Climatic determinants of Benguela SST variability

Climatic determinants of Benguela SST variability

~ ) Pergamon ContinentalShelfResearch, gol. 15, No. 11/12, pp. 1339-1354, 1995 Copyright © 1995 Elsevier Science Ltd Printed in Great Britain. All r...

987KB Sizes 1 Downloads 71 Views

~ )

Pergamon

ContinentalShelfResearch, gol. 15, No. 11/12, pp. 1339-1354, 1995 Copyright © 1995 Elsevier Science Ltd Printed in Great Britain. All rights reserved 0278-4343/95 $9.50 + 0.0(1

0278--4343(94)00090-5

Climatic determinants of Benguela SST variability MARK R. JURY* and SHAUN COURTNEY* (Received 4 March 1994; in revised form 12 August 1994; accepted 16 August 1994) Abstract--Climatic determinants of summer sea surface temperature (SST) variations in the Benguela upwelling zone of the SE Atlantic are studied through statistical associations with field variables of SST and surface winds in the ocean basins surrounding SW Africa. Shelf SST indices are formulated from COADS monthly ship data for the southern and central Benguela for February and October respectively, months of active upwelling. Linear correlation maps are produced at lags - 4 and 0 months and provide some unexpected results, February, southern area SST trends are downward over the period 1950-1988. Correlations between the shelf SST index and SST in the rest of the Benguela zone are positive and suggest that interannual signals are widespread in the longshore direction, but out of phase with the central ocean basins, Correlations between SST and meridional winds at 0 lag are - 0 . 4 in the south and - 0 . 8 in the central Benguela. This confirms the expected relationship between southerly wind and upwelling. The lower correlation in the south indicates that SST variability is often advectively forced whereas, in the central Benguela, SST variability is driven by local upwelling. To assess the impact of an El Nifio on winds in the SE Atlantic, a general circulation model (GCM) simulation is performed. A fixed +2°C SST anomaly is imposed in the central Indian Ocean for a period of three years. Resultant surface layer winds in February are compared with climatology and inferences are made with regard to upwelling. Southerly wind stress is nearly doubled in the central Benguela. On the other hand, the Cape Town area obtains increased onshore flow from mid-latitude westerlies, and a reduction in upwelling is inferred. The GCM simulation is consistent with the observed pattern of an enhanced anticyclonic gyre in the SE Atlantic and cooler SST in the central Benguela. The GCM and statistical results demonstrate regional response patterns with respect to coastal upwelling, and offer deterministic inputs to environmental long-range forecasts.

INTRODUCTION Research on interannual variability of the physical environment in the southeast (SE) Atlantic Ocean has not attained the detailed level of understanding as that of the Pacific, where the El Nifio originates and where more extensive observational coverage is afforded. Pacific basin atmospheric and oceanic variability has been described through EOF and composite lag analysis (Davis, 1976), while Brink et al. (1978) have investigated the dynamical forcing of intra-seasonal oscillations off the coast of Peru using statistical techniques.

*Oceanography Department, University Cape Town, Rondebosch 7700, South Africa. 1339

1340

M.R. Jury and S. Courtney

In the SE Atlantic, sea surface temperatures (SST) are cooled by the Benguela Current which is sourced with coastal upwelled water. The upwelling enriches the physical environment along the west coast of southern Africa, contributing to a productive marine ecosystem and fisheries (Shannon et al., 1988). Active upwelling and SST < 16°C characterize the "Benguela zone" from 34 to 17°S, seawards to the 1000-m isobath (Shelton et al., 1985). The offshore extent and intensity of upwelling over the shelf is modulated by longshore winds driven by the South Atlantic Anticyclone (SAA). The upwelling-favourable winds are modulated at intra-seasonal frequencies of 10-25 days by eastward moving Rossby waves in the circumpolar westerlies, and associated high pressure cells and continental shelf waves (Jury, 1988; Jury and Brundrit, 1992). SST are coldest in the spring (August-October) in the central Benguela (24-30°S) and during summer (November-February) in the south (30-34°S). Studies of inter-annual variability of Benguela SST have been conducted with the aim of establishing scientific links between fisheries and major changes in the local and regional physical environment. In a comparative study, McLain et al. (1985) suggests that SST anomalies in the Benguela are more persistent in time than space, in contrast with other upwelling zones. Warm events in the northern Benguela (1963, 1984) have been analysed by Shannon et al. (1986) and related to incursions of equatorial water. Remote forcing of the tropical Atlantic, in a manner similar to the global E1 Nifio, is suggested as a primary causal mechanism. Walker (1987) established spatial and temporal coherence patterns of Benguela SST variability through principal components (PC) analysis. PC1 (58%) reveals a 5-6 y cycle of major warm and cool events. PC2 (12%) describes opposing variability in the northern and southern Benguela and displays a 3.5-3.9 y cycle consistent with the E! Nifio. North-south differences are attributable to meridional shifts of the anticyclonic surface wind gyre in the SE Atlantic and to the influence of the warm Agulhas Current (Shannon et al., 1990). Taunton-Clark and Shannon (1988) investigated interannual variability in the SE Atlantic in light of the 1982/83 E1Nifio and its impact on fisheries. They note shifts in trade winds leading to a reduction in upwelling in the south in 1983 and a warm event in the north in 1984. SST records reveal gradual upward trends in the first half of the 20th century, and decadal cycles are found to be of relatively low amplitude. The propagation of events, generally towards the central Benguela from both the north and south, is indicated through comparison of shelf SST indices and SST anomaly maps (Shannon and Agenbag, 1990). Taunton-Clark and Shannon (1988) find significant intercorrelation between SST in an area 30-35°S, 10-15°E and coastal SST in the central and southern Benguela. They caution that SST trends and cycles depend on record length and data quality. Taunton-Clark and Kamstra (1988) analyse environmental variability via pressure, wind, SST and the southern oscillation index (SOI). The time of onset and duration of seasonal wind driven upwelling is found to be highly variable, owing to standing Rossby waves in the westerly winds which re-position the South Atlantic anticyclone. Seasonal climatic scenarios are offered to explain decadal trends in SST. In "summer mode", the SAA is located further south, trade winds are of longer duration and upwelling is more intense. In "winter mode", i.e. 1955-1965 and 1975-1985, the SAA is further north and upwelling near Cape Town is suppressed. In this paper the intention is to extend the work on inter-annual variability of Benguela SST in a predictive manner through correlation with regional climatic determinants. In the following section the data and methods are reviewed. The results are divided into sections

Climaticdeterminantsof BenguelaSST variability

1341

on February southern index SST correlations, October central index SST correlations and a general circulation model (GCM) simulation of SE Atlantic wind response to an El Nifio warm event transmitted from the Indian Ocean. DATA AND METHODS Statistical correlation analysis

Climatic determinants of variability in the Benguela upwelling zone are investigated through the formulation of SST time series in two areas. The SST and surface wind data are from the Comprehensive Ocean-Atmosphere Data Set (COADS) which is described in Slutz et al. (1985) and Woodruff et al. (1987). The ship, buoy and other marine data are accumulated into monthly bins every 2° latitude/longitude over the world's oceans and carefully screened for data quality. Duplicate reports and outliers are removed prior to monthly averaging. The observations are spatially weighted according to proximity to the center of each grid square (Cressman, 1959). In regions of poor data coverage, the interpolation routine provides a climatological weight based on historical means which ensures a representative data field. The index areas considered in this study include a 2° grid square off Namaqualand, South Africa: 30-32°S, 16--18°E; and a second 2° box which encompasses the upwelling zone off Luderitz, Namibia: 24-26°S, 12-14°E. These areas are shown in Fig. 1 in the context of shelf width and summer SST. The SST time series indices are given in Fig. 2. The Namibian SST box is selected as it is the persistently coldest part of the Benguela upwelling zone. The South Africa box is selected as the southernmost grid cell within the upwelling zone. It is also an important fisheries region. As a check on data quality, the COADS SST time series were compared with SST time series reported in the literature and locally available. Favourable comparisons are noted with SST time series reported by Walker (1987) and Taunton-Clark and Shannon (1988) for similar areas. Comparisons with the local South African Data Centre for Oceanography (SADCO) data set are not favourable since United Kingdom Meteorological Office (UKMO) SST data were incorporated in 1989. The incompatibility is attributable to a spatial bias in the data distribution. Much of the SST and wind inputs come from the shipping lane located off the shelf edge. SADCO SST in the selected grid squares closely correspond with COADS SST in the adjacent squares to the south and west; SADCO mean values are 14-16°C in the central Benguela and 18-20°C in the southern Benguela. The COADS SST time series from 1950-1988 are in the range 12-15°C for October in the central Benguela, and 14-17°C for February in the southern Benguela and reflect the presence of coastal upwelling. The time series is sufficiently long (N = 39) that infrequent environmental perturbations do not unduly bias the statistical inferences. In determining possible associations with Benguela SST, the index and field variables are first normalized with respect to each month's historical mean and standard deviation, so that seasonal amplitudes are reduced and comparable. The linear regression correlation of normalized index and field values is computed at each grid point using the Pearson's product moment technique (Pathack, 1993). The statistical significance of the correlation results is assessed via the degrees of freedom (Sciremammano, 1979). The time series extend from 1950-1988, however a linear trend in February SST (0.06°C y i) and inter-annual persistence in October SST reduce the degrees of freedom. Autocorrelation

1342

M.R. Jury and S. Courtney BENGUELA REGION MEAN SUMMER SST 20S 22S 24S 26S 28S 30S 32S 34S 36S 10E

12E

14E

16E

1BE 20E

Fig. 1. Map of the Benguela region showing mean summer (December-February) SST over the SE Atlantic (thin lines in °C intervals), selected grid squares used in correlation study and the location of Luderitz and Cape Town. Continental shelf edge is identified by 1000-rn isobath (dashed).

at various lags reveals that standard errors accumulate similarly for de-trended and original records. The degrees of freedom are estimated as 30 for February, and 35 for October. SST vs field absolute correlations >0.36 are therefore significant at the 95% confidence limit. As C O A D S SST are represented in most grid cells in the period analysed, significance levels are relatively uniform across the domain. Point-to-field correlation maps were produced at various lags from - 6 to +4 months. H e r e the lags - 4 and 0 months are selected for interpretation, as it corresponds to the offset of minimum SST in the two grid cells. Four months is also a time period which could make predictions operationally useful if subsequent relationships between SST and fisheries productivity can be found. The lags - 4 and 0 refer to the pattern four months prior to (June w.r.t. October, October w.r.t. February), and coincident with the month analysed. Correlation values are contoured within the domain 20°N-40°S, 70°W-110°E: the South Atlantic and Indian Oceans surrounding the continent of Africa. Contours are analysed at 0.2 intervals, the zero line is unplotted and positive values are contoured in bold. Limited objective smoothing is incorporated in the contouring routine such that peak correlation values in individual grid cells appear slightly reduced. Surface wind correlations are c o m p u t e d for the V and U components independently. Positive correlations refer to SST

Climatic determinants of Benguela SST variability SST anom for February 3 0 - 3 2 ° S

1343

16-18°E

2.50

200 ~I 150 O0 0.50 I

ooo

~//><,~e..~~

050-t.00150 _200 i

50

5'5

6;

6;

7;

Years

75

SST anom for October 2 4 - 2 6 ° 5 5.00

-

2.50

-

2.00

-

8;

K5

12-14°E

150 -

OI

0.50 0.00 /~~/ -0.50

-

-1.00

-

1.50

-

-2°°5o

5;

~o

6;

7'0

7;

5o

B;

90

Years

Fig. 2. Time series of SST standardized departures from 1950-1988 for the selected grid squares shown in Fig. 1. Top is for southern Benguela in the month of February, bottom is for central Benguela in October.

increases when the field variable is above normal in the locations indicated. For SST this implies (sympathetic) warming, and for surface wind a southerly (meridional) or westerly (zonal) anomaly. Interpretations are provided when significant correlations are noted in a number of adjacent grid cells. Correlations below an absolute value of 0.36 are considered "noise".

GCM El Nitio simulation and SE Atlantic wind response A s a secondary goal, the effects of an El Nifio on winds over the Benguela upwelling z o n e is investigated using the Australian Centre for Scientific and Industrial Research

1344

M.R. Juryand S. Courtney

Operations (CSIRO) four level general circulation model (GCM). Some of the model's attributes are briefly outlined. Vertical layers are centered at 900,650,350 and 100 hPa in the atmosphere. The horizontal resolution is represented by spherical harmonics truncated at wavenumber R-21, with a resolution of 3.2 ° latitude × 5.6 ° longitude. The model includes realistic continental geometry, smoothed orography and climatologically defined SST. The model has been used successfully to understand how SST anomalies perturb the atmosphere to an anomalous state, thereby producing changes in circulation and rainfall which are consistent with observations (Gordon and Hunt, 1991). The model response to SST is through thermodynamic principles, whereby surface heat fluxes increase moisture convergence and subsequent cumulus convection. The process can result in positive feedback through latent heating, particularly in tropical latitudes. In the African region it is recognized, through statistical correlation of the SOI with regional climatic elements, that the E1 Nifio has two main expressions (Jury et al., 1994a). The first being the enhancement of upper westerlies over the equatorial Atlantic Ocean and the second being an increase in SST in the central equatorial Indian Ocean; which are associated with similar increases in the eastern Pacific basin. Departures from mean climate can be simulated in the CSIRO GCM only through the imposition of SST anomalies. Here, SSTs are increased 2°C above climatology in the area 5°N-15°S, 55°80°E in the central Indian Ocean. This is consistent with an El Nifio--induced drought over southern Africa (Pathack, 1993). Following a 1000 day run, resultant surface layer winds are compared with mean winds for the month of February. Differences between anomaly and climatology winds demonstrate the response of the SAA to the global E1 Nifio. The tropical "Atlantic Nifio", an infrequent phenomena which often follows a global El Nifio (Shannon et al., 1986), is not considered. RESULTS

Southern Benguela February SST in the southern Benguela exhibit a declining trend over the period considered (Fig. 2). The downward trend is consistent with the Bakun model of more intense upwelling in the global warming scenario of enhanced ocean basin gyres (Bakun, 1990). The trend could be the result of: - - a lengthening of the summer upwelling season, --the more rapid uplift of sub-surface water, - - a wider offshore extent of the upwelling zone, or - - a bias in the observations: more observations over the inner shelf in recent years during the active phase of upwelling. Highest SST occur in 1957, and lowest in 1964, 1974 and 1983, with a decadal period approximately in-phase with the solar flux (Mason, 1992). SST in the 1980s are all below the mean. The trend and cycle is consistent with that found in the literature, for example area 5 of Taunton-Clark and Shannon (1988). In the following section, correlation maps are interpreted according to index area, variable field and lag. The February southern SST index vs the SST field at lags - 4 and 0 months is shown in Fig. 3. Because of the linear trend, the results may reflect inter-decadal climatic change in addition to the intended inter-annual fluctuations. At lag - 4 months, which refers to the preceding October, the contour plots reveal:

Climatic determinants of Benguela SST variability

=~

~ o/~/

. . . . . . . . . .

-70

-50

0

~ I Feb SST vs Feb SST ~ ' ( . . ~ \

o-__.

-30

_ ~

-10

^

10

,,

K~/ .,~ f~'fL~ c~(~r~'.~

~ . ~ < ~ ' ~ Z Z D

,; ~ -

30

1345

50

70

__

90

110

Fig. 3. Correlation map of southern area February SST vs SST field at lag - 4 months (October) and simultaneous (February). Contours are analysed at 0.2 intervals, the zero line is unplotted, and positive values are in bold. The domain includes the South Atlantic and Indian Oceans. Reference grid square is identified.

--positive correlations of > +0.6 with SST in the Benguela shelf zone from 10-35°S, --negative correlations in the Angola bight ( < - 0 . 8 ) , --negative correlations with the central ocean basins of the South Atlantic (< -0.6) and central Indian Ocean ( < - 0 . 4 ) . At lag 0, the simultaneous correlation of February southern SST index vs the SST field indicates: --positive correlation (>+0.4) along the entire west coast of southern Africa from 10°N-35°S, extending offshore to 10°E, --positive correlation with SST along the SE coast of Africa from 15-25°S, and along the NE and SE coast of Brazil, --continued negative correlation with the central ocean basins, although values are weaker, and - - n o clear spatial evolution. The correlation maps for the February southern SST index vs the meridional wind field at lags - 0 . 4 and 0 months are illustrated in Fig. 4. At lag - 4 months in the preceding October: --negative correlations ( < - 0 . 6 ) along the west coast between 20-34°S, suggest that southerly winds equatorward of the index area anticipate upwelling,

1346

_

M . R . J u r y a n d S. C o u r t n e y

' - -,.,

*

.~~ "x . / \ ]

~•

b ',ST vs Oct V wind~

Fe

~6~-

~

o

~

.,

-2C

-4¢

,,J

-~o

i

"_

0

-~o

o

'

, o

4o

% ZD •

'

0

0

-;o

10

30

50

(~ [,Feb SST vs Feb V wind ~ t - x t

70

90

1t O

-- o>~ \ r/'--.O _'2~((~

~o X .

(~)

0 0

',

-1 o21 -31 ..4~

-70

i

-50

Fig. 4.

-30

-10

10

30

50

70

90

110

C o r r e l a t i o n m a p as in Fig. 3, but for SST vs surface m e r i d i o n a l wind.

--positive correlations (> +0.8) are noted over the Angola bight and in the SE Atlantic, -----correlation patterns of the Benguela zone are reflected on the east coast of Africa at similar latitudes, --correlations are positive from 10°W-10°E from 3540°S in the South Atlantic midlatitudes, and --patterns over the ocean basins are incoherent and evenly divided between negative and positive values. At lag 0, the simultaneous correlation of February southern SST index vs the meridional wind field reveals: --weak negative correlations locally (-0.2 to -0.4), which imply that longshore winds explain at most 16% of the SST variance in the southern area, --positive correlations in the Angola bight and over the southern tip of Africa, hence southerly winds there are associated with warm events in the south. Correlation maps for the southern SST index vs the zonal winds at lags - 4 and 0 months are illustrated in Fig. 5. In the preceding October: --positive correlations, which refer to enhanced westerlies, in the Angola bight anticipate warming in the south, --negative correlations are found off the southern tip of Africa, hence easterly wind anomalies are associated with warming, --positive correlations are noted in the mid-latitudes of the SE Atlantic and over the ocean basins to the north of the equator.

1347

Climatic determinants of Benguela SST variability

Oct U windt~d'(~0 ,/~:~0 ~ \ " 7

~

,,k~ 1

d ° : -" -70 21

~, )

-50

,

-30

-10

10

30

"

50

, -(kFebSSTvs]Feb U wind'.~c-~-~vo

O~

°

a 70

g

° \ /

90

11Q

x.v9 1 ~ ' ~ 1 ~

o g

-70

.50

-30

Fig. 5.

.1 a

10

30

.# 2<. 2 SO

70

90

1t 0

Correlation map as in Fig. 3, but for SST vs surface zonal wind.

At lag 0, the simultaneous correlation of February southern SST index vs the zonal wind field shows: --positive local correlations, therefore westerly anomalies over the shelf contribute to warming in the southern Benguela, and --generally positive correlations over the ocean basins and mid-latitudes, hence enhanced westerlies are associated with higher SST in the south.

Central Benguela Reference to Fig. 2 shows that the central Benguela October SST time series exhibits a prominent positive anomaly in 1961 and a negative anomaly in 1979. Unlike the southern region February SST, no gradual downward trend is evident. During the late 1950s the central region is generally cooler than normal in the early summer, while the southern upwelling zone is above normal in the late summer. The central region thus contains signals which are unique and worthy of investigation. Correlation maps for the central Benguela index with respect to October SST variability are shown in Figs 6, 7 and 8. The correlation of October SST vs the SST field (Fig. 6) at - 4 months lag illustrates: --weak sympathetic correlations over the Benguela shelf zone in the latitudes 10-30°S --generally positive correlations in the central ocean basins, --positive correlations along the SE coast of South America.

1348

M . R . Jury and S. Courtney

/~L[ Oct SST vs Jun ssT--~ ~ L

-70

-50

2.

-30

~

<~

-10

:~ ' # {

10

30

50

OctSSTvsOctSST ~-----~

*

-70

-50

Fig. 6.

-30

~ ) ~"-~,--,_cj ~ ' (

-10

10

30

,

O

50

70

90

110

90

110

o . ~

00°

c;~

70

Correlation map as in Fig. 3, but for central area October SST vs SST field at lag - 4 months (June) and simultaneous (October).

Simultaneous correlation of October central SST index vs SST at 0 lag reveals: --positive local correlations extending southward along the west coast from 25-35°S and in the Angola bight, - - a small area of negative correlations from 15-20°S, --weakening of correlations over the ocean basins. The correlation of October SST vs the meridional winds (Fig. 7) at - 4 months lag displays: - - n o local correlation, --negative correlations in the Angola bight and positive correlations around the southern tip of Africa, --positive correlations off the bulge of west Africa, --incoherent patterns over the ocean basins. The simultaneous correlation of the October central area SST index vs the meridional wind field indicates: --the expected negative correlation ( < - 0 . 8 ) over the central Benguela region, --negative correlations extend into the southern Benguela and around the southern tip of the continent, --weak and incoherent patterns over the ocean basins. The correlation of October SST in the central Benguela vs zonal winds (Fig. 8) at - 4 months lag illustrates:

Climatic determinants of Benguela SST variability

SST vs Jun V Wind"../

1349

~

k/V')

7 ~° u

U

~0

0

~

k~-~(

~0

~v,~

0

Oo -70

r~

-50

o

-30

.

-10

q~o

10

~O-~t

30

50

70

90

110

SST vs Oct V wind

~v/o ~ 00

-1

I~.~(

M

),g . 0 -~

o~



..'1

-2C -3(~

-4(

,

-70

~.~

-So

Fig. 7.

~

-30

_

~

. -10

~

_

~ 10

~ o 30

SO

~b~:::)" 10

" So

110

Correlation map as in Fig. 6, but for SST vs surface meridional wind.

- - w e a k positive correlations in the central South Atlantic, --positive correlations in the Angola bight and to the east of South Africa, --relatively strong negative correlations in the South Atlantic mid-latitudes from 50 °20°W. The simultaneous correlation of October SST index vs the zonal wind field highlights: - - w e a k local correlations, - - w i d e s p r e a d weak negative correlations over the ocean basins and in the mid-latitudes. Inferences which may be drawn from the correlation results are discussed in the final section.

El Nitro GCM simulation and response of SE Atlantic winds The G C M simulation provides insight to environmental conditions in the SE Atlantic during an E1 Nifio transmitted from the Indian Ocean. February climatological mean surface layer winds are compared with El Nifio-influenced winds, 1000 days after a +2°C SST anomaly has been imposed in the central equatorial Indian Ocean (Fig. 9). It is useful to confirm whether the climatology is well represented. The centre of the SAA is located near 29°S, 4°W. An axis of light wind flow, separating the mid-latitude westerlies from the southeasterly trade winds extends along a zonal axis toward Cape Town from 29°S, 0 ° to 37°S, 17°E. These features are consistent with climatological mean patterns of the surface high pressure cell in the SE Atlantic in February. The mid-latitude westerlies at 40°S and SE trade winds at 20°S over the open ocean are 5-6 m s -j . These

1350

M . R . Jury and S. Courtney

?

,

"

'

-!

0

o'n

/(,[ Oct SST vs Jun U wind~-2/ ti ~

O

~

:

/, o

~

Yo

o

I

\-~/"~-,

t..J

"'

-o

b

c,, o ' ~

v

.

,

-2C -3©

/U.

~ ~.~<~_oTA~

<'"

, ",-'

c~t::::p__ o o . < ~

c,",.,

-

-4C -70 2 t~-~

-50 °

-30

-10

,

10

~.~

~-~-~---

o

30

SO

SST vs Oct U wind .

70

90

110

Ol~ ~'~_~ •

(~

~

-11

.~

j'~. Akk ~

, g t ~ _ f._~ ~ o .~ <:::>,'T"-.<~...~ t o

/L 5D0o o g " ' :o. . _ ' " -70

-50

Fig. 8.

-30

~°,'~

-10

10

30

50

70

gO

110

Correlation map as in Fig. 6, but for SST vs surface zonal wind. Dashed box in central Indian Ocean identifies SST anomaly imposed in GCM study.

values are in agreement with climatological means reported by ships (i.e. C O A D S ) and island stations. Winds along the Benguela coast increase from 4 m s - I at 34°S, 17°E off Cape Town to 6.5 m s - t at 26°S off Luderitz in the February climatology. In the G C M E1 Nifio simulation, the SE Atlantic anticyclonic surface wind gyre "spins up": mid-latitude westerlies, sub-tropical easterlies and longshore coastal winds are all increased. Significantly, the position of the gyre does not shift equatorwards. Rather, the ridge axis extending toward Cape Town retreats westwards. The southern Benguela upwelling zone experiences a m o r e westerly c o m p o n e n t of flow. The mean wind direction off Cape Town changes from 170 ° to 195 °, and wind speed decreases slightly to 3.8 m s -1 . The westerlies near 40°S increase by 40% to > 8 m s -1 and the trade winds near 20°S increase about 15% to 6-7 m s 1. In the context of coastal upwelling a fundamental difference is the 30% increase in SSE wind in the central and northern Benguela. This aspect is illustrated in Fig. 10. Coastal longshore pseudo wind stress (V260) increases >25 m 2 s -2 off Namibia 21-26°S. In the southern Benguela longshore wind stress decreases 10 m 2 s 2. The area of diminished stress is confined to the latitudes 32-35°S and extends around the coast to the southeast of Cape Town. The shift of m a x i m u m upwelling favourable winds to the central Benguela suggests that SST there would be below normal during a global E1 Nifio. This is confirmed in both 1983 and 1987, when SST off Cape Town (Luderitz) were > I ° C above

1351

Climatic determinants of Benguela SST variability

i

.i

p

j'

~

.,i

,,

'" L'"

.

i'i"

I

t ..,.,..4

ANOM

(a) (b) Fig. 9. Surface layer wind in the SE Atlantic for February as simulated by CSIRO G C M l000 day run with climatology (a) and El Nifio anomaly (b). Wind blows towards the dot, magnitude scale is provided.

0S

L-L

!

BENGUELA REGION V1602 ANOM-CLIM

E

15.~ • 20S

""

2'3-':~h ~

"->I /

405

0E Fig. 10.

f

I

20E

40E

Longshore wind stress difference between climatology and anomaly field computed from GCM results.

i

I

1352

M.R. Jury and S. Courtney

(below) normal (Parker et al., 1992). The shift of winds in the GCM simulation is accomplished without the SAA moving appreciably. Rather the wind gyre obtains increased anticyclonic vorticity primarily through the mid-latitude westerlies. It is well known that circulations associated with the El Nifio lead to a reduction of moisture over the interior of southern Africa and a coincident increase of mean surface temperature by 1-2°C (Lindesay, 1988). Jury et al. (1994b) have demonstrated that February rainfall is halved in this E1 Nifio simulation and that surface sensible heat fluxes over the western interior increase by 10 W m -2. The result is a stronger continental thermal low and sharper pressure contrast with the offshore marine high. During an El Nifio, the pressure gradient and equatorward winds over the west coast are enhanced preferentially to the north of 32°S. DISCUSSION AND CONCLUSIONS Climatic determinants of summer SST variations over the Benguela upwelling zone have been statistically analysed. Of note is the downward trend of February SST in the 3032°S, 16-18°E area. This may be attributed to the gradual enhancement of the anticyclonic gyre in the SE Atlantic and an associated increase in the offshore extent and intensity of coastal upwelling. The lengthening of the summer upwelling season is another potential factor. A feature of SST variability in the southern Benguela is the decadal cycle, consistent with previous research. Unexpected inter-relationships between the SST indices and SST in the rest of the Benguela zone are found. Positive correlations extend over 1000 km in the longshore direction and indicate that inter-annual signals are relatively widespread. SST fluctuations in the South African area are out of phase with SST in the central ocean basins, while the Namibian area is more in phase. Simultaneous correlations between SST and meridional winds at 0 lag are - 0 . 4 in the south and -0.8 in the central Benguela, and confirm the expected relationship between southerly wind and upwelling. The lower correlation in the south indicates that SST variability may often be advectively forced by oceanic intrusions, for example from the warm Agulhas Current to the southeast. In the central Benguela, changes in SST are consistent with local wind-driven upwelling. Zonal wind components are small in the Benguela, but locally positive correlations suggest that warming of SST is associated with onshore wind flow. An interesting result is that westerly wind anomalies over the ocean basins and mid-latitudes of the southern hemisphere are associated with higher SST in the southern area, but lower SST in the central area. The westerlies spin up the anticyclonic gyre, promoting upwelling off Namibia. However the northward intrusion of westerlies limits upwelling favourable winds in the south. Hence the opposing response, as also indicated in the GCM wind anomaly results (Fig. 10). The correlation maps demonstrate regional response patterns with respect to coastal upwelling, and offer deterministic inputs to environmental long-range forecasts. For example, departures of SST in the preceding October in the southern Benguela are usually of the same sign as SST departures in the following February. Environmental indicators of February warm events in the southern Benguela include easterly surface wind anomalies to the south of Africa and northerly wind anomalies along the Namibian coast in the preceding October. Southerly wind anomalies to the south of Africa in the preceding June are associated with above normal SST in the central Benguela. The Angola bight is

Climatic determinants of Benguela SST variability

1353

f e a t u r e d p r o m i n e n t l y in a n u m b e r o f c o r r e l a t i o n p a t t e r n s , h o w e v e r d a t a in t h e r e g i o n a r e sparse and the associations may be considered unreliable. A s e c o n d a r y goal of this s t u d y was an a s s e s s m e n t o f E1 Nifio influences. A G C M s i m u l a t i o n o f surface c i r c u l a t i o n s was a n a l y s e d to infer a shift in u p w e l l i n g activity in the B e n g u e l a . T h e s i m u l a t i o n , initialized with i n c r e a s e d S S T in the c e n t r a l I n d i a n O c e a n , r e s u l t e d in e n h a n c e d s o u t h e r l y surface l a y e r winds in the c e n t r a l B e n g u e l a a n d i n c r e a s e d m i d - l a t i t u d e w e s t e r l i e s n e a r C a p e T o w n . A n e q u a t o r w a r d shift in u p w e l l i n g is i n f e r r e d . T h e G C M s i m u l a t i o n is c o n s i s t e n t with o b s e r v a t i o n s d u r i n g r e c e n t E1Nifios: the anticyclonic g y r e in the S E A t l a n t i c spins up a n d s u m m e r S S T off N a m i b i a are 1-2°C b e l o w the m e a n ( P a r k e r et al., 1992). F u r t h e r w o r k c o u l d i n c l u d e an analysis of c o m p o s i t e cool a n d w a r m p h a s e s of the B e n g u e l a a n d an e v a l u a t i o n o f r e g i o n a l a t m o s p h e r i c forcing a n d r e s p o n s e p a t t e r n s . A t m o s p h e r i c G C M studies which i m p o s e S S T a n o m a l i e s in the Pacific a n d A t l a n t i c c o u l d be c o n d u c t e d to e x p l a i n p h a s e r e l a t i o n s h i p s . It is o f i n t e r e s t to u n d e r s t a n d h o w SST varies in the S E A t l a n t i c , given its m o d u l a t i n g effect o n e a r l y s u m m e r rainfall o v e r s o u t h e r n A f r i c a ( N i c h o l s o n a n d E n t k e h a b i , 1987; P a t h a c k , 1993). A n a m b i t i o u s e n d e a v o u r w o u l d be t h e utilisation o f o c e a n i c G C M s to g e n e r a t e S S T a n o m a l i e s which a r e c o n s i s t e n t with c o m p o s i t e results. I n this w a y t h e b u d g e t o f SST v a r i a b i l i t y c o u l d be d i s t i n g u i s h e d in t e r m s o f local a n d r e m o t e w i n d forcing, a n d o c e a n i c h e a t i n g a n d a d v e c t i o n .

Acknowledgements--The analysis of COADS data was co-sponsored by the Water Research Commission and the Foundation for Research Development. B. Pathack of the Mauritius Meteorological Services provided useful guidance in the early phases of the project. GCM model runs were performed at the University of Pretoria by C. J. de W. Rautenbach.

REFERENCES Bakun A. (1990) Global climate change and intensification of coastal upwelling. Science, 247, 198-201. Brink K. H., J. S. Allen and R. S. Smith (1978) A study of low-frequency fluctuations near the Peru coast. Journal of Physical Oceanography, 8, 1025-1041. Cressman G. P, (1959) An operational objective analysis system. Monthly Weather Review, 87,367-374. Davis R. E. (1976) Predictability of sea surface temperature and sea level pressure anomalies over the north Pacific Ocean. Journal of Physical Oceanography, 6, 249-266. Gordon H. B. and B. G. Hunt (1991) Drought, floods and sea-surface temperature anomalies: a modelling approach. International Journal of Climatology, 11,347-365. Jury M. R. (1988) Case studies of the response and spatial distribution of wind-driven upwelling off the coast of Africa: 29-34°S. Continental Shelf Research, 8(11), 1257-1271. Jury M, R. and G. B. Brundrit (1992) Temporal organization of upwelling in the southern Benguela ecosystem by resonant coastal trapped waves in the ocean and atmosphere. South African Journal of Marine Science, 12, 21%224. Jury M. R., C. McQueen and K. M. Levey (1994a) SOI and QBO signals in the African region. Theoretical & Applied Climatology. Jury M. R., B. Pathack, C. J, de W. Rautenbach and J. VanHeerden (1994b) The relationship between drought over southern Africa and warming of SST in the central Indian Ocean: statistical correlations and GCM results. Lindesay J. A. (1988) The southern oscillation and atmospheric circulation changes over southern Africa. Ph. D. Thesis, University of Witwatersrand, Johannesburg, 284 pp. Mason S. J. (1992) Sea surface temperatures and South African rainfall variability. Ph.D. Thesis, University of Witwatersrand, Johannesburg, 235 pp. McLain D. R., R. E. Brainard and J. G. Norton (1985) Anomalous warm events in eastern boundary current systems. Report of the California Cooperative Oceanic Fisheries Invest, 26, 51~4.

1354

M . R . Jury and S. Courtney

Nicholson S. E. and D. Entekhabi (1987) Rainfall variability in equatorial and southern Africa: relationship with sea surface temperatures along the SW coast of Africa. Journal of Climate & Applied Meteorology, 26,561578. Parker B. A., B. Pathack and M. R. Jury (1992) Atlas of southern Africa climatic variability. Technical Report, Department of Oceanography, University of Cape Town. Pathack B. (1993) Modulation of South African summer rainfall by global climatic processes. Ph.D. Thesis, Oceanography Department, University of Cape Town, 293 pp. Sciremammano F. Jr (1979) A suggestion for the presentation of correlations and their significance levels. Journal of Physical Oceanography, 9, 1273-1276. Shannon L. V. and J. J. Agenbag (1990) A large scale perspective on interannual variability in the environment of the SE Atlantic. South African Journal of Marine Science, 9, 161-168. Shannon L. V., J. J. Agenbag, N. D. Walker and J. R. E. Lutjeharms (1990) A major perturbation in the Agulhas retroflection area in 1986. Deep Sea Research, 37(3A), 493-512. Shannon L. V., A. J. Boyd, G. B. Brundrit and J. Taunton-Clark (1986) On the existence of an El Nifio-type phenomenon in the Benguela system. Journal of Marine Research, 44(3), 495-520. Shannon L. V., R. J. M. Crawford, G. B. Brundrit and L. G. Underhill (1988) Responses offish populations in the Benguela ecosystem to environmental change. Journal du Conseil Permanent pour l'Exploration de la Mer, 45(1), 5-12. Shelton P. A., A. J. Boyd and M. J. Armstrong (1985) The influence of large-scale environmental processes on neritic fish populations in the Benguela Current system. Report of the California Cooperative Oceanic Fisheries Invest, 26, 72-92. Slutz R. J., S. J. Lubkcr, J. D. Hiscox, S. D. Woodruff, R. L. Jenne, D. H. Joseph, P. M. Steurer andJ. D. Elms (1985) COADS: Comprehensive Ocean-Atmosphere Data Set, Release 1, Climate Research Programmc, Environmental Research Laboratory, Boulder, CO, 262 pp. Taunton-Clark J. and F. Kamstra (1988) Aspects of marine environmental variability near Cape Town, 19601985. South African Journal or Marine Science, 6,273-283. Taunton-Clark J. and L. V. Shannon (1988) Annual and interannual variability in the southeast Atlantic during the 20th century. South African Journal of Marine Science, 6, 97-106. Walker N. D. (1987) lntcrannual sea surface temperature variability and associated atmospheric forcing within the Benguela system. South African Journal of Marine Science, 5,121-132. Woodruff S. D., R. J. Slutz, R. L. Jennc and P. M. Steurer (1987) A comprehensive ocean-atmosphere data set. Bulletin--A merican Meteorological Society, 68, 1239-1250.