REPORTS
Warming of the Eurasian
Landmass Is Making the Arabian
Sea More Productive
Joaquim I. Goes,1* Prasad G. Thoppil,2. Helga do R Gomes,1
John T. Fasullo3
The recent trend of declining winter and spring snow cover over Eurasia is
causing a land-ocean thermal gradient that is particularly favorable to stronger
southwest (summer) monsoon winds. Since 1997, sea surface winds have been
strengthening over the western Arabian Sea. This escalation in the intensity of
summer monsoon winds, accompanied by enhanced upwelling and an increase
of more than 350% in average summertime phytoplankton biomass along the
coast and over 300% offshore, raises the possibility that the current warming
trend of the Eurasian landmass is making the Arabian Sea more productive.
The Arabian Sea_s seasonally reversing
monsoons drive one of the most energetic
current systems in the world and the greatest
seasonal variability observed in any ocean
basin (1, 2). It is the only ocean basin that
fully reverses its circulation on a semiannual
basis (3, 4), a phenomenon in which the
Indian Ocean, the Eurasian continent, and
the Pacific Ocean play important roles (5). In
summer (June-September), the heating of the
Eurasian landmass results in low pressure
over Asia, while high pressure prevails over
the Indian Ocean. The geostrophically balanced airflow results in a strong topographically steered southwesterly wind and the
formation of a low-level atmospheric feature
called the Findlater Jet (6), which induces a
northeastward flow of the surface current,
causing strong coastal upwelling near the
coasts of Somalia, Yemen, and Oman (7).
In contrast, during the northeast monsoon
(winter, November-February), the cooling of
the Northern Hemispheric landmass results
in high pressure over land and low pressure
over the Indian Ocean, which causes a reversal in the direction of the winds from
southwesterly to northeasterly (7). Because
the reversal of the monsoons has a major
influence on mixed-layer dynamics (8) and
on physical oceanographic processes that
facilitate the input of nutrients to the normally nutrient-impoverished waters of the
Arabian Sea (9, 10), its importance for
phytoplankton growth and biogeochemical
processes is profound (11).
1
Bigelow Laboratory for Ocean Sciences, West Boothbay Harbor, ME 04575, USA. 2Oceanography Department, Naval Postgraduate School, Monterey, CA
93943, USA. 3Program in Atmospheric and Oceanic
Sciences, University of Colorado, Boulder, CO, 80309,
USA.
*To whom correspondence should be addressed.
E-mail: [email protected]
.Present address: Department of Marine Science,
University of Southern Mississippi, Stennis Space
Center, MS 39529, USA.
From 1994 to 1996, the multinational Joint
Global Ocean Flux Study (JGOFS) expeditions
to the Arabian Sea helped unravel several
linkages between physical forcing and carbon
cycling in the northern Arabian Sea, but these
were mostly on seasonal and shorter time scales
(7, 11, 12). Here we present results of rapid
and profound interannual changes being expe-
rienced by the Arabian Sea and, furthermore,
evidence that ascribes these changes to the
warming trend and the declining wintertime
snow cover over the Eurasian landmass.
In 1997, the tropical Indian Ocean experienced a dipole mode (IOD) event: a pattern of
zonal (east-west) variability across the ocean,
with anomalously low sea surface temperatures
(SSTs) off Sumatra, high temperatures in the
western Indian Ocean, and accompanying
wind and precipitation anomalies (13, 14).
This was also the year of one of the strongest
El NiDo events in recent history (15). Although
uncertainty exists as to whether the dipole
structure was triggered remotely by the El
NiDo event in the tropical Pacific or generated
locally (16), SSTs along the entire western and
central parts of the Arabian Sea were warmer
than normal (17–19). Our analysis of a 7-year
record of satellite ocean color data (20) from
1997, encompassing the period of the IOD
event, revealed that concentrations of chlorophyll a in the coastal region of the western
Arabian Sea (47- to 55-E, 5- to 10-N) (fig. S1)
were lower than normal during the summer
upwelling season of 1997 (Fig. 1A). Although
satellite chlorophyll data are not available for
A
B
C
Fig. 1. Annual trends of (A) satellite-derived chlorophyll a data; (B) Reynolds blended (R&S, 1- 1-)
SSTs; and TMI (0.25- 0.25-)–derived SSTs; and (C) wind stress curl values derived from NCEP-NCAR
reanalysis data (open histograms) and TMI-derived wind speed for the region off the coast of Somalia
(5- to 10-N and 47- to 55-E) in the western Arabian Sea. Positive wind stress curl values and lower
SST values indicate upwelling, whereas negative wind stress curl values indicate downwelling.
M, March; J, June; S, September; D, December.
www.sciencemag.org
SCIENCE
VOL 308
22 APRIL 2005
545
REPORTS
the entire summer monsoon period of 1997
because of the premature demise of the
ADEOS-1 satellite (20), the low chlorophyll a
concentrations are explainable given that sea
surface wind stress in May and June, the
primary driver of upwelling during the summer
monsoon in the western Arabian Sea, was
much weaker than normal (21). The timing of
the onset and the intensities of sea surface
winds are both critical to the development of
the Findlater Jet, which in turn is responsible
for coastal divergent upwelling off the Somali
coast and offshore Ekman forced upwelling off
the Omani continental shelf (3, 4). Coincident
with the IOD event of 1997, sea surface winds
(21) picked up only by June, almost a month
later than in a normal year, followed by a
peak in July that was short-lived. The impact
on upwelling of the early collapse of the
monsoon winds in the coastal region is clearly
visible in the higher-than-normal SSTs (22) in
June and July (Fig. 1B), an indicator of
weaker upwelling that year.
In the time series record of chlorophyll a,
however, the most conspicuous observation
was the consistent year-by-year increase in
phytoplankton biomass over the 7-year period
(Fig. 1A). By the summer of 2003, chlorophyll
a concentrations were 9350% higher than those
observed in the summer of 1997. The increase
in chlorophyll a was accompanied by a yearby-year decline in summertime SSTs and cyclonic wind stress curl values (Fig. 1C) (23),
both indicators of a progressive intensification
of upwelling along the coast of Somalia resulting from a progressive strengthening of sea
surface winds over the 7-year period (Fig. 1C).
Upwelling off Somalia is also associated with
the development of the Somali Current gyres,
such as in the Great Whirl, where the vorticity
balance forces an uplift of the thermocline to
the left of the offshore flows (24, 25).
This year-by-year increase in chlorophyll a
concentrations was not confined to the coast
alone but was also observed over a wider area
of the western (52- to 57-E, 5-S to 10-N)
Arabian Sea (Fig. 2A). Outside the region of
coastal upwelling, chlorophyll a concentrations
in the summer of 2003 attained values that
were 9300% higher than those observed in the
summer of 1997. This increase in chlorophyll a
was also accompanied by an intensification of
sea surface winds, in particular of the zonal
(east-to-west) component (Fig. 2A). It is clear
from the offshore observations that the influence of southwest monsoon winds on phytoplankton in the Arabian Sea is not through their
impact on coastal upwelling alone but also via
the ability of zonal winds to laterally advect
newly upwelled nutrient-rich waters to regions
away from the upwelling zone. When colder
waters are advected offshore, they cause a reduction in the latent heat flux to the atmosphere
and an increase in the net heat input into the
oceans. Increased heat flux into the ocean sta-
546
bilizes the water column, causing the mixed
layer to shoal (26). Thus, although sea surface
winds showed a progressive year-by-year increase after 1997, mixed layer depths during the
summer monsoon shallowed progressively over
the 7-year period (Fig. 2B). Increased water
column stability during the summer monsoon
associated with a shallower mixed layer is particularly crucial for retaining phytoplankton
in the euphotic layer, especially when overcast
skies and insufficient light can limit phytoplankton photosynthesis and growth (10, 11).
The summer monsoon winds are a coupled
atmosphere-land-ocean phenomenon, whose
strength is significantly correlated with tropical
SSTs and Eurasian snow cover anomalies on a
year-to-year basis (27, 28). The intensification
of the winds across the Arabian Sea during the
southwest monsoon is largely governed by the
land-sea thermal gradient that develops over
the Arabian Sea in late spring and early
summer. Therefore, the extent of winter and
spring snow cover over the Eurasian landmass
and the latent heat released by the sea during
spring have a major impact on this land-sea
thermal gradient (29). In general, positive
snow anomalies in winter and spring can give
rise to colder ground temperatures in the sub-
sequent summer, because a substantial fraction
of the available solar energy during spring and
early summer goes toward melting the snow
and evaporating water from the wet soil rather
than toward heating the ground (30). Excessive
snowfall in the early part of winter also tends
to reduce solar radiation heating in winter by
increasing the surface albedo, resulting in
persistently colder temperatures over the land
(31). Conversely, reduced snow cover over
Eurasia strengthens the spring and summer
land-sea thermal contrast and is considered to
be responsible for the stronger southwest
monsoon winds and positive rainfall anomalies
over the subcontinent (32, 33).
Analysis of snow cover data (34) for the
period beginning in 1997 revealed a progressive decline of winter and spring snow cover
over the Eurasian landmass (Fig. 2C), which
is consistent with the mid-latitude continental
warming trend reported in the Northern
Hemisphere (33). Since 1979, the decline in
snow cover has been particularly pronounced
over northern Eurasia poleward of 70-N, over
Western Europe, to the northeast of Russia,
over southwest Asia, and over the northern
Indian Himalayan Tibetan Plateau region (fig.
S2). Of greatest relevance to the strength of the
A
B
C
Fig. 2. Annual trends of (A) satellite-derived chlorophyll a data (CHL) and zonal wind stress and (B)
mixed layer depth (MLD) and Reynolds SSTs from the region (52- to 57-E, 5-S to 10-N) in the
western Arabian Sea. (C) Anomalies (departures from monthly means for the period between 1996
and 2002) of Eurasian snow cover (ESC). The trend line shown in bold is a 14-point moving average.
22 APRIL 2005
VOL 308
SCIENCE
www.sciencemag.org
REPORTS
southwest monsoon winds are the latter two
regions, on account of their proximity to the
Arabian Sea. A plot of Eurasian snow cover
extent versus wind stress data (Fig. 3A) and
wind stress versus SST (Fig. 3B) for the period
from late spring to midsummer (May-July)
suggests that the year-by-year decline in winter
and spring snow cover over Eurasia is creating
conditions that are conducive to stronger winds
and lower summertime SSTs across the western Arabian Sea. By regressing SST against
satellite-derived chlorophyll a concentrations over this region for the summer season
(Fig. 3C), we conclude that the escalating
strength of sea surface winds is largely
responsible for the increase in phytoplankton
WIND STRESS (N M-2)
A
0.2
0.15
0.1
0.05
0
0
8
16
24
32
EURASIAN SNOW COVER (x106 KM2)
B
biomass in the western Arabian Sea over the
past 7 years. The fact that disparate satellitederived and observational data sets of SSTs
and winds come together to fit into a physically
consistent scenario gives us a great deal of
confidence in our results.
Our findings raise the intriguing possibility
that the western as well as the central regions of
the Arabian Sea could witness more widespread blooms of phytoplankton if the midlatitudinal continental warming trend and the
decline in winter snow cover over the Northern
Hemisphere continue. Although our findings
have an immediate and important bearing on
regional fisheries, the implications of a more
productive Arabian Sea go far beyond that; for
example, to our planet_s climate. The Arabian
Sea hosts a distinct, basin-wide oxygen minimum zone between 150 and 1000 m (35–37),
whose presence has a substantial impact on
marine elemental cycles, in particular those
linked to the production of climatically relevant
trace gases (37). The changing productivity of
the Arabian Sea could thus have far-reaching
consequences for the oxygen minimum zone,
whose existence is regulated by a balance
between the ventilation of intermediate depths
and oxygen consumption during the oxidation
of organic matter produced in the euphotic
column (36, 37).
31
References and Notes
SST (°C)
29
27
25
0
CHLOROPHYLL (MG M-3)
C
0.05
0.1
0.15
0.2
WIND STRESS (N M-2)
0.9
0.6
0.3
0.0
25
27
29
SST (°C)
31
Fig. 3. Scatter plots of (A) resultant wind stress
and Eurasian snow cover for May to July; (B)
SSTs and resultant wind stress for May to July;
and (C) satellite-derived chlorophyll a data and
SSTs for May to September. Oceanographic
data are for the region (52- to 57-E, 5-S to
10-N), and Eurasian snow cover is from the
Northern Hemisphere EASE grid. Linear leastsquares fits to scatter plots yielded r2 values of
0.84, 0.66, and 0.70 (P G 0.01) for (A), (B), and
(C), respectively.
1. F. A. Schott, J. P. McCreary, Prog. Oceanogr. 51, 1 (2001).
2. S. Clemens, W. D. Prell, D. Murray, G. Shimmield, G.
Weedon, Nature 353, 720 (1991).
3. F. Schott, Prog. Oceanogr. 12, 357 (1983).
4. J. C. Swallow, Deep-Sea Res. 31, 639 (1984).
5. A. S. Bamzai, J. Shukla, J. Clim. 12, 3117 (1999).
6. J. Findlater, Q. J. R. Meteorol. Soc. 95, 362 (1969).
7. C. M. Lee, B. H. Jones, K. H. Brink, A. S. Fischer, DeepSea Res. II 47, 1177 (2000).
8. T. G. Prasad, J. Geophys. Res. 109, C03035 (2004).
9. J. D. Wiggert, R. G. Murtugudde, C. R. McClain, DeepSea Res. II 49, 2319 (2002).
10. K. Banse, D. English, Deep-Sea Res. II 47, 1623 (2000).
11. S. L. Smith, L. A. Codispoti, J. M. Morrison, R. T.
Barber, Deep-Sea Res. II 45, 1905 (1998).
12. M. J. R. Fasham, B. M. Balino, M. C. Bowles, Ambio
(special report) 10, 4 (2001).
13. N. H. Saji, B. N. Goswami, P. N. Vinayachandran, T.
Yamagata, Nature 401, 360 (1999).
14. P. J. Webster, A. M. Moore, J. Loschnigg, R. R. Leben,
Nature 401, 356 (1999).
15. J. E. Overland, N. A. Bond, J. M. Adams, Fish.
Oceanogr. 10, 69 (2001).
16. T. M. Shinoda, A. Alexander, H. H. Hendon, J. Clim. 7,
929 (2004).
17. J. M. Slingo, H. Annamalai, Mon. Weather Rev. 128,
1778 (1999).
18. R. G. Murtugudde, J. P. McCreary, A. J. Busalacchi, J.
Geophys. Res. 105, 3295 (2000).
19. T. G. Prasad, J. L. McClean, J. Geophys. Res. 109,
C02019 (2004).
20. Chlorophyll a data from November 1996 to June 1997
are reprocessed (V4.1), Level 3–binned, ADEOS-1,
Ocean Color Temperature Sensor (OCTS) monthly data
obtainable from the Earth Observation Center, National
Space Development Agency (NASDA) of Japan. For the
period from September 1997 to April 2004, we used
reprocessed Sea-viewing Wide Field-of-view Sensor
(SeaWiFS) (V4.1) Level 3, Global Area Coverage, monthly images from the Distributed Active Archive Center of
the Goddard Space Flight Center, NASA, USA. These
monthly binned products have been corrected for
atmospheric light scattering and for sun angles differing
www.sciencemag.org
SCIENCE
VOL 308
21.
22.
23.
24.
25.
26.
27.
28.
29.
30.
31.
32.
33.
34.
35.
36.
37.
38.
39.
40.
41.
from the nadir. In addition, the influence of clouds has
been substantially reduced. To account for sensor degradation over time, the instrument is calibrated using
internal lamps, solar diffuser observations, and lunar
images, as well as vicarious methods.
The wind data used in this study come from two
sources: the Tropical Rainfall Measurement Mission
Imager (TMI) derived monthly mean (0.25- 0.25grid) and the monthly mean surface wind stresses
(tx, t y) from the National Centers for Environmental
Prediction–National Center for Atmospheric Research (NCEP-NCAR) reanalysis product, calculated
according to E. Kalnay et al. (38).
Monthly mean SSTs were obtained from two sources:
the optimally interpolated (1- 1- grid) Reynolds
reanalysis product, which is a blend of Advanced High
Resolution Radiometer and Comprehensive Ocean
Atmosphere Data Set observations [see (39)]; and
TMI, available from January 1998 onward. These data
are processed by the Remote Sensing Systems
algorithm and mapped to a 0.25- 0.25- grid.
The curl of the wind stress is calculated as ¯t y/¯x –
¯t x/¯y, where t x and t y are the zonal and meridional
wind stress, respectively.
G. L. Hitchcock, E. L. Ley, J. Masters, Deep-Sea Res. II
47, 1605 (2000).
F. A. Schott, M. Dengler, R. Schoenefeldt, Prog.
Oceanogr. 53, 57 (2002).
The mixed layer depth was taken as the depth at
which the temperature declined to 1-C below SST,
using Expandable Bathythermograph (XBT) data
optimally interpolated to a 5- 2- longitudelatitude grid from the Joint Environmental Data
Analysis Center (40).
G. A. Meehl, Science 266, 263 (1994a).
P. J. Webster et al., J. Geophys. Res. 103, 14451 (1998).
H. F. Blanford, Proc. R. Soc. London 37, 3 (1884).
T. P. Barnett, L. Dumenil, U. Schlese, E. Roekler, M.
Latif, J. Atmos. Sci. 46, 661 (1989).
B. Parthashastry, S. Yang, Adv. Atmos. Sci. 12, 143 (1995).
G. A. Meehl, J. Clim. 7, 1033 (1994b).
K. Krishna Kumar, B. Rajagopalan, M. A. Cane, Science
284, 2156 (1999).
Snow cover extents from 1979 onward were obtained from the National Snow and Ice Data Center’s
CD-ROM of Northern Hemisphere EASE-Grid weekly
snow cover and from ice extent data sets of R. L.
Armstrong and M. J. Brodzik (available at http://
nsidc.org/data/snow.html).
R. Sen-Gupta, S. W. A. Naqvi, Deep-Sea Res. 31, 671
(1984).
J. M. Morrison et al., Deep-Sea Res. II 46, 1903 (1999).
S. W. A. Naqvi et al., Nature 408, 346 (2000).
E. Kalnay et al., Bull. Am. Meteorol. Soc. 77, 437 (1996).
R. W. Reynolds, T. M. Smith, J. Clim. 7, 929 (1994).
W. B. White, S. E. Pazan, G. W. Withee, C. Noe, Eos
69, 122 (1988).
This work is supported by grants NNG04GH50G and
NNG04GM64G from NASA to J.I.G. and H.R.G. Funding
from the Maine Space Grants Consortium, USA, and the
Takeda Foundation, Japan, to J.I.G. and H.R.G. and a
postdoctoral fellowship to T.G.P. from the Naval Postgraduate School, Monterey, USA, are gratefully acknowledged. The authors thank the Goddard Earth Sciences
Data and Information Services Center/Distributed Active
Archive Center, NASA, USA, and the Earth Observation
Research Center, NASDA, Japan, for ocean color data
from SeaWiFS and OCTS, respectively; the Climate Diagnostics Center, National Oceanic and Atmospheric Administration, USA, for the reanalysis data products; and
the Remote Sensing Systems, NASA, USA, for the data
from TMI. We are especially grateful to R. Armstrong and
M. Brodzik of the National Snow and Ice Data Center,
USA, for the Northern Hemisphere EASE-Grid snow cover data; and to K. Banse of the School of Oceanography,
University of Washington, USA, for helpful comments.
Supporting Online Material
www.sciencemag.org/cgi/content/full/308/5721/545/
DC1
Materials and Methods
Figs. S1 and S2
21 October 2004; accepted 18 February 2005
10.1126/science.1106610
22 APRIL 2005
547