple dry Kelvin wave,

as derived in Section 8.2. Another question concerns the mechanism which

might excite the Kelvin wave. The evidence is that tropical moist convection

may provide the necessary energy through some feedback between the large

scale flow and the convective scale. 'Conditional instability of the second

kind' (CISK) has been suggested as a model of how such a feedback might

operate. The mean tropical atmosphere is conditionally unstable, but it is

not saturated. Thus, for convection to begin in a given region, there must

be convergence of a water vapour flux into that region. The release of latent

heat within the convecting elements drives midlevel ascent and so serves to

intensify the low level moisture convergence into the convecting region. The

pattern of convergence and divergence driven in this way can also serve to

generate large scale relative vorticity through the Rossby source term S (see

Eqs.

8.7 and 8.8). Such a mechanism has also been suggested to account for

the formation of hurricanes as well as other larger scale tropical weather

Low frequency variability of the circulation

Fig. 8.12. EOFs 1 (a) and 2 (b) of the divergence field for eight DJF seasons of

ECMWF data.

systems such as easterly waves. The problem with such mechanisms is

that they depend upon some postulated relationship between cumulus scale

motion and the large scale synoptic fields, that is, upon a parametrization of

the cumulus convection. Given this, it is no surprise that different theoretical

models of the intraseasonal oscillation give rather different predictions of

the phase speed of the wave. The CISK mechanism can certainly act to

slow a Kelvin wave down, by changing the coupling between the upper level

and low level flow. Many current GCMs are capable of reproducing some

8.5 The

Southern Oscillation

281

Fig. 8.12

(cont.).

PCI while dashed is PC2. (Courtesy A. Matthews.)

kind of intraseasonal oscillation, but they overestimate the phase speed. The

results are certainly sensitive to the particular convective parametrization

used.

Since the intraseasonal oscillation was first identified, improved observa-

tions of the tropical atmosphere (especially the wider availability of satellite

data) and careful analysis of earlier records have revealed its signature in

many meteorological fields. For example, the 'active' and 'break' periods

in the Indian monsoon, during which the monsoonal rainfall fluctuates, are

revealed as a modulation of the monsoonal circulation by the intraseasonal

oscillation. Cloudiness over South America and Africa also appears to vary

according to the phase of the intraseasonal oscillation. Some global re-

sponses are also seen. For example, the length of the day, which directly

reflects variations of the atmospheric angular momentum, also shows a 30 -

60 day oscillation.

8.5 The Southern Oscillation

Fluctuations with timescales of one year or longer contribute to the inter-

annual variability, whereby one winter or summer season differs significantly

from another. EOF or correlation analysis of seasonal or annual mean fields

highlights whether any significant, spatially coherent patterns are associated

with the interannual variability. One such pattern has been identified, and

is recognized as the best-defined such structure. It accounts for a significant

282

Low frequency variability of the circulation

EAST LONGITUDE WEST LONGITUDE

20° 60° 100° 140° 180° 140° 100° 60° 20°

1 I i I I I I I I I I I I i

Tropopause

Sea Level Pressure

AFRICA INDONESIA S. AMERICA

Fig. 8.13. Schematic summary of the relationship between anomalies of the surface

pressure field, the azimuthal circulation, the tropopause height and the tropical

cumulus convection through a typical cycle of the intraseasonal oscillation. (From

Madden and Julian, 1972.)

8.5 The Southern Oscillation 283

fraction of the total interannual variance over much of the globe. It has two

aspects: one is an oscillation of the atmospheric circulation and the other is

an oscillation of the tropical ocean circulation. The former has been known

for many years in the form of a negative correlation between the surface

pressure at Darwin, Australia, and Easter Island, in the mid-Pacific. This

was called the 'Southern Oscillation'. The oceanic part of the fluctuation

was revealed by the sea surface temperatures off the coast of Peru. Norm-

ally, upwelling brings cold, nutrient-rich water to the surface here. But

every few years, the upwelling ceases and the sea surface temperatures rise

several degrees, with devastating consequences for the local fisheries. The

phenomenon is called 'El Nino' by the Peruvian fishermen, for it tends to

occur around Christmas.

Global monitoring reveals that both the Southern Oscillation and El

Nino are large scale events which involve changes over most of the tropical

Pacific. Teleconnections link the Southern Oscillation to circulation changes

in the midlatitudes. Furthermore, the phenomena are intimately linked to

each other, forming a coupled nonlinear oscillation of the tropical ocean

and global atmospheric circulations. The coupled oscillation is called

Nino-Southern Oscillation' or, regrettably, 'ENSO'. Like other nonlinear

oscillations, ENSO does not have a precisely defined period, but is a broad

band feature. El Nino events differ in their intensity: some events are minor,

while others can be very large indeed. The largest ENSO event in the last

40 years was recorded during 1982-3, and has been studied extensively. The

interval between ENSO events can be as short as two years, or as long as

seven years; the average period is about 40 months.

Figure 8.14 illustrates the two aspects of the 1982-3 ENSO. In Fig. 8.14(a)

are shown anomalies of the

kPa wind for DJF 1982-3, compared to the

six years 1983-9. A strong westerly anomaly extends over much of the

central tropical Pacific, with weaker easterly anomalies over Indonesia. This

pattern would be expected to result from the movement of the centre of

tropical convection from Indonesia to the central or eastern Pacific. At the

same time, strong cyclonic anomalies are seen at higher latitudes. These form

wavetrains linking the tropics and midlatitudes. The northern hemisphere

train is reminiscent of the PNA pattern shown in Fig. 8.1(b). Figure 8.14(b)

shows anomalies of the sea surface temperatures during the El Nino in

January 1983. The main features are cooling over the Indonesian region and

warming towards the South American coast.

Attempts to model the ENSO oscillation are mainly based around a

coupled oscillation of the combined atmospheric and oceanic circulation.

Figure 8.15 is a schematic illustration of the sort of interaction which is

284

Low frequency variability of the circulation

5.0

(a)

:;T:tr:::

- ^ / ,.,f

V 1

/ / /

.-f

' / '

-*•:/

/ /

!::iiSi!l

^Vft-A^-VV

»—i—

>^//r

. . . . j.

-V\\\V-

>--

— :..

\ « x >

N *

\ 1 /

S * '

- /:»

30N

ION

10S

30S

150W 120W

90W

Fig. 8.14. The structure of ENSO during 1982-3, the largest ENSO event in the

last 40 years, (a) Anomaly of the

kPa winds (ECMWF data); the sample vector

represents

ms"

(b) Anomalies of sea surface temperature over the Pacific during

1982-3.

Contour interval

K, values in excess of +1 K shaded.

probably involved. The usual state of affairs is shown in Fig. 8.15(a).

The warmest sea surface temperatures are in the vicinity of Indonesia. These

drive intense convection in this region, which releases latent heat and drives a

Walker circulation, as discussed in Section

7.1.

The low level winds associated

with the Walker cell exert an eastward stress on the sea surface, feeding warm

water into the western Pacific, and maintaining the upwelling along the

$.5 The Southern Oscillation

285

(b)

Fig. 8.15. A schematic illustration of the coupling between the ocean and atmo-

spheric circulation during El Nino events. The diagrams show longitude-height

sections of the atmosphere and the ocean along the equator over the Pacific Ocean:

(a) Normal circulation, with convection over Indonesia; (b) anomalous or El Nino

circulation, with the convection moving to the mid or western Pacific.

coast of South America. But now suppose that the sea surface temperature

maximum moves towards the mid-Pacific. The convective maximum and

the Walker cell will move further east. The low level easterly winds are

replaced by westerlies over much of the tropical Pacific, moving the warm

water further east and suppressing the South American upwelling. It can be

seen that the changes will feed back on themselves and help to establish and

maintain the anomalous El Nino state. Idealized models which represent

such a coupling between the low level tropical winds and the flow in the

upper layers of the ocean suggest that the process is an instability; an initial

condition resembling that shown in Fig. 8.15(a) begins to oscillate between

the El Nino state and the normal state (sometimes called the 'La Nina'

286 Low frequency variability of the circulation

state).

The amplitude of the oscillation saturates at some maximum value.

In a more complex situation, the influence of such irregularities as baroclinic

instability in the midlatitudes will mean that the El Nino oscillations will

be somewhat irregular. It has been suggested that their timescale may be

determined essentially by the time taken for equatorial Kelvin waves in the

ocean to propagate across the Pacific basin, and that changes in the ocean

circulation could be used to forecast the evolution of such events.

The global implications can be understood in terms of Rossby wavetrains

propagating away from the anomalous convective heating. In this way, the

El Nino creates circulation anomalies over midlatitude North and South

America, Australia and possibly even further away. Indeed, the El Nino is a

major, if not the largest, component of interannual variability over much of

the globe. Improved understanding of the El Nino holds out the possibility of

imparting some sort of skill to weather predictions on the seasonal timescale

or even longer. As a result of this hope, and because of the impact of the

highly anomalous weather regimes associated with the massive 1982-3 event,

studies of El Nino have become very widespread and fashionable.

8.6 Blocking of the midlatitude flow

The examples of low frequency behaviour that we have discussed so far

have their seat in the tropical latitudes. The midlatitudes have much larger

low frequency variance. But the midlatitude low frequency variability has a

less well-defined range of frequency, and generally a much less clear spatial

structure than tropical variability. In this section, we will describe one very

characteristic low frequency fluctuation of the midlatitude circulation which

is still imperfectly understood and which attracts considerable attention.

This feature is called 'blocking'; it is often thought of as an alternation

between a relatively zonal midlatitude flow with superimposed fast moving

disturbances, and a more meridional flow with stationary or near stationary

disturbances.

This alternation of flow regime is traditionally described by a 'zonal index'.

This might, for example, be based on the mean midlatitude zonal geostrophic

wind. A typical index is derived from the zonal mean geopotential heights

°N and

°N.

Zonal flow is characterized by large values of the zonal

index, while blocked flow has a smaller zonal index. Such a zonal index is

observed to oscillate erratically between low and high values, with typical

periods in the range of several weeks. The low index flow is often marked

by intense, nearly steady anticyclones. These have preferred locations; the

eastern Atlantic and western European sector is one and the other is in

8.6

Blocking

the midlatitude

flow 287

the eastern Pacific. Both lie downstream of the major storm tracks, and in

a region where the time mean wave pattern tends to be characterized by

ridging (see Chapters 6 and 7). A blocking anticyclone may last for several

weeks; in some major European droughts, it seems that a succession of such

anticyclones keep forming in much the same location.

Blocking is a much less prominent feature of the southern hemisphere

midlatitudes. Some regions where intense anticyclones are more prevalent

have been identified, but they are less less long lived than their northern

hemisphere counterparts. One such region lies to the south of New Zealand,

where the tropospheric jet tends to split (see Fig. 7.11); another lies near

South America.

Once a major block is established, it prevents the usual eastward motion of

cyclonic weather systems. Instead, they are steered around the block, mainly

around the poleward side. In so doing, they are distorted and tilted. This is

illustrated for a characteristic case in Fig. 8.16(a), which shows the high pass

filtered geopotential height field superimposed on the low pass filtered field

at the same time during a blocking event over the UK. The discussion in

Section 7.4 immediately suggests how a characteristic pattern of E-vectors

will be associated with such distorted high frequency eddies. The E-vectors

are shown in Fig. 8.16(b). They tend to converge into the block, suggesting

that the interaction between the high frequency transients and the block will

tend to reduce westerly flow in the region of the block. That is, the transients

will tend to reinforce the block. This effect is one important way in which

blocks tend to be maintained. On the other hand, one can imagine that

a particularly intense cyclone may be strong enough to disrupt the block,

explaining the apparently random collapse of the system. But these ideas

do not explain the geographical location of the blocks especially well. If

they were formed entirely by interactions with high frequency transients, one

might expect to see them more frequently in the downstream part of a storm

track, but one might also expect to see them form fairly frequently at other

locations also. One might also expect to see a particular block drift away

from its original location. It seems that interactions with features of the

Earth's surface must also be involved in the formation of blocks. This might

be directly, or indirectly, via dynamical interactions involving the steady

waves forced by mountains or land-sea contrasts.

A particularly elegant theory of how the forcing of steady waves could

lead to both blocked and zonal flow regimes was proposed by Charney and

Devore (1981). It is a straightforward extension of the theory introduced in

Section 6.2 of barotropic flow over a mountain. For simplicity, consider flow

along a /?-channel; the topography of the lower surface is simply wavelike

288

Low frequency variability of the circulation

(a)

6CTW

20°W

60*E

20°E

(b)

60°W

40°

20°W

SCALE IN

Fig. 8.16. The interaction of high frequency transient eddies with a blocking an-

ticyclone over north-west Europe: (a) high pass eddies and low pass geopotential

height; (b) E-vectors. (From Hoskins et. al, 1983.)

in the x- and y-directions. This will force a steady wave whose phase and

amplitude were given by Eqs. (6.9) and (6.10), and were illustrated in Fig. 6.8.

The steady wave response is resonant, with a maximum amplitude when the

zonal wind is:

U = (8.13)