15
Global and Regional Climate Response to Late Twentieth-Century Warming over the Indian Ocean JAMES J. LUFFMAN,ANDRE ´ A S. TASCHETTO, AND MATTHEW H. ENGLAND Climate Change Research Centre, University of New South Wales, Kensington, New South Wales, Australia (Manuscript received 10 February 2009, in final form 15 October 2009) ABSTRACT The global and regional climate response to a warming of the Indian Ocean is examined in an ensemble of atmospheric general circulation model experiments. The most marked changes occur over the Indian Ocean, where the increase in tropical SST is found to drive enhanced convection throughout the troposphere. In the extratropics, the warming Indian Ocean is found to induce a significant trend toward the positive phase of the northern annular mode and also to enhance the Southern Hemisphere storm track over Indian Ocean longitudes as a result of stronger meridional temperature gradients. Convective outflow in the upper levels over the warming Indian Ocean leads to a trend in subsidence over the Indian and Asian monsoon regions extending southeastward to Indonesia, the eastern Pacific, and northern Australia. Regional changes in Australia reveal that this anom- alous zone of subsidence induces a drying trend in the northern regions of the continent. The long-term rainfall trend is exacerbated over northeastern Australia by the anomalous anticyclonic circulation, which leads to an offshore trend in near-surface winds. The confluence of these two factors leads to a drying signal over north- eastern Australia, which is detectable during austral autumn. The rapid, late twentieth-century warming of the Indian Ocean may have contributed to a component of the observed drying trend over northeastern Australia in this season via modifications to the vertical structure of the tropical wind field. 1. Introduction The Indian Ocean plays a fundamental role in deter- mining regional climate. Changes in sea surface tem- perature (SST) in the Indian Ocean have been linked to a significant proportion of rainfall variability across Asia (Ashok et al. 2001), Africa (Giannini et al. 2003; Washington and Preston 2006), and Australia (e.g., Nicholls 1989; Drosdowsky and Chambers 2001; England et al. 2006). Modes of variability unique to the Indian Ocean have recently been identified and are now under- stood as important drivers of regional climate variability. The Indian Ocean dipole (IOD) was described by Saji et al. (1999) as a temporary modification to the east– west structure of the tropical wind field, which can be measured as the difference between SST in the western and eastern tropical Indian Ocean. The IOD exhibits significant correlations with regional rainfall, including Africa, southern Asia (Saji and Yamagata 2003), and western and southern Australia (Ashok et al. 2003; Ummenhofer et al. 2008, 2009; Cai et al. 2009). Despite the relatively large number of studies of seasonal-to-interannual Indian Ocean variability, few studies have been conducted on the impact of Indian Ocean warming on regional climate. This is the central goal of the present study. The warming trend has been signifi- cant across most of the basin, with only a small area in the southwest showing a decreasing trend (Fig. 1a). Despite prominent interannual to decadal variations (Fig. 1b), the most notable feature of the temperature record is the long-term trend seen during 1970–2005. The basin- averaged Indian Ocean SST has increased since 1970 at a rate of 0.938C century 21 , with local rates as high as 2.928C century 21 . The trend in SST over the Indian Ocean exhibits only minimal seasonal variation (figure not shown), perhaps unsurprisingly given the high per- sistence of SST on monthly time scales. This uniformity between seasons in the trend in both spatial pattern and magnitude also suggests that it is forced by mechanisms operating over significant time and space scales, most likely as a result of radiation changes associated with anthropogenic climate change (Du and Xie 2008). However, reconstructions of heat flux into the Indian Corresponding author address: Andre ´a S. Taschetto, Climate Change Research Centre, University of New South Wales, Kensington, NSW 2052, Australia. E-mail: [email protected] 1660 JOURNAL OF CLIMATE VOLUME 23 DOI: 10.1175/2009JCLI3086.1 Ó 2010 American Meteorological Society

Global and Regional Climate Response to Late Twentieth ...web.science.unsw.edu.au/~andrea/papers/Luffman_etal_2010.pdf · Global and Regional Climate Response to Late Twentieth-Century

  • Upload
    vomien

  • View
    216

  • Download
    2

Embed Size (px)

Citation preview

Page 1: Global and Regional Climate Response to Late Twentieth ...web.science.unsw.edu.au/~andrea/papers/Luffman_etal_2010.pdf · Global and Regional Climate Response to Late Twentieth-Century

Global and Regional Climate Response to Late Twentieth-CenturyWarming over the Indian Ocean

JAMES J. LUFFMAN, ANDREA S. TASCHETTO, AND MATTHEW H. ENGLAND

Climate Change Research Centre, University of New South Wales, Kensington, New South Wales, Australia

(Manuscript received 10 February 2009, in final form 15 October 2009)

ABSTRACT

The global and regional climate response to a warming of the Indian Ocean is examined in an ensemble of

atmospheric general circulation model experiments. The most marked changes occur over the Indian Ocean,

where the increase in tropical SST is found to drive enhanced convection throughout the troposphere. In the

extratropics, the warming Indian Ocean is found to induce a significant trend toward the positive phase of the

northern annular mode and also to enhance the Southern Hemisphere storm track over Indian Ocean longitudes

as a result of stronger meridional temperature gradients. Convective outflow in the upper levels over the warming

Indian Ocean leads to a trend in subsidence over the Indian and Asian monsoon regions extending southeastward

to Indonesia, the eastern Pacific, and northern Australia. Regional changes in Australia reveal that this anom-

alous zone of subsidence induces a drying trend in the northern regions of the continent. The long-term rainfall

trend is exacerbated over northeastern Australia by the anomalous anticyclonic circulation, which leads to an

offshore trend in near-surface winds. The confluence of these two factors leads to a drying signal over north-

eastern Australia, which is detectable during austral autumn. The rapid, late twentieth-century warming of the

Indian Ocean may have contributed to a component of the observed drying trend over northeastern Australia in

this season via modifications to the vertical structure of the tropical wind field.

1. Introduction

The Indian Ocean plays a fundamental role in deter-

mining regional climate. Changes in sea surface tem-

perature (SST) in the Indian Ocean have been linked

to a significant proportion of rainfall variability across

Asia (Ashok et al. 2001), Africa (Giannini et al. 2003;

Washington and Preston 2006), and Australia (e.g.,

Nicholls 1989; Drosdowsky and Chambers 2001; England

et al. 2006). Modes of variability unique to the Indian

Ocean have recently been identified and are now under-

stood as important drivers of regional climate variability.

The Indian Ocean dipole (IOD) was described by Saji

et al. (1999) as a temporary modification to the east–

west structure of the tropical wind field, which can be

measured as the difference between SST in the western

and eastern tropical Indian Ocean. The IOD exhibits

significant correlations with regional rainfall, including

Africa, southern Asia (Saji and Yamagata 2003), and

western and southern Australia (Ashok et al. 2003;

Ummenhofer et al. 2008, 2009; Cai et al. 2009).

Despite the relatively large number of studies of

seasonal-to-interannual Indian Ocean variability, few

studies have been conducted on the impact of Indian

Ocean warming on regional climate. This is the central goal

of the present study. The warming trend has been signifi-

cant across most of the basin, with only a small area in the

southwest showing a decreasing trend (Fig. 1a). Despite

prominent interannual to decadal variations (Fig. 1b), the

most notable feature of the temperature record is the

long-term trend seen during 1970–2005. The basin-

averaged Indian Ocean SST has increased since 1970

at a rate of 0.938C century21, with local rates as high as

2.928C century21. The trend in SST over the Indian

Ocean exhibits only minimal seasonal variation (figure

not shown), perhaps unsurprisingly given the high per-

sistence of SST on monthly time scales. This uniformity

between seasons in the trend in both spatial pattern and

magnitude also suggests that it is forced by mechanisms

operating over significant time and space scales, most

likely as a result of radiation changes associated with

anthropogenic climate change (Du and Xie 2008).

However, reconstructions of heat flux into the Indian

Corresponding author address: Andrea S. Taschetto, Climate

Change Research Centre, University of New South Wales,

Kensington, NSW 2052, Australia.

E-mail: [email protected]

1660 J O U R N A L O F C L I M A T E VOLUME 23

DOI: 10.1175/2009JCLI3086.1

� 2010 American Meteorological Society

Page 2: Global and Regional Climate Response to Late Twentieth ...web.science.unsw.edu.au/~andrea/papers/Luffman_etal_2010.pdf · Global and Regional Climate Response to Late Twentieth-Century

Ocean over the late twentieth century disagree on both

the sign and magnitude of the net heat flux (Yu et al.

2007), with most datasets actually exhibiting a negative

flux south of 158S, suggesting that internal oceanic cir-

culation processes may have played a significant role in

a component of the observed surface warming.

Although the subject of Indian Ocean variability and

teleconnections with regional climate has been studied

extensively, the role of this emerging SST trend has re-

ceived far less attention. The sustained SST trend in the

Indian Ocean will probably modulate the historic tele-

connections in this region—for example, via the IOD—to

rainfall. The observed warming in Indian Ocean SST

adjacent to Africa has been suggested to have forced a

net shift of convection from land toward the ocean, re-

ducing rainfall in the Sahel region (Giannini et al. 2003).

Using an atmospheric general circulation model (AGCM)

ensemble forced with observed global SST over the pe-

riod 1930–2000, the authors found a low-frequency trend

in precipitation over the Sahel similar to observations,

with a marked drying from the 1960s to the 1980s. Their

study suggests that low-frequency trends in precipitation

can be accounted for by corresponding trends in Indian

Ocean SST.

It has also been argued that the warming of the trop-

ical Indian Ocean might have contributed to unusually

dry and warm conditions over North America and west-

ern Europe during the 1998–2002 period (Hoerling and

Kumar 2003). A subsequent study considered tropical

Indian Ocean SSTs as a separate case and found that a

large proportion of the observed Northern Hemisphere

circulation changes over the late twentieth century could

be forced simply by the linear trend in Indian Ocean

SSTs (Hoerling et al. 2004). The authors also found that

in 80% of model realizations forced only with warm

Indian Ocean SST, the North Atlantic Oscillation index

exhibited a positive trend, leading to a decrease in pre-

cipitation over the Mediterranean region and an increase

over northern continental Europe. The tropical SST

trend in the Indian Ocean was found to project onto the

Northern Hemisphere extratropical climate through at-

mospheric eddy feedbacks and modifications to the win-

tertime storm track. It is as yet unknown whether such a

mechanism exists in the Southern Hemisphere, and in-

deed what the impacts of an Indian Ocean warming on

the Southern Hemisphere extratropical climate might be.

It is likely that the surface warming of the Indian Ocean

also has a signal in regional precipitation, including over

Australia. The rainfall decline in Australia has been

particularly pronounced over the northeast, where de-

creases in mean annual rainfall have been as large as

200 mm decade21 since 1971 (Taschetto and England

2009). As Australian climate is affected by Pacific SST

variability, one may speculate that the decrease along

the east coast could be a result of changes in the El Nino

pattern over the past few decades. However, given the

absence of substantial trends in ENSO indices over the

twentieth century (Nicholls 2008), it is unclear whether

changes in El Nino activity could have forced the ob-

served Australian east coast drying trend. The causes of

the post-1950 drying trend over northeastern Australia

remain unknown (Nicholls 2006).

Trends in regional rainfall over Australia and Indian

Ocean rim nations warrant further investigation, espe-

cially considering their continuation into the early part of

the twenty-first century. Given the extensive linkages be-

tween regional climate and the Indian Ocean on seasonal

to interannual time scales, the role of lower-frequency

variability and trends in the Indian Ocean needs to be

considered. In this study, we investigate the atmospheric

response to a warming Indian Ocean using an ensemble

set of experiments with an AGCM forced by a linear

trend component of observed SST during the late twen-

tieth century. The configuration and forcing of the model

experiments and an evaluation of the relative contribu-

tions from the Indian and Pacific oceans are outlined in

section 2. Section 3 describes the forced changes in the

FIG. 1. (a) Linear trend in annual SST (8C century21) in the In-

dian Ocean over the period 1970–2005. Dashed lines indicate the

linear ramp at the southern boundary of the model forcing region.

(b) Time series of annual Indian Ocean basin-averaged SST (8C)

between 208S–208N and 408–1108E. Anomalies are relative to the

1970–2005 climatology. The HadISST1 dataset (Rayner et al. 2003)

was used to compute these trends.

1 APRIL 2010 L U F F M A N E T A L . 1661

Page 3: Global and Regional Climate Response to Late Twentieth ...web.science.unsw.edu.au/~andrea/papers/Luffman_etal_2010.pdf · Global and Regional Climate Response to Late Twentieth-Century

global atmospheric circulation, and the resultant re-

gional trends are analyzed in section 4, with a focus on

Australia. Finally, section 5 discusses and concludes the

main findings of this study.

2. Atmospheric model and experimental design

a. Atmospheric model

The atmospheric model used in this study is the Com-

munity Atmosphere Model, version 3 (CAM3), devel-

oped by the National Center for Atmospheric Research

(NCAR). The CAM3 uses Eulerian spectral dynamics

with a T42 horizontal resolution (approximately 2.88 in

both latitude and longitude) and 26 levels in a hybrid

sigma pressure system in the vertical. The AGCM is

forced by prescribed global monthly SST values, as well

as sea ice fraction, linearly interpolated at each time step

to form the bottom boundary conditions for the atmo-

sphere. A more thorough description of CAM3 is given

by Collins et al. (2006), and an analysis of the model’s

hydrological cycle is given by Hack et al. (2006).

b. Experimental forcing and configuration

The premise of our experimental design is to analyze

the modifications to regional climate forced by the ob-

served warming of the Indian Ocean. In the main per-

turbation experiment we are interested in particular in the

climate modifications arising from the lowest-frequency

aspects of the observed SST changes—that is, the linear

trend component. By conducting experiments forced

solely by the observed changes in Indian Ocean SST, we

can elucidate the response of regional climate in iso-

lation from other potential forcing agents. Therefore,

the main experiment ensemble set applies SST pertur-

bations to the monthly varying climatological mean only

over the Indian Ocean region, defined as the ocean basin

north of 258S, bordered to the west by the African con-

tinent, to the east by the Indonesian archipelago, and to

the north by the Asian continent (Fig. 1a). Everywhere

else the boundary forcing is the climatological monthly

mean SST and sea ice fraction data from the Hadley

Centre (HadISST1; Rayner et al. 2003).

In this main Indian Ocean warming experiment, re-

ferred to as ITR, a linear trend over the period 1970–2005

is fitted to the SST time series at each model grid point,

with a mean 1970–2005 SST change of 10.938C century21

over the basin. The linear trend in SST at each grid

point is added to the monthly climatological values

over the period 1970–2005, so that the only variability

in SST is the seasonal cycle superimposed upon the

observed linear trend. At 158S, a ramp of width 108 in

latitude is used to border the forced region to minimize

edge effects (see Fig. 1a), and across this ramp the im-

posed trend is damped linearly to a value of zero at the

outside boundary.

To filter out the internal variability generated by the

atmosphere, an ensemble of experiments was carried out

with identical SST forcing as described previously, but

each with slightly varied atmospheric initial conditions,

taken at consecutive two-day intervals. The number of

ensemble members was five, yielding a total of 175 yr of

integration time. Although there is no upper bound on the

number of integrations desired, the computational load of

this ensemble size is already substantial (Taschetto and

England 2008). We employ a bootstrap statistical analysis

to quantify the significance of modeled trends (Efron

1982). This technique can be particularly effective in

dealing with climate data from non-Gaussian fields, be-

cause other hypothesis tests often make invalid assump-

tions about the normality of the sampling distribution

(Kiktev et al. 2003). Our approach determines whether

the mean of the bootstrap distribution of ensemble

member linear trends at each model grid point is signifi-

cantly different from zero at the 90% confidence level

using a bootstrap t test. In sections 3 and 4, we find a

number of major features in the atmospheric circulation

common to the output of all five ensemble members,

suggesting that this modest ensemble size is sufficient for

our purposes.

c. Assessing the relative contributions from Indianand Pacific ocean warming

Additional experiment sets were performed in order

to assess the relative contributions from surface warm-

ing of the Indian and Pacific oceans during the late

twentieth century (see Table 1). These additional sim-

ulations allow us to examine the linearity or otherwise of

the regional climate response to Indian Ocean surface

warming and also to support the results of our primary

experiment set forced only by the warming trend in

the Indian Ocean. The AGCM experiments consisted

of seven-member multidecadal ensemble sets using

CAM3, integrated from 1950 to 2006 and forced by

monthly varying SST from HadISST1 in the different

domains. These experiments differ from the ITR by in-

cluding not only the observed trend in SST but also the

interannual variability. For consistency, only the 1971–

2005 period from these additional experiments is ana-

lyzed here.

The geometry of the forcing regions used in the addi-

tional experiment set is as follows: The first set was forced

by the trending SST field only in the Indian Ocean north of

258S, which also includes year–year variability (ITR1VR).

Outside this region, the monthly varying climatological

1662 J O U R N A L O F C L I M A T E VOLUME 23

Page 4: Global and Regional Climate Response to Late Twentieth ...web.science.unsw.edu.au/~andrea/papers/Luffman_etal_2010.pdf · Global and Regional Climate Response to Late Twentieth-Century

SST is applied (i.e., no trend and no year–year variabil-

ity). Similarly, the second set was forced by the trending

and interannually varying SST only in the Pacific Ocean,

from 308S to 308N, with monthly mean SST climatology

elsewhere (PTR1VR). A third experiment set was forced

by the trending and interannually varying SST in both the

Indian and Pacific oceans (IPTR1VR). Beyond the domain

of varying SST in each experiment set, anomalies were

damped to zero over a 108 latitude and longitude band to

avoid unrealistic atmospheric circulation resulting from

edge effects from the boundary forcing.

Figure 2 shows the mean linear trend in regional

precipitation from each of these three ensemble sets. Of

course, we do not expect a priori that any of the model

ensembles should look like the observed trends, as the

experiments include just one forcing agent (i.e., SST) out

of many possible drivers (greenhouse, ozone, aerosols,

etc). Nevertheless, a general comparison with National

Centers for Environmental Prediction (NCEP)–NCAR

reanalysis (not shown) indicates that the large-scale rainfall

pattern is captured by the IPTR1VR experiment (Fig. 2c).

For instance, a general increase of precipitation is seen

over the tropical Indian Ocean in the reanalysis product,

consistent with the SST warming over the region. In

addition, NCEP–NCAR reanalysis shows a rainfall reduc-

tion over India, the Arabian Sea, and the Bay of Bengal

(not shown), which is also captured by the IPTR1VR

simulation (Fig. 2c). Despite this large agreement be-

tween the simulated trends and the precipitation pattern

from the NCEP–NCAR reanalysis over certain regions,

the same cannot be said for the Climate Prediction Center

Merged Analysis of Precipitation (CMAP) dataset—

for example, CMAP and NCEP–NCAR reanalysis ex-

hibit opposite trends over the tropical Indian Ocean

(not shown). Previous studies have reported a discrep-

ancy between the long-term rainfall trends over the trop-

ical Indian Ocean between observations and reanalysis

products (e.g., Arakawa and Kitoh 2004; Kumar et al.

2004; Quartly et al. 2007). It is possible that the re-

analysis product is biased toward a larger positive rain-

fall trend, because their models use SST as boundary

conditions in the data assimilation. While the rainfall

trends in the reanalysis are consistent with the SST

warming, precipitation trends from CMAP agree with

the long-term changes in surface salinity over the Indian

Ocean (Durack and Wijffels 2009, manuscript submitted

to J. Climate).

Despite discrepancies in Indian Ocean rainfall trends

between observations, both CMAP and the reanalysis

data show the east–west pattern over Australia, consis-

tent with the Australian Bureau of Meteorology dataset.

None of the experiments, however, simulate an east–

west pattern of rainfall trend over Australia, unlike the

observations. Instead, the Indian Ocean SST warming

generates a negative trend in the north and a positive

trend in the southern regions of Australia (Fig. 2a),

whereas broadly the opposite pattern is exhibited for the

experiment forced by trending Pacific SST (Fig. 2b). The

experiment IPTR1VR shows a decreasing trend across

the continent, which is particularly strong in the north

and east. This suggests that while the drying trend in the

east may be partially explained by changes in the trop-

ical Indian and Pacific SST, the rise of rainfall in the

northwest must be related to other factors.

In general, there is a marked linearity in atmospheric

response to the forcing in the experiment with SST trends

in the Indian and Pacific oceans combined, relative to the

response in the individually forced sets. Nonlinear re-

sponses in rainfall trend are seen over the South China

Sea and Malaysia. However, the overall linearity is quite

clear and this increases the confidence in the in-

terpretation of results from the ITR experiment forced

only by the linear trend in Indian Ocean SST. Focusing on

the region over the central Indian Ocean and northern

Australia, the experiment set forced only by Pacific

Ocean SST exhibits a precipitation trend of opposite sign

to the Indian Ocean and combined sets. This suggests that

SST trends over the Indian Ocean may be a potential

factor in forcing trends over these regions. In contrast, the

Australian region south of approximately 188S shows

a rainfall decline apparent primarily in the PTR1VR ex-

periment. The modeled decrease over northern and

northeastern Australia in the combined forced IPTR1VR

experiment, and in the ITR1VR set, is in agreement with

the results of the ITR experiment as outlined in section 4.

This aspect of the climate response to Indian Ocean

TABLE 1. Outline of the experiments analyzed in this study. Unless otherwise indicated in the text, the results are described for the ITR

experiment from 1971 to 2005.

Experiment Region Forcing Ensemble size

Control Global oceans Monthly SST climatology 5

ITR Tropical Indian Ocean SST trend 5

ITR1VR Tropical Indian Ocean SST trend and variability 7

PTR1VR Tropical Pacific Ocean SST trend and variability 7

IPTR1VR Tropical Indian and Pacific oceans SST trend and variability 7

1 APRIL 2010 L U F F M A N E T A L . 1663

Page 5: Global and Regional Climate Response to Late Twentieth ...web.science.unsw.edu.au/~andrea/papers/Luffman_etal_2010.pdf · Global and Regional Climate Response to Late Twentieth-Century

warming is in agreement with observed trends over far

northeastern Australia, as outlined in section 4c.

3. Global response to Indian Ocean warming

Here, we analyze the circulation changes imposed by

the linear trend component of Indian Ocean SST and

their projection onto global climate from the ITR exper-

iment. A global response to the warming Indian Ocean

can be seen in zonally averaged zonal winds (Fig. 3)

throughout the depth of the troposphere, suggesting a

large-scale reorganization of the wind field at multiple

levels in response to the imposed surface heating. The

trends in the Northern Hemisphere extratropics re-

semble the zonal wind anomaly structure of the positive

phase of the northern annular mode (NAM) shown by

Thompson et al. (2003), with an increase in westerlies

north of about 408N and a weakening to the south. To

investigate this modification further, we conduct an em-

pirical orthogonal function (EOF) analysis of modeled

sea level pressure (SLP) poleward of 208 latitude. The

leading spatial pattern and associated principal compo-

nent time series of SLP under the condition of a warm-

ing Indian Ocean are displayed in Fig. 4.

In the Northern Hemisphere, the first EOF mode

(Fig. 4b) resembles the SLP signature of the NAM, with

positive eigenvectors extending from North America to

western Europe and also over the North Pacific. The

principal component time series for the leading EOF

mode (Fig. 4d) exhibits a notable positive trend, which is

found to be significant at the 90% confidence level

using the bootstrap t test. This result is consistent with

the findings of Hoerling et al. (2004)—suggesting that

a positive trend in the NAM, and the associated re-

gional climate change in the Northern Hemisphere ex-

tratropics, can be forced significantly by a warming Indian

Ocean.

In the Southern Hemisphere, the strongest response in

zonal winds (Fig. 3) is an enhancement of the subtropical

jet, which is driven by divergent outflow associated with

increased convection over the tropical Indian Ocean. The

positive trend in zonal winds associated with this en-

hancement is strongest near 200 hPa, but also extends to

the surface from 208 to 408S, corresponding to a weaken-

ing of the subtropical easterly trade winds in this band.

Over the Southern Hemisphere extratropics, the pattern

of zonally averaged trends is weaker than that exhibited

in the Northern Hemisphere and has a different zonal and

vertical structure. The spatial pattern of the leading EOF

in the Southern Hemisphere SLP (Fig. 4a) has a roughly

annular form similar to the observed southern annular

mode (SAM), described by Thompson and Wallace (2000),

although negative eigenvectors extend equatorward at

the Indian Ocean longitudes, weakening the ‘‘annular’’

shape. The principal component time series (Fig. 4c) for

the leading EOF shows a slight negative trend and is not

significant at the 90% confidence level. Given these re-

sults, we conclude that it is unlikely that the observed

positive trend in the SAM (Marshall 2003) has been

forced in any significant way by a warming of the Indian

Ocean. Other factors such as changes in stratospheric

FIG. 2. Ensemble mean linear trends in annual precipitation

(mm century21) during 1971–2005 from the atmospheric model

forced by observed trending and monthly and interannually vary-

ing SST in the (a) Indian, (b) Pacific, and (c) Indian and Pacific

oceans combined. Areas within the dashed lines are significant at

the 90% level based on a Student’s t test.

1664 J O U R N A L O F C L I M A T E VOLUME 23

Page 6: Global and Regional Climate Response to Late Twentieth ...web.science.unsw.edu.au/~andrea/papers/Luffman_etal_2010.pdf · Global and Regional Climate Response to Late Twentieth-Century

ozone (Gillett and Thompson 2003) and greenhouse

gases (Cai et al. 2003) have clearly played a greater role.

It is worth noting that negative eigenvectors in the

Southern Hemisphere extend equatorward at Indian

Ocean longitudes in Fig. 4a. This suggests a regional

weakening of the SAM (i.e., toward its negative phase),

indicating that the Indian Ocean warming may have a

compensating effect upon extratropical climate over these

longitudes in the Southern Hemisphere. We examine the

mechanisms associated with this zonally asymmetric mod-

ification to the SAM in a discussion of regional dynamics

in section 4.

4. Regional atmospheric response

a. Annual mean

The direct response to the warming of Indian Ocean

SST is an overall decrease in sea level pressure over the

ocean basin (Fig. 5b) and enhanced deep convection in

the atmosphere, which is clearly visible in Fig. 5a as an

increase in upward motion over the tropical Indian

Ocean. This feature in our experiment is consistent with

the results of Giannini et al. (2003), who found that the

observed SST warming caused preferential and in-

creased convection over the Indian Ocean. We find this

trend toward enhanced uplift over the ocean to be a ro-

bust response; it is present in each of the five ensemble

members of the ITR experiment as well as the IPTR1VR

experiment forced by the actual observed SST over both

the Indian and Pacific oceans.

The convection-driven uplift over the tropical Indian

Ocean leads to upper-tropospheric divergence, visible in

the divergent component of horizontal winds at 220 hPa

in Fig. 5a. At lower levels a marked convergence in near-

surface winds can be seen over much of the tropical

Indian Ocean (Fig. 5b), causing a reorganization of the

local wind field in the region. Farther east is a corre-

sponding suppression of tropical convection over the

Bay of Bengal, northern Australia, and the Coral Sea,

shown as a broad area of subsiding air, associated upper-

level convergence and positive trends in sea level pres-

sure. The zonal and vertical structure of the vertical

velocity anomalies is examined in Fig. 6, revealing that

the strong east–west sign reversal seen in the 500-hPa

velocities is consistent throughout the depth of the

troposphere. Again, this feature is also present in the

IPTR1VR ensemble forced by SST in the Indian and

Pacific oceans combined.

Another vertical circulation response to the imposed

SST warming is enhanced uplift over the southern Indian

Ocean, outside the region of forced warming. Surface

winds (Fig. 5b) reveal an anomalous cyclonic circulation

centered near 408S, 658E and enhanced westerlies from

FIG. 3. Ensemble mean linear trend in global annual zonally averaged zonal winds [(m s21)

century21] in the atmosphere in response to Indian Ocean warming. The contour interval is

0.5 (m s21) century21. The zero contour is shown in heavy black line, and positive (eastward)

contours are dashed.

1 APRIL 2010 L U F F M A N E T A L . 1665

Page 7: Global and Regional Climate Response to Late Twentieth ...web.science.unsw.edu.au/~andrea/papers/Luffman_etal_2010.pdf · Global and Regional Climate Response to Late Twentieth-Century

308 to 408S, indicating a tendency toward stronger baro-

clinicity and intensification of the storm track over the

southern Indian Ocean, evidenced by a reduction in geo-

potential height (not shown). This increased cyclogenesis

over the southern Indian Ocean is also apparent in the

spatial structure of the SLP response (Fig. 4a), where the

annular structure deviates equatorward at 408–808E.

Although we do not conduct a detailed analysis of the

model storm track trends, the enhancement of mid-

latitude depressions over the southern Indian Ocean in

the model is likely a result of the increased meridional

temperature gradient arising from the warmed tropical

SST. It is also likely a result of the increase in moisture

content of prefrontal air masses from 708 to 908E, where

lifting results in condensation and the release of latent

heat, warming the overlying air and resulting in further

intensification of midlatitude systems.

To analyze the nature of changes to advection of

moisture in the lower and middle troposphere arising

from the warming of the Indian Ocean, Fig. 7 shows the

trend in specific humidity and winds, integrated from

800 to 500 hPa. In the control run forced by climato-

logical monthly SST, the levels over 800–500 hPa ac-

count for 56% of global-averaged moisture transport,

calculated using the bulk formula Q � kUk. Figure 7a

shows that the increased uplift over the tropical Indian

Ocean and subsidence to the north of Australia result in

an east–west dipole in specific humidity trends over these

areas. This dipole is consistent with the vertical circulation

described previously and is a robust feature seen in both

the ITR and IPTR1VR simulations. Over the tropical In-

dian Ocean, warmer SSTs increase evaporation rates and

lead to higher specific humidity, while the deep convec-

tion drives a strong convergence of the tropical wind field.

FIG. 4. (top) Spatial loading pattern of the first EOF of (a) Southern and (b) Northern Hemisphere SLP (Pa)

calculated poleward of 208 latitude for the ensemble mean of the Indian Ocean warming experiments, and associated

(bottom) normalized annual means of the principal component time series of the (c) Southern and (d) Northern

Hemisphere.

1666 J O U R N A L O F C L I M A T E VOLUME 23

Page 8: Global and Regional Climate Response to Late Twentieth ...web.science.unsw.edu.au/~andrea/papers/Luffman_etal_2010.pdf · Global and Regional Climate Response to Late Twentieth-Century

FIG. 5. Ensemble mean trends in atmospheric circulation over 1971–2005 in response to

Indian Ocean warming: (a) annual vertical velocity (shaded) at 500 hPa [(Pa s21) century21] and

divergent component of horizontal annual wind (arrows) at 220 hPa [(m s21) century21]; (b) sea

level pressure (hPa century21) and near-surface annual winds [(m s21) century21]. Negative

(positive) values in the vertical velocity represent upward (downward) motion. Trend values

determined to be significantly different from zero at the 90% confidence level using a bootstrap

t test in (a) and Student’s t test in (b) are bounded by dashed lines, trend vectors significantly

different from zero at the 90% confidence level are shown in black.

1 APRIL 2010 L U F F M A N E T A L . 1667

Page 9: Global and Regional Climate Response to Late Twentieth ...web.science.unsw.edu.au/~andrea/papers/Luffman_etal_2010.pdf · Global and Regional Climate Response to Late Twentieth-Century

Farther east, an anomalous anticyclonic circulation cen-

tered near 108S, 1308E coincides with an area of de-

creasing moisture that takes a maximum value over the

Coral Sea and northeastern Australia. This region of

atmospheric drying coincides with the region of anom-

alous subsidence seen in Fig. 5a.

Linear trends in modeled annual rainfall during

1971–2005 are shown in Fig. 8. The regional response

is dominated by an increase over the tropical Indian

Ocean and a decrease over India, the Bay of Bengal, the

South China Sea extending to the Indonesian archipel-

ago and the Coral Sea. This pattern is highly correlated

with changes to vertical velocity (Fig. 5a) and horizontal

moisture transport (Fig. 7a), indicating that the precip-

itation change in the model is accounted for by the mod-

ifications to the regional atmospheric circulation described

earlier. The increase over the tropical Indian Ocean and

decrease over surrounding tropical land areas is consis-

tent with the results of Giannini et al. (2003), whereby a

reduced land–sea temperature gradient forces preferen-

tial convection over the ocean and weakens the monsoon

mechanism.

It is of interest to compare the spatial pattern of the

regional rainfall response in the ITR experiment set with

the response in the IPTR1VR experiment with combined

forcing from both the Indian and Pacific oceans (Fig. 2c).

The pattern of increase over the tropical Indian Ocean,

and decrease over India, the Bay of Bengal, parts of the

Indonesian archipelago, the Coral Sea, and northeastern

Australia is largely retained when the SST forcing is

extended east across the Pacific Ocean. This result in-

dicates that the atmospheric response in the ITR exper-

iment is not particular to the isolated warming imposed

over the Indian Ocean. In contrast, the PTR1VR exper-

iment set (Fig. 2b), forced solely by Pacific Ocean SST

trends plus interannual variability, exhibits a rainfall

increase over northeastern Australia and the Coral Sea,

which is opposite to both the combined forced IPTR1VR

set and the observed trends. To the extent that a linear

superposition between individual and combined atmo-

spheric forcings from the Indian and Pacific oceans ex-

ists, our results suggest that the Indian Ocean is leading

the atmospheric response over land areas of southern

Asia, Indonesia, and tropical Australia.

b. Seasonal evolution

Although the magnitude and spatial pattern of the

applied SST trend forcing is annually constant, a sea-

sonal component to the atmospheric response is possi-

ble, because the trend forcing is projected onto natural

unforced seasonal variability in the atmospheric circu-

lation. In the case of horizontal moisture transport, we

FIG. 6. Ensemble mean trends in atmospheric annual vertical velocity [(Pa s21) century21] in

response to Indian Ocean warming from 1000 to 100 hPa over the Southern Hemisphere

tropics (08–238S), across longitudes spanning the Indian Ocean (608–1158E) and Australia

(1158–1508E). The contour interval is 0.05 (Pa s21) century21. The zero contour is shown in

a heavy black line, and positive (downward) contours are dashed.

1668 J O U R N A L O F C L I M A T E VOLUME 23

Page 10: Global and Regional Climate Response to Late Twentieth ...web.science.unsw.edu.au/~andrea/papers/Luffman_etal_2010.pdf · Global and Regional Climate Response to Late Twentieth-Century

expect that the effect of the trend in Indian Ocean SST

will be nonuniform, given that the flow regimes vary

with season across the region.

Specific humidity and wind trends (Fig. 7b) for the

austral summer [December–January–February (DJF)]

reveal a decrease in atmospheric moisture content over

eastern parts of the Indonesian archipelago, the South

China Sea, and northeastern Australia, which is associ-

ated with the region of subsidence also seen in the annual

trends over this region. In addition, a reduction of mois-

ture is seen over the southeastern Indian Ocean in DJF.

Trends in moisture transport during the austral au-

tumn [March–April–May (MAM)] exhibit a marked

dipole in moisture (Fig. 7c) qualitatively resembling the

annual plot (Fig. 7a) but of greater magnitude. The in-

creasing moisture and wind field convergence associated

with enhanced convective activity reaches a maximum

over the Indian Ocean during MAM (Fig. 7c), with the

corresponding area of subsidence over northeastern

Australia also at a maximum.

With the increasing moisture content of the lower

troposphere over the tropical Indian Ocean at a maxi-

mum in austral autumn, the annual trend toward in-

creased inflow of moisture to extratropical latitudes also

reaches a peak during these months. The increasing trend

of this moisture advection pathway can be seen clearly in

Fig. 7c extending from near 108S, 908E to western and

southern Australia.

During austral winter [June–July–August (JJA)], pat-

terns of atmospheric moisture in the Southern Hemi-

sphere tropics east of 1008E are weak as the ITCZ is at its

northernmost position, corresponding to the dry season

in this region. An area of increasing moisture over the

Indian Ocean south of the equator is advected toward the

southeast (Fig. 7d) as a result of enhanced midlatitude

baroclinic eddies over the southern Indian Ocean, lead-

ing to a trend of increasing moisture transport into the

Southern Hemisphere extratropics.

During austral spring [September–October–November

(SON)], the ITCZ would ordinarily be held north over

the strongly heated south Asian landmass; however, the

warming trend in SST enhances deep convection over the

FIG. 7. Ensemble mean of the (a) annual and (b)–(e) seasonal

trends in the specific humidity [(g kg21) century21] and horizontal

wind [(m s21) century21] in response to Indian Ocean warming,

integrated over vertical levels in the 800–500-hPa band. Trend

values determined to be significantly different from zero at the

90% confidence level using a bootstrap t test are bounded by

dashed lines, trend vectors significantly different from zero at the

90% confidence level are shown in black.

1 APRIL 2010 L U F F M A N E T A L . 1669

Page 11: Global and Regional Climate Response to Late Twentieth ...web.science.unsw.edu.au/~andrea/papers/Luffman_etal_2010.pdf · Global and Regional Climate Response to Late Twentieth-Century

tropical Indian Ocean (Fig. 5a) by weakening the land–

sea temperature gradient and drawing the ITCZ south-

ward of its mean climatological position. The increased

specific humidity over the southern equatorial Indian

Ocean resulting from this enhanced uplift is evident in

Fig. 7e, as is the drying trend over continental South Asia.

With the ITCZ enhanced over the tropical Indian Ocean,

there is an increase in eastward advection of moisture

south of Java, Indonesia, and over subtropical Australia.

c. Australian rainfall response to Indian Oceanwarming

Observed annual trends (Lavery et al. 1997) over

1971–2005 (Fig. 9f) show an east–west pattern associ-

ated with a substantial increase in the northwest and a

drying over the northeast. This observed east–west pat-

tern is not captured in the simulation ensemble means.

Instead, the modeled annual rainfall trends (Fig. 9a) show

a significant decrease over the north and northeast and an

increase along the east coast and over southwest Western

Australia. Although the observed rainfall decrease over

far northeastern Australia corresponds to the modeled

responses in the ITR experiment (Fig. 2a), the strongest

magnitude of the drying trend is obtained in IPTR1VR,

suggesting a combined response to both tropical Indian

and Pacific oceans (Fig. 2c).

The situation is different south of approximately 188S,

where the Indian Ocean warming tends to increase rain-

fall (see Fig. 9a for ITR and Fig. 2a for ITR1VR), whereas

the Pacific Ocean warming tends to decrease rainfall

(Fig. 2b). This suggests that the Pacific Ocean may play

a role in the decline of rainfall over the southeastern re-

gions of Australia. It is important to note, however,

that the magnitude of the modeled rainfall trend across

Australia is generally much weaker than the observations.

The cause of the modeled annual rainfall decrease

over northern and northeastern Australia is an induced

subsidence trend in the Southern Hemisphere tropics

from 1258 to 1608E (Figs. 5 and 6), causing an anoma-

lous divergent anticyclonic circulation over northern

Australia (Fig. 5b). The subsidence-induced drying is

exacerbated by the anomalous anticyclonic circulation,

causing an offshore trend in horizontal winds, both at

the surface (Fig. 5b) and integrated over 800–500 hPa

(Fig. 7a). This offshore trend opposes the northerly wind

FIG. 8. Ensemble mean linear trends in (a) annual and (b)–(e)

seasonal precipitation (mm century21) during 1971–2005 from

CAM3 forced by increasing Indian Ocean SST. Trend values de-

termined to be significantly different from zero at the 90% confi-

dence level using a bootstrap t test are bounded by dashed lines.

1670 J O U R N A L O F C L I M A T E VOLUME 23

Page 12: Global and Regional Climate Response to Late Twentieth ...web.science.unsw.edu.au/~andrea/papers/Luffman_etal_2010.pdf · Global and Regional Climate Response to Late Twentieth-Century

pattern of Watterson (2001), which was found in a model

simulation to be correlated with monthly rainfall over

northeastern Australia (coefficients greater than 0.80

during both January and July). Over northern Australia,

the decreasing trend in specific humidity (Fig. 7a)

combines with an easterly trend in near-surface winds

north of 158S and east of 1258E (Fig. 5b) to weaken the

northwesterly monsoon surge in this region.

The increase in onshore moisture transport from the

southeastern Indian Ocean to the northwestern and west-

ern portions of the continent (Figs. 5b and 7a) combined

with enhanced uplift within the prevailing west coast

trough (Fig. 5a) may account for the slight increase in

modeled precipitation over Western Australia (Fig. 9a),

although the signal is not significant at the 90% level. It

has been shown by Smith (2004) that observed rainfall

trends over the twentieth century showed their most

significant increase over a similar area; however, the

observed 1971–2005 trends are mostly in the range 100–

300 mm century21, whereas our ensemble mean has a

maximum increase of only 110 mm century21. A rainfall

increase over the northwestern regions of Australia closer

in magnitude to the observed was shown by Rotstayn

et al. (2007) to appear in experiments, including Asian

anthropogenic aerosols in a coupled GCM. It is thus

possible that the additional inclusion of increasing an-

thropogenic aerosols would produce a modeled rainfall

increase over the northwestern regions of Australia

closer in magnitude to observations.

The modeled rainfall response in DJF (Fig. 9b) over

Australia is not large, exhibiting a slight decrease of ap-

proximately 150 mm century21 over the north and north-

east, which is not significant at the 90% level. Trends in

specific humidity and wind (Fig. 7b) suggest that the de-

crease over northeastern Australia is accounted for by an

atmospheric drying and offshore trend in winds.

The modeled decrease in precipitation over the north

and northeast is most extensive during MAM, with the

trend significant at the 90% confidence level across a

large area. During MAM the ITCZ is in a transient state,

centered near the equator where the observed SST in-

crease since 1970 (Fig. 1) has been greatest. With the

mean ITCZ location coinciding with the region of greatest

oceanic warming, the available thermal energy for con-

vective instability is maximized and the regional west–east

pattern of uplift and subsidence is greatest.

Farther south, a region of increased MAM rainfall ex-

tends from Western Australia to the southern and south-

eastern (Fig. 9c) regions of the continent. This band of

increased rainfall is associated with an enhanced moisture

advection from the tropical eastern Indian Ocean to the

southern regions of Australia (Fig. 7c).

During JJA, modeled rainfall trends over Australia are

weak and not significant in the ITR experiment (Fig. 9d).

During SON, there is a band of increased rainfall ex-

tending from the northwestern regions of the continent to

the eastern inland (Fig. 9e), as a result of the anomalous

FIG. 9. Linear trends in (a),(f) annual, (b),(g) DJF, (c),(h) MAM,

(d),(i) JJA, and (e),( j) SON mean precipitation over the period

1971–2005 (mm century21) for (left) the ITR experiment and (right)

observations estimated from a monthly gridded rainfall dataset

interpolated from high-quality station data (Lavery et al. 1997).

Areas within the thin black contours are significant at the 90% level

based on a Student’s t test.

1 APRIL 2010 L U F F M A N E T A L . 1671

Page 13: Global and Regional Climate Response to Late Twentieth ...web.science.unsw.edu.au/~andrea/papers/Luffman_etal_2010.pdf · Global and Regional Climate Response to Late Twentieth-Century

moisture advection pathway (Fig. 7e). The observed rain-

fall trend is, however, not significant during this season.

5. Discussion and conclusions

We have sought to examine the climate response to

observed surface warming of the tropical Indian Ocean

through an ensemble of AGCM experiments, forced

only by the linear trend component of Indian Ocean

SST, with monthly varying climatological values else-

where across the globe. To test statistical significance we

conducted a five-member ensemble of the 35-yr runs

with identical forcing. Our forcing methodology allowed

us to examine, to a linear approximation, the particular

atmospheric signal of a warming Indian Ocean in iso-

lation from other potential forcing agents. An additional

set of experiments with the trend-forced SST region

extending east across the Pacific Ocean exhibited a cli-

mate response in agreement with our major findings.

Over the Northern Hemisphere extratropics, the warm-

ing tropical Indian Ocean induces a significant positive

trend in the NAM, which is in agreement with the findings

of Hoerling et al. (2004). No significant trend is forced in

the SAM; however, the Southern Hemisphere storm track

is enhanced over Indian Ocean longitudes as a result of an

anomalous increase in the meridional temperature gra-

dient and release of latent heat in baroclinic eddies.

It is worth noting that the changes simulated in the ITR

experiment are consistent with the responses of the hy-

drological cycle to global warming reported by previous

studies. For instance, Held and Soden (2006) show that in

a warmer-world scenario, the horizontal moisture trans-

port and the pattern of evaporation minus precipitation

tend to be enhanced. The intensification of the hydro-

logical cycle is also seen in our experiments over the In-

dian Ocean, with an enhancement of precipitation over

the tropical latitudes and a reduction over the sub-

tropics, especially in the Northern Hemisphere.

The regional response of the tropical atmosphere to

the linear trend component of Indian Ocean SST occurs

via significant modifications to the vertical circulation

in the tropics and a reorganization of moisture advec-

tion pathways. Surface warming over the tropical Indian

Ocean generates a decrease in sea level pressure and

enhances atmospheric convection, causing a convergence

of the wind field in the lower levels and a weakening of

the monsoon over adjacent land areas of southern Asia.

Convective outflow in the upper levels leads to a broad

area of anomalous subsidence over the South China Sea,

extending south over the western Pacific Ocean to the

Coral Sea. These major features remain robust when the

imposed warming is extended east into the Pacific Ocean.

A schematic diagram showing the regional climate re-

sponse to Indian Ocean warming is given in Fig. 10.

In general the Indian Ocean warming was found to be

efficient in forcing dry conditions over northern and

northeastern Australia. It also captures a broad region of

increased rainfall over the Indian Ocean itself, consistent

with warmer SST driving increased local convective cloud

formation and precipitation. The simulated decrease

over northern and northeastern Australia is a result of an

anomalous zone of subsidence caused by the divergent

outflow from increased convective activity over the trop-

ical Indian Ocean, coupled with an offshore trend in the

lower-tropospheric wind field. The confluence of these

two factors leads to a drying signal over northeastern

Australia detectable in the annual precipitation and during

MAM. Experiments forced with Pacific Ocean warming

FIG. 10. Schematic diagram representing the anomalous regional atmospheric response to

surface heating over the Indian Ocean. Horizontal vectors shown are indicative of the near-

surface layer (black) and the 800–500-hPa layer (red/blue). The vertical bold blue (red) arrows

indicate convection (subsidence), and the green arrow shows anomalous circulation at high

levels of the atmosphere.

1672 J O U R N A L O F C L I M A T E VOLUME 23

Page 14: Global and Regional Climate Response to Late Twentieth ...web.science.unsw.edu.au/~andrea/papers/Luffman_etal_2010.pdf · Global and Regional Climate Response to Late Twentieth-Century

show a significant precipitation decrease in the south and

east but do not generate a rainfall decline over the far

northeastern region where the observed annual trend is

strongest. On the other hand, the IPTR1VR experiment

found to better capture the annual drying trend over the

eastern and northeastern regions Australia.

The modeled increase in annual and MAM rainfall

over Western Australia, caused by a weakening of the

prevailing offshore winds and increased atmospheric

moisture content, is of the same sign as the observed

changes but of smaller magnitude. In addition, the ob-

served east–west pattern of rainfall trends over Australia

is not clearly obtained in our experiments, indicating the

importance of other factors in these changes, such as

greenhouse, ozone, or aerosol forcing; for example, it

could be speculated that the increase of anthropogenic

aerosols over Asia has contributed to the intensification

of rainfall over northwestern Australia. Indeed, Rotstayn

et al. (2007) suggest that rainfall over this area has been

enhanced by a combination of increasing Asian anthro-

pogenic aerosols and warming Indian Ocean sea temper-

atures. However, Shi et al. (2008a) argue that this increased

rainfall in models may be a response to an unrealistic re-

lationship between the modeled eastern Indian Ocean SST

and northwest Australian rainfall. These hypotheses must

be tested with further experiments.

One limitation of our results linking Indian Ocean

warming to Australian rainfall trends is that the model

trends across northern Australia reach a maximum in the

austral autumn, whereas the observed trends peak in the

summer months. This may be a consequence of the linear

assumption implicit in using a single forcing mechanism

in the ITR experiment. Indeed, the ITR1VR and IPTR1VR

experiment sets, which include both the SST variability

and trends over the Indian Ocean and the Indian and

Pacific oceans, respectively, exhibited their strongest

modeled decrease over north and northeastern Australia

during DJF (figure not shown). The negative trend in

annual rainfall over the east coast of Australia is better

simulated in the IPTR1VR experiment, suggesting that

changes in both the Indian and Pacific oceans may ac-

count for a component of the observed drought.

Previous studies have suggested an influence of the

SAM on Australian rainfall trends (e.g., Meneghini et al.

2006 and Sen Gupta and England 2006). It is likely that

recent trends in the SAM may account for some fraction

of rainfall changes, particularly over the southern regions

of Australia. In addition, the warming of the Tasman Sea

may play an additional role in modulating Australian

rainfall, via alterations in the SAM as suggested by Shi

et al. (2008b). Here, we focused on atmospheric changes

arising from the recent warming over the tropical Indian

Ocean. Further investigation must be carried out to assess

the relative impact of the southern extratropics compared

to the tropical forcing studied here.

Over the Pacific Ocean, anthropogenic forcing by

greenhouse gases may have weakened the Walker circu-

lation, reducing the zonal gradient of SLP between the

western and central tropical Pacific Ocean (Vecchi et al.

2006). However, Vecchi et al. (2006) show that the region

of relative SLP increase reaches west into the Indian

Ocean, with a maximum adjacent to northwestern

Australia where precipitation has increased significantly in

recent decades. A broad increase of SLP across northern

Australia from the Pacific Ocean to the Indian Ocean

would fail to account for the observed east–west gradient

in rainfall trends over Australia observed in recent

decades.

The major modifications to the atmospheric circula-

tion imposed by the observed trend in Indian Ocean SST

suggest that this is an important climatic perturbation

through which changes in the global heat budget are pro-

jected onto regional climate. Given the resulting changes

in tropical atmospheric circulation and the projected

continuation of the Indian Ocean SST trend, studies into

the future evolution of this teleconnection should be

conducted.

Acknowledgments. This research was supported by the

Australian Research Council (ARC). Alex Sen Gupta

provided some assistance with Fig. 10.

REFERENCES

Arakawa, O., and A. Kitoh, 2004: Comparison of local precipitation–

SST relationship between the observation and a reanalysis data-

set. Geophys. Res. Lett., 31, L12206, doi:10.1029/2004GL020283.

Ashok, K., Z. Guan, and T. Yamagata, 2001: Impact of the Indian

Ocean dipole on the relationship between the Indian monsoon

rainfall and ENSO. Geophys. Res. Lett., 28, 4499–4502.

——, ——, and ——, 2003: Influence of the Indian Ocean dipole on

the Australian winter rainfall. Geophys. Res. Lett., 30, 1821,

doi:10.1029/2003GL017926.

Cai, W., P. H. Whetton, and D. J. Karoly, 2003: The response of the

Antarctic Oscillation to increasing and stabilized atmospheric

CO2. J. Climate, 16, 1525–1538.

——, T. Cowan, and A. Sullivan, 2009: Recent unprecedented

skewness towards positive Indian Ocean dipole occurrences

and its impact on Australian rainfall. Geophys. Res. Lett., 36,

L11705, doi:10.1029/2009GL037604.

Collins, W. D., and Coauthors, 2006: The formulation and atmo-

spheric simulation of the Community Atmosphere Model

version 3 (CAM3). J. Climate, 19, 2144–2161.

Drosdowsky, W., and L. E. Chambers, 2001: Near-global sea sur-

face temperature anomalies as predictors of Australian sea-

sonal rainfall. J. Climate, 14, 1677–1687.

Du, Y., and S. Xie, 2008: Role of atmospheric adjustments in the

tropical Indian Ocean warming during the 20th century in

climate models. Geophys. Res. Lett., 35, L08712, doi:10.1029/

2008GL033631.

1 APRIL 2010 L U F F M A N E T A L . 1673

Page 15: Global and Regional Climate Response to Late Twentieth ...web.science.unsw.edu.au/~andrea/papers/Luffman_etal_2010.pdf · Global and Regional Climate Response to Late Twentieth-Century

Efron, B., 1982: The Jackknife, the Bootstrap, and Other Resampling

Plans. Society for Industrial and Applied Mathematics, 92 pp.

England, M. H., C. C. Ummenhofer, and A. Santoso, 2006: Inter-

annual rainfall extremes over southwest Western Australia

linked to Indian Ocean climate variability. J. Climate, 19, 1948–

1969.

Giannini, A., R. Saravanan, and P. Chang, 2003: Oceanic forcing of

Sahel rainfall on interannual to interdecadal time scales. Sci-

ence, 302, 1027–1030.

Gillett, N. P., and D. W. J. Thompson, 2003: Simulation of recent

Southern Hemisphere climate change. Science, 302, 273–275.

Hack, J. J., J. M. Caron, S. G. Yeager, K. W. Oleson, M. M. Holland,

J. E. Truesdale, and P. J. Rasch, 2006: Simulation of the global

hydrological cycle in the CCSM Community Atmosphere Model

Version 3 (CAM3): Mean features. J. Climate, 19, 2199–2221.

Held, I. M., and B. J. Soden, 2006: Robust responses of the hy-

drological cycle to global warming. J. Climate, 19, 5686–5699.

Hoerling, M. P., and A. Kumar, 2003: The perfect ocean for

drought. Science, 299, 691–694.

——, J. W. Hurrell, T. Xu, G. T. Bates, and A. S. Phillips, 2004:

Twentieth century North Atlantic climate change. Part II:

Understanding the effect of Indian Ocean warming. Climate

Dyn., 23, 391–405.

Kiktev, D., D. M. H. Sexton, L. Alexander, and C. K. Folland, 2003:

Comparison of modeled and observed trends in indices of

daily climate extremes. J. Climate, 16, 3560–3571.

Kumar, A., F. Yang, L. Goddard, and S. Schubert, 2004: Differing

trends in the tropical surface temperatures and precipitation

over land and oceans. J. Climate, 17, 653–664.

Lavery, B., G. Joung, and N. Nicholls, 1997: An extended high-

quality historical rainfall dataset for Australia. Aust. Meteor.

Mag., 46, 27–38.

Marshall, G. J., 2003: Trends in the southern annular mode from

observations and reanalyses. J. Climate, 16, 4134–4143.

Meneghini, B., I. Simmonds, and I. N. Smith, 2006: Association

between Australian rainfall and the southern annular mode.

Int. J. Climatol., 27, 109–121.

Nicholls, N., 1989: Sea surface temperature and Australian winter

rainfall. J. Climate, 2, 965–973.

——, 2006: Detecting and attributing Australian climate change: A

review. Aust. Meteor. Mag., 55, 199–211.

——, 2008: Recent trends in the seasonal and temporal behaviour

of the El Nino–Southern Oscillation. Geophys. Res. Lett., 35,

L19703, doi:10.1029/2008GL034499.

Quartly, G. D., E. A. Kyte, M. A. Srokosz, and M. N. Tsimplis, 2007:

An intercomparison of global oceanic precipitation climatol-

ogies. J. Geophys. Res., 112, D10121, doi:10.1029/2006JD007810.

Rayner, N. A., D. E. Parker, E. B. Horton, C. K. Folland,

L. V. Alexander, D. P. Rowell, E. C. Kent, and A. Kaplan,

2003: Global analyses of sea surface temperature, sea ice, and

night marine air temperature since the late nineteenth cen-

tury. J. Geophys. Res., 108, 4407, doi:10.1029/2002JD002670.

Rotstayn, L. D., and Coauthors, 2007: Have Australian rainfall

and cloudiness increased due to the remote effects of Asian

anthropogenic aerosols? J. Geophys. Res., 112, D09202,

doi:10.1029/2006JD007712.

Saji, N. H., and T. Yamagata, 2003: Possible impacts of Indian

Ocean dipole mode events on global climate. Climate Res., 25,151–169.

——, B. N. Goswami, P. N. Vinayachandran, and T. Yamagata,

1999: A dipole mode in the tropical Indian Ocean. Nature, 401,

360–363.

Sen Gupta, A., and M. H. England, 2006: Coupled ocean–

atmosphere–ice response to variations in the southern an-

nular mode. J. Climate, 19, 4457–4486.

Shi, G., W. Cai, T. Cowan, J. Ribbe, L. Rotstayn, and M. Dix,

2008a: Variability and trend of northwest Australia rainfall:

Observations and coupled climate modeling. J. Climate, 21,

2938–2959.

——, J. Ribbe, W. Cai, and T. Cowan, 2008b: An interpretation

of Australian rainfall projections. Geophys. Res. Lett., 35,

L02702, doi:10.1029/2007GL032436.

Smith, I., 2004: An assessment of recent trends in Australian

rainfall. Aust. Meteor. Mag., 53, 163–173.

Taschetto, A. S., and M. H. England, 2008: Estimating ensemble size

requirements of AGCM simulations. Meteor. Atmos. Phys.,

100, 23–36.

——, and ——, 2009: An analysis of late twentieth century trends in

Australian rainfall. Int. J. Climatol., 29, 791–807.

Thompson, D. W. J., and J. M. Wallace, 2000: Annular modes in the

extratropical circulation. Part I: Month-to-month variability.

J. Climate, 13, 1000–1016.

——, S. Lee, and M. P. Baldwin, 2003: Atmospheric processes

governing the Northern Hemisphere annular mode/North

Atlantic Oscillation. The North Atlantic Oscillation: Climatic

Significance and Environmental Impact, Geophys. Monogr.,

Vol. 134, Amer. Geophys. Union, 81–112.

Ummenhofer, C. C., A. Sen Gupta, M. J. Pook, and M. H. England,

2008: Anomalous rainfall over southwest Western Australia

forced by Indian Ocean sea surface temperatures. J. Climate,

21, 5113–5134.

——, ——, A. S. Taschetto, and M. H. England, 2009: Modulation

of Australian precipitation by meridional gradients in east

Indian Ocean sea surface temperature. J. Climate, 22, 5597–

5610.

Vecchi, G. A., B. J. Soden, A. T. Wittenberg, I. M. Held, A. Leetmaa,

and M. J. Harrison, 2006: Weakening of tropical Pacific atmo-

spheric circulation due to anthropogenic forcing. Nature, 441,

73–76.

Washington, R., and A. Preston, 2006: Extreme wet years over

southern Africa: Role of Indian Ocean sea surface tempera-

tures. J. Geophys. Res., 111, D15104, doi:10.1029/2005JD006724.

Watterson, I. G., 2001: Wind-induced rainfall and surface tem-

perature anomalies in the Australian region. J. Climate, 14,

1901–1922.

Yu, L., X. Jin, and R. Weller, 2007: Annual, seasonal, and in-

terannual variability of air–sea heat fluxes in the Indian

Ocean. J. Climate, 20, 3190–3209.

1674 J O U R N A L O F C L I M A T E VOLUME 23