Saltzman B. (editor) Anomalous Atmospheric Flows and Blocking

Подождите немного. Документ загружается.

This Page Intentionally Left Blank

MECHANISTIC EXPERIMENTS TO

SOUTHERN HEMISPHERE STATIONARY

ROSSBY WAVES

DETERMINE THE ORIGIN

SHORT-SCALE

EUGENIA

KALNAY

AND

KINGTSE

C. Mo*

Laboratory for Atmospheres

National Aeronautics and Space Administration

Goddard Space Flight Center

Greenbelt, Maryland

20771

INTRODUCTION

Studies by van Loon and Jenne

(1 972),

van Loon

al.

(1 973),

Trenberth

(1979),

and others indicate that stationary waves in the Southern Hemi-

sphere are of planetary scale, and generally weaker than in the Northern

Hemisphere. On the other hand, Kalnay and Halem

(1981)

reported the

presence of unexpectedly large-amplitude, short-wavelength stationary

waves over and in the lee of South America during January

1979,

the first

month of the First GARP Global Experiment (FGGE) Special Observing

Period

(SOP-1).

Even more remarkable was the fact that the waves disap-

peared during February, the second month of

SOP-

Kalnay and Paegle

(1 983)

and Kalnay

al.

986)

have discussed several

possible mechanisms that could have given rise to the waves during January

and not during February: orographic forcing by the Andes, sea surface tem-

perature (SST) anomalies, tropical heating, and resonant enhancement

forcing due to convection over the Amazonia region and over the South

Pacific Convergence Zone (SPCZ) region.

Briefly, they concluded that the January waves were not forced by the

Andes because of their phase: instead of

lee trough their phase corresponds

to a lee ridge or poleward flow over the Andes. In an equivalent barotropic

fluid this indicates transfer

kinetic energy from stationary waves to the

mean flow (Kalnay-Rivas and Merkine,

198

I),

that the waves seem to

exist

despite

the Andes. Kalnay and Paegle

983)

also pointed out that SST

anomalies of the same wavelength

the atmospheric waves were observed

in the South Atlantic during SOP-

However, their phase and time change

indicated that the SST anomalies were driven by the atmospheric stationary

Department

Meteorology, University

Maryland, College Park, Maryland

20742.

415

1986

Academic

Press,

Inc.

All

rights

reproduction

in any

form

reserved.

ADVANCES

GEOPHYSICS.

VOLUME

416

EUGENIA KALNAY AND KINGTSE

waves, rather than being their cause. Finally, they suggested that the waves

were related to the stronger convective heating over the Amazon and central

Pacific during January. This stronger source of stationary forcing was proba-

bly the cause of weaker tropical easterlies and stronger tropical westerlies in

the upper atmospheric flow during January. They concluded that the pres-

ence of stronger tropical westerlies during January allowed freer propagation

of energy from the stationary low-latitude forcing into the extratropics.

Simple numerical experiments with a hemispheric two-level quasi-geo-

strophic model

(J.

Paegle, personal communication) and with a shallow

water barotropic model

(F.

Semazzi, personal communication), are in gen-

eral agreement with these arguments. They indicate that placing a source of

tropical heating in the Amazon region leads to the development within a few

days ofa train of short subtropical waves not unlike the ones observed during

January. The results also show a strong sensitivity of the stationary waves in

the South Atlantic to the precise position and extension of a heat source

representing the SPCZ. However, the experiments with simple models are

not conclusive because the waves have a phase relationship to the Andes

corresponding to a

lee trough

rather than the observed

lee ridge.

In the present study, we perform experiments with a comprehensive

three-dimensional general circulation model

(GCM).

In the first two experi-

ments, we attempt to reproduce

closely as possible the observed January

and February

1979

atmospheric circulation by performing 15-day forecasts

from real initial conditions. Because of the well known limited predictability

of the atmosphere, we cannot expect

priori

to simulate well the observed

stationary circulation. However, since the

GCM

succeeds, at least partially,

reproducing the stationary waves in January and their absence in Febru-

ary, we have the possibility of modifying the model and performing other

mechanistic

experiments designed to determine the precise role of orog-

raphy, tropical heating, etc.

Section

we present the

two

control experimental forecasts: 15-day

integrations with the Goddard Laboratory for Atmospheric Sciences

(GLAS) fourth-order

EM,

starting from

January and

February

1979.

We compare them with the

GLAS

analyses for the same periods and discuss

the extent to which the observed January waves are reproduced in the fore-

cast. The rest of the paper is devoted to several mechanistic experiments in

which we artificially modify the Andean orography (Section

3),

tropical

heating (Section

4),

and regional heating (Section

and observe the effect of

these changes on the stationary waves. In Section

we show that a slight

deceleration of the initial westerly flow results in a remarkable amplification

regular train of zonal wavenumber 5 waves, similar to

those

observed in

the Southern Hemisphere during the FGGE summer (Salby,

1982;

Randel

and Stanford,

1983).

Finally, Section

contains a summary of the experi-

ments and conclusions.

THE ORIGIN OF STATIONARY ROSSBY

WAVES

417

ANALYSIS

AND

CONTROL EXPERIMENTS

In this section we compare the time averages of 15-day control forecasts

started from

January 1979 and from 4 February 1979 with analyses aver-

aged over the same period. For simplicity, in the rest of the paper we will refer

to these 15-day averages as “January” and “February,” respectively. Since

operational forecasts retain useful skill for only about

week (e.g., Bengts-

son, 1984) and we are trying to reproduce the persistent anomalous waves

observed in January 1979, it does not seem warranted to make forecasts

longer than

weeks.

The forecasts are performed with the GLAS fourth-order GCM as docu-

mented in Kalnay

al.

(1983), which has a resolution of 4” latitude,

5”

longitude, and 9 vertical levels. The rather coarse resolution is partially

compensated by the use of fourth-order horizontal differences,

that the

model has a short-range forecast skill comparable to that of models of finer

resolution (e.g., Atlas, 1984; Halem

al.,

1982; Baker

al.,

1p84). The

model has a complete parameterization of subgrid-scale physical processes,

and when used in a general circulation simulation mode, reproduces rather

well the summer and winter atmospheric circulation (Kalnay

al.,

1983).

For verification, we use the GLAS Analysis, developed by Baker

983),

which has the same horizontal resolution and has been used for many satel-

lite data impact studies (e.g., Halem

al.,

1982; Baker

al.,

1984). There

very good agreement between the time-averaged fields of rneridional veloc-

ity, which we present here, and the corresponding fields of the European

Centre for Medium Range Weather Forecasts (ECMWF) analysis.

Figure la and b presents the observed

5-day averaged) January and

February mean meridional velocity at

200

mbar.

comparison with the

same maps averaged over 30 days during January and February (Kalnay and

Paegle, 1983) shows the same characteristics almost everywhere except that

the stationary waves appear somewhat stronger in the 15-day average than in

the 30-day average. The focus of our attention is the very large amplitude (up

to 30 m sec-’ mean meridional velocity) and short zonal wavenumber

(approximately 7) stationary waves centered over and lee of South America

during January at about 30”s. These subtropical waves are virtually absent

in the February average, which also shows weaker waves in the SPCZ region.

As indicated by Kalnay and Paegle (1983), independent satellite observa-

tions agree well with this picture; showing the presence ofstrong convergence

zones in the South Pacific and South Atlantic in January, and their virtual

absence in February.’ The corresponding zonal velocity averages at 200

It should

noted that the January

SPCZ

is displaced eastward from its normal position, a

phenomenon similar to that observed during

“El

Niiio”

years for longer periods.

418

419

420

EUGENIA KALNAY

AND

IUNGTSE

mbar (not shown) indicate stronger upper level tropical westerlies and

weaker easterlies in January, in agreement with Kalnay and Paegle (1983). It

is also important

notice that during February, the center of maximum

cyclonic vorticity

(&/ax

min)

located at about

20"s

30°W,

which

very

close to

its

climatological position (Taljaard and van Loon, 1969; Newel1

al.,

1972).

contrast, during January, the cyclone center is shifted to 30"s

and

40"

and

much more intense.

Figure 2a and b presents Hovmoller diagrams of the meridional velocity at

200 mbar at

30"s

for the same January and February periods. The difference

in the stationary wave behavior in January and February is truly striking,

indicating that during January, stationary waves over South America were

not only present, but they dominated the subtropical circulation.

Next we present the 15-day average control forecasts for January and

February (Fig. 3a and b). With some notable exceptions (e.g., northwest

Africa in February), the model succeeds in predicting the average direction

of the meridional flow, and its changes between January and February. In

our region of interest, South America, we observe that during January there

-tm

-14

-100

-80

-a

lea

FIG.

Hovmoller

diagram

ofthe meridional velocity

200

mbar

and

30"s

for

(a)

January

and (b) February. Contour interval is

sec-'.

THE ORIGIN

STATIONARY ROSSBY WAVES

are short stationary waves elongated mainly in the north-south direction

which are similar but weaker than the January waves of the analysis.

In the February forecast, the stationary waves over South America are

weaker than in January, and, in agreement with the analysis, are tilted in the

northwest

southeast direction. The forecast has also succeeded in placing

the cyclone center in the correct position, 20"s and 40"W.

Figure 4a presents the Hovmoller diagram for 200 mbar for the January

forecast. It

apparent that during the first 10 days, the model is able to

predict and even amplify the stationary waves over South America. On

10-

January, the waves reach an amplitude of 40

sec-' in the meri-

dional wind, about 10 m sec-I stronger than in the analysis. After that,

however, there is a sudden breakdown of the stationary waves, with a partial

regeneration at about

January. In order to test the possibility that the

sudden breakdown of the forecasted stationary waves may

due to baro-

tropic instability, we computed the barotropic stability parameter

(@u/

dyz).

Hovmoller diagram of this parameter computed at the same latitude

30"s

confirms this hypothesis (Fig. 4b). The stationary waves are asso-

ciated with a strong minimum in the zonal velocity, which results in negative

-a

422

423