Content uploaded by Themis Chronis
Author content
All content in this area was uploaded by Themis Chronis on May 01, 2014
Content may be subject to copyright.
Abrupt warming of the Red Sea
D. E. Raitsos,
1
I. Hoteit,
1
P. K. Prihartato,
1
T. Chronis,
2
G. Triantafyllou,
2
and Y. Abualnaja
1
Received 30 April 2011; revised 5 June 2011; accepted 8 June 2011; published 19 July 2011.
[1] Coral reef ecosystems, often referred to as “marine
rainforests,” concentrate the most diverse life in the
oceans. Red Sea reef dwellers are adapted in a very warm
environment, fact that makes them vulnerable to f urther
and rapid warmin g. The detection and understa nding of
abrupt temperature changes is an important task, as
ecosystems have more chances to adapt in a slowly rather
than i n a rapid changing environment. Using satellite
derived sea surface and ground based air tem peratures, it is
shown that the Red Sea is going through an intense
warming initiated in the m id‐90s, with evidence for an
abrupt increase after 1994 (0.7°C difference pre and post the
shift). The air temperature is fo und to be a key para meter
that influences the Red Sea marine temperature. The
comparisons with Northern Hemisphere temperatures
revealed that the observed warming is part of global climate
change trends. The hitherto results also raise additional
questions regarding other broader climatic impacts over
the area.
Citation: Raitsos, D. E., I. Hoteit, P. K. Prihartato,
T. Chronis, G. Triantafyllou, and Y. Abualnaja (2011), Abrupt
warmingoftheRedSea,Geophys. Res. Lett., 38, L14601,
doi:10.1029/2011GL047984.
1. Introduction
[2] The Red Sea holds one of the most diverse marine
ecosystems in the world, although fragile and vulnerable to
oceanic warming [Cantin et al., 2010]. While global
warming is evident across most of the tropical and sub-
tropical seas, the Red Sea warming in particular appears to
exceed the average marine tropical tempera tures [Kleypas
et al., 2008]. It is well documented that the consequences
of the intense warming are apparent across the entire
marine food web, i.e., from primary producers to top‐
predators, which can potentially lead to ecological break-
downs [Beaugran d, 20 04; Richardson and Schoeman,
2004]. For instance, the increased warming has signifi-
cantly slowed down the coral growth in the central Red Sea
[Cantin et al., 2010], while in other cases it has resulted in
the calcification cessation, coral bleaching (zooxanthellae
loss) and mortality. In the past, the Earth’s ecosystem has
repeatedly gone through large climate changes [Alley et al.,
2003]. However, it is the speed of change that controls the
level of response of the ecological communities. In other
words, the ecological systems would have more chances to
adapt in a slowly, rather than in a fast changing environment
[Alley et al., 2003; deYoung et al., 2008]. Thus, particularly
during the last decade there is a growing scientific interest in
the detection and understanding of sudden changes or eco-
system regime shifts.
[
3] The Advanced Very High Resolution Radiometer
(AVHRR) Pathfinder data set has been characterized as the
most extended time series of global sea surface temperature
(SST) currently available [Nykjaer, 2009]. Although it was
not initially intended to be used as a proxy in climate
studies, it has been proved highly valuable for studying
trends and anomalies over long time periods [Marullo et al.,
2007; Nykjaer, 2009]. Despite the importance, there is no
up‐to‐date detailed study assessing the Red Sea thermal
regime through AVHRR. The goal of this study is to report
the satellite derived spatiotemporal changes of the Red Sea
temperatures, and assess whether the alterations observed
during the last two decades originate from regional phe-
nomena, or they are in part driven by global climate change
trends.
2. Methods
2.1. Data Sets
[
4] For this study, three different sources of temperature
data were used, encompassing a period between 1985 and
2007. Here we used the monthly SST means (4 × 4 km
2
)of
the AVHRR Pathfinder V5 data set, which is produced
jointly by the National Oceanic and Atmospheric Admin-
istration (NOAA), and the National Aeronautics and Space
Administration (NASA). To avoid the solar radiation bias
that occurs during the day‐time from surface heating, only
the nighttime overpasses were employed [Nykjaer, 2009;
Raitsos et al., 2006]. The accuracy of the satellite derived
SSTs are documented in numerous studies [Nykjaer, 2009;
Marullo et al., 2007, and references therein]. A recent
detailed ground‐truth validation study in the Mediterranean
Sea reported that the AVHRR Pathfinder data related to an
average error of less than 0.1 K, while although satellites
retrieve the temperature of the skin of the sea surface, the best
fit was found at 4m depth [Marullo et al., 2007]. The authors
further reported that the error appeared to be weekly depen-
dent on season, while it did not drift with time. Both afore-
mentioned features make the Pathfinder SST a dependable
data set for studying global and regional trends and anomalies
[Marullo et al., 2007].
[
5] Monthly means of the in situ air temperature data
were obtained from the King Abdulaziz International Air-
port (KAIA), provided by the Presidency of Meteorology
and Environment (PME) of Saudi Arabia. The Northern
Hemisphere Temperatures (NHT) were acquired from the
1
Red Sea Research Centre, King Abdullah University for Science and
Technology, Thuwal, Saudi Arabia.
2
Hellenic Centre for Marine Research, Institute of Oceanography,
Anavyssos, Greece.
Copyright 2011 by the American Geophysical Union.
0094‐8276/11/2011GL047984
GEOPHYSICAL RESEARCH LETTERS, VOL. 38, L14601, doi:10.1029/2011GL047984, 2011
L14601 1of5
Climatic Research Unit and H adley Centre [Jones et al.,
2008] and employed as anomalies relative to the 1961–
90 reference period [Brohan et al., 2006]. The 1000hPa
level monthly gridded geopotential height (m) from the
National Center for Environmental Prediction‐Department
Of Energy Reanalysis II (NCEP‐DOE) was also used for
the same period [Kalnay et al., 1996; Saha et al., 2006].
2.2. Data Analysis
[
6] A regime shift index (RSI) combined with an automatic
sequential algorithm [Rodionov, 2004] was employed to
confirm the existence and statistical significance of abrupt
changes in the data. The absolute value of RSI represents the
magnitude of the shift(s) while its sign determines the change
in direction of mean between regim es (see Rodionov [2004]
for more information). The data were de‐seasonalised
(standardized anomalies) to reduce temporal autocorrela-
tion. Pearson correlation and cross‐correlation analysis were
used to examine the relationships between the data sets.
Here it has to be mentioned that the correlation coefficient
and its significance level were reduced when performed on
de‐seasonalised data (compared to the original one). How-
ever, the seasonality within the monthly temperature time
series is very strong and thus should be removed prior to the
statistical analysis.
3. Results and Discussion
[7] The monthly SST anomaly time series revealed an
abrupt warming that was initiated in 1994, while it was
stabilized to the new warmer state few years later (Figure 1a).
The warming trend is apparent in every month, implying that
the change is not driven by seasonality. The most pro-
nounced signals in this anomaly plot are the relatively colder
winter of 1992 and the relatively warmer summer of 1995
and 1998. Using the NHT anomalies [Jones et al., 2008] it
has been shown that the years 1995 and 1998 have ranked as
the warmest years of the Northern Hemisphere, while the
temperature shift at the end of the 1990s was the most intense
change in the last 160 years [Raitsos et al., 2010]. While the
satellite derived Red Sea SST data set revealed an intense
and abrupt warming in the mid‐90s, this may be part of a
more widespread temperature shift seen in NHT at the
beginning of the 80s [Brohan et al., 2006; Jones et al., 2008].
In addition, the features revealed from the SST differences
prior (1985–1993) and after (1994–2007) the abrupt change,
clearly underlines that the oceanic warming is evident over
Figure 1. Temporal and spatial patterns of long ‐ term satellite derived sea surface temperatures of the Red Sea for the
period of 1985 to 2007. (a) Monthly time series of SST standardized anomalies (the blue colors represent a relatively colder
period, while the red colors a warmer one), and (b) temperature differences (°C) between the two decades, prior and post the
abrupt SST shift (1985–1993 and 1994–2007 respectively). (Note that the scale represents only positive anomalies as neg-
ative ones were not apparent.)
RAITSOS ET AL.: RED SEA CLIMATE DRIVEN WARMING L14601L14601
2of5
the entire Red Sea (Figure 1b). The average difference
between the two decades registers around +0.62°C and in
some areas exceeds the 1°C. It is the central part of the study
area that portrays the highest deviations from the mean. This
is in accordance with Cantin et al. [2010] reporting that the
level of warming of the central Red Sea exceeds the observed
mean tropical warming, an area that holds a very high
diversity of corals.
[
8] In order to examine whether the observed warming
trends is a regional (Red Sea) characteristic or part of the
current global warming trends, we used the NHT data set
(Figure 2a). The results are presented as the 6 month moving
averages to reduce the excess noise of monthly means. The
NHT and the regional SST anomaly data sets parallel one
another at a significant level, showing that the Red Sea
temperature is influenced by the Northern Hemisphere
trends (on a monthly r = 0.44, p = 0.0001, and annual
scale r = 0.73, p = 0.0001). To further assess the mecha-
nism behind the marine warming in the study area, monthly
anomalies of the in situ air temperatures were plotted against
the regional SST ones (Figure 2b). The marine temperature
anomalies strongly parallel the in situ air temperature one
(r = 0.62, p = 0.001), indicating the climatic influence on
the Red Sea warming. Strong correlation is also docu-
mented at an annual base (r = 0.78, p = 0.0001). (Note that
the correlation between the original monthly SST and air
temperature, prior to seasonality removal, was r = 0.96, p <
0.00001.) Cross‐correlation analysis indicated that the best fit
between SST and air temperatures occur at 1 month lag (for
SST). In other words, the air temperature is found to be a key
parameter that influences the Red Sea marine temperature, as
it changes first and approximately 1 month later, the alteration
is observed in the sea. The effect and the timing of the
atmospheric forcing (air temperature) on the Red Sea shown
herein, is consistent with findings from surrounding regions
such as the Mediterranean Sea [Astraldi et al., 1995; Raitsos
et al., 2010]. Finally, the NHT appeared to be significantly
Figure 2. Moving averages (6 month) of the monthly de‐seasonalised regional (marine and air) and Northern Hemisphere
temperatures (1985–2007). (a) Monthly standardized anomalies of the Red Sea SST against NHT, (b) monthly standardized
anomalies of the Red Sea SST versus in situ air temperature. The solid gray line stands for the SST, while the black one for
the NHT and air temperature anomalies respectively.
Figure 3. Red Sea annual time series of SST and minimum
air tem perature values since 1985, along with their seasonal
cycles. (a) Annual SST (°C) against minimum air temperature
values (°C). The gray solid line represents the statistical sig-
nificant regime shifts using Rodionov’s automatic sequential
algorithm. (b and c) Seasonal cycles of SST and minimum air
temperature, along with the averaged difference ( °C) before
and after the shift (1985–1993 and 1994–2007 respectively).
RAITSOS ET AL.: RED SEA CLIMATE DRIVEN WARMING L14601L14601
3of5
correlated with the in situ annual and monthly air temperature
anomalies (r = 0.56, p = 0.003 and r = 0.3, p = 0.001,
respectively) indicating that the anomaly signals and trends
evident in regional air temperatures, which in sequence
influence the Red Sea SST are related to the climate
anomalies of the Northern Hemisphere.
[
9] The analysis of the long‐term (1985–2007) annual
AVHRR SST revealed that not only the Red Sea is warming
rapidly, but there is evidence of a temperature abrupt shift
around 1994 (Figure 3a). It can be observed that before 1994
the annual SST mean remains below the overall mean,
whereas after the abrupt shift the opposite pattern is
observed. During the first decade (1985–1993) the annual
SST mean was 27.4°C, whereas during 1994–2007 the
mean was 28.1°C (0.7°C difference). This result is con-
firmed (using Hadley SST data), as it is reported that the
Red Sea has changed by 0.74°C between 1982 and 2006
[Belkin, 2009]. The coldest registered year was 1992 after
which a stepwise increase is evident (Figure 3a). Further-
more, the abrupt SST shift is related to a statistically sig-
nificant change as the latter is assessed by the Rodionov’s
regime shift detection algorithm (year = 1994, RSI = 1.02,
p < 0.0001). It is noteworthy that after the aforementioned
stepwise increase, the SST has remained warmer without
returning to its initial state. Additional analysis on in situ air
temperature data revealed that the most extensive change
was evid ent in terms of the minimu m values. Fig ure 3b
shows that the lower minimum air temperatures have
increased rapidly after 1994 (RSI = 0.46, p = 0.002). Sea-
sonal cycle analysis of both temperatures showed that the
most prominent alterations occurred during the summer
months (July and August), along with smaller changes dur-
ing the winter months with a peak in February (Figures 3b
and 3c). Similar results are obtained if analysis is per-
formed in maximum as well as in total values, however the
changes are more pronounced in the minimum air tempera-
ture values. August and February are also the months that
have changed the most in the eastern Mediterranean Sea
[Raitsos et al., 2010]. The winter months and particularly
February are very critical for primary production (for both
the Red and Mediterranean Seas) as they are related to the
phytoplankton blooms that are primarily formed due to the
intense water column mixing. Increasing SSTs lead to more
stratified conditions associated with reduced vertical mixing
and thus nutrient availability in the upper oceanic layer. Such
a situation can be detrimental to phytoplankton, which forms
the base of the marine food chain [e.g., Raitsos et al., 2011;
Richardson and Schoeman, 2004].
[
10] The majority of the hitherto findings concern SST
spatiotemporal trends, and how these are related to the
broader changes seen in the Northern Hemisphere. To extend
our working hypothesis, we investigate the involved proxy
variable (i.e., SST) as a potential climate forcing regulator for
the Red Sea area. Although not a direct recipient, the Red Sea
bears few characteristics of the summer monsoonal regime
(southwestern phase) dominant over the Arabian Sea and the
Indian Ocean [Patzert, 1974]. These pertain to the north-
eastward expansion of the Arabian peninsula thermal low,
which further interacts with the low level circulation known
as the Somali Jet [Krishnamurti et al., 1 976]. Figure 4a
illustrates the summer (June to ‐September– monsoonal
season) 1000hPa level geopotential height composite for
the period 1985–2007, where the dominant low pressure
system over the Arabian peninsula extends over the Red
Sea. At the same time and despite the fact that precipitation
over the area is very low (2–3 inches per year [Wang , 2006])
other feedback mechanisms may strongly implicate SST as a
climate regulating factor. The herein documented SST
warming trends are expected to favor additional evaporation,
hence the deepening of the regional low pressures over the
area is likely. In turn, Figure 4b illustrates the respective
numerical geopotential height difference prior to and after the
herein documented SST abrupt change (1985–1993 period
minus 1994–2007). It becomes evident that the period 1985–
1993 is followed by an overall lowering of the pressure gra-
dients during 1994–2007 (i.e., positive geopotential height
differences, 1000hPa, in meters) over the Red Sea, fact that
supports the previous hypothesis. Additional implications
arise from the aforementioned pressure re‐distribution and
these pertain to the zonal wind and cloudiness modulation
over the study area. In fact, additional data analysis (not
shown) depicts the intensification of the Somali Jet along
with increased Outgoing Longwave Radiation values for the
period 1994–2007.
4. Conclusion
[11] Regardless of its origin, natural and/or anthropogenic
[Alley et al., 2003], it is widely accepted that oceanic
warming is evident worldwide. The satellite retrieved SST
revealed that the Red Sea is going through an intense
warming that was initiated in the mid‐90s, with evidence of
an abrupt increase after 1994 (a result also shown from the
air temperature data sets). Only during the last decade, the
SST has increased by 0.7°C. Spatiotemporal analysis
showed that the thermal change is apparent in the whole Red
Sea and in the entire year. Comparison with temperatures of
the Northern Hemisphere suggested that the Red Sea
warming is not a local phenomenon but a part of a wide-
spread warming trend that is observed worldwide. In this
view, the marine temperatures appeared to follow the air
temperatures with 1 month delay, indicating the approxi-
mate time needed to see the changes in the sea. It is evident
that the broader climate including atmospheric forcing and
intensity of monsoons have been altered during the period of
Figure 4. Numerical barometric pressure gridded at the
1000hPa level. (a) Composite of June–September surface
geopote ntial height (m) f rom 1985 to 2007, and (b) differ-
ence prior an d after th e SST abr upt change (1985–1993
minus 1994–2007) of June–September surface geopotential
height.
RAITSOS ET AL.: RED SEA CLIMATE DRIVEN WARMING L14601L14601
4of5
SST abrupt shift. This work brings a step closer toward
reporting and understanding a temperature abrupt change
seen in the Red Sea. However, the question as which factor
has triggered this sudden alteration remains unanswered.
Alley et al. [2003] mentioned that even a small and slow
forcing can trigger an abrupt change. The description of the
potential ocean temperature‐lower atmosphere feedback
would need further and detailed investigation.
[
12] An abrupt climate change can potentially occur when
the climate system is forced to overpass a particular
threshold, while their economic and ecological impact could
be significant [Alley et al., 2003]. It is predicted that the Red
Sea temperature will increase as climate change continues
[Cantin et al., 2010], a prediction that will further alter the
regional ecosystem. The level of the Red Sea temperature
change is one of the highest seeing, a result that is confirmed
by other studies. Oceanic warming may have a direct or
indirect impact on marine entities and ecosystems, thus,
there is a need to assess further available past biological data
(e.g., coral, fisheries, plankton) for potential responses to the
new thermal state, and to closely monitor the relatively
unexplored and fragile Red Sea ecosystem.
[
13] Acknowledgments. We thank the Presidency of Meteorology
and Environment (PME) of Saudi Arabia, for providing the air tempera-
ture data. We also thank Igor M. Belkin and one anonymous reviewer for
their constructive comments. This research was supported by the King
Abdullah University for Science and Technol ogy (KAUST), Kingdom
of Saudi Arabia.
[14] The Editor thanks two anonymous reviewers for their assistance in
evaluating this paper.
References
Alley, R. B., et al. (2003), Abrupt climate change, Science, 299, 2005–2010,
doi:10.1126/science.1081056.
Astraldi, M., C. N. Bianchi, G. P. Gasparini, and C. Morri (1995), Climatic
fluctuations, current variability and marine species distribution: A case
study in the Ligurian Sea (northwest Mediterranean), Oceanol. Acta,
18, 139–149.
Beaugrand, G. (2004), The North Sea regime shift: Evidence, causes, me-
chanisms and consequences, Prog. Oceanogr., 60, 245– 262,
doi:10.1016/j.pocean.2004.02.018.
Belkin, I. M. (2009), Rapid warming of large marine ecosystems, Prog.
Oceanogr., 81, 207–213, doi:10.1016/j.pocean.2009.04.011.
Brohan, P., J. J. Kennedy, I. Harris, S. F. B. Tett, and P. D. Jones (2006),
Uncertainty estimates in regional and global observed temperature
changes: A new dataset from 1850, J. Geophys. Res., 111, D12106,
doi:10.1029/2005JD006548.
Cantin, N. E., A. L. Cohen, K. B. Karnauskas, A. M. Tarrant, and D. C.
McCorkle ( 2010), Ocean warming slows coral growth in the central
Red Sea, Science, 329, doi:10.1126/science.1190182.
deYoung, B., M. Barange, G. Beaugrand, R. Harris, R. I. Perry, M. Scheffer,
and F. Werner (2008), Regime shifts in marine ecosystems: Detection, pre-
diction and management, Trends Ecol. Evol., 23, 402–409, doi:10.1016/j.
tree.2008.03.008.
Jones, P. D., D. E. Parker, T. J. Osborn, and K. R. Briffa (2008), Global and
hemispheric te mpera ture a nomalies —Land and marine instrumental
records, in Trends: A Compendium of Data on Global Ch ange, Carbon
Dioxide Inf. Anal. Cent., Oak Ridge Natl. Lab., U.S. Dep. of Energy,
Oak Ridge, Tenn.
Kalnay, E., et al. (1996), The NCEP/NCAR 40‐year re‐analysis project,
Bull. Am. Meteorol. Soc., 77, 437–471, doi:10.1175/1520-0477(1996)
077<0437:TNYRP>2.0.CO;2.
Kleypas, J. A., G. Danabasoglu, and J. M. Lough (2008), Potential role of
the ocean the rmostat in determining regional differences in coral re ef
bleaching events, Geophys. Res. Lett., 35, L03613, doi:10.1029/
2007GL032257.
Krishnamurti, T. N., J. Molinari, and H. L. Pan (1976), Numerical simula-
tion of the Somali Jet, J. Atmos. Sci., 33, 2350–2362, doi:10.1175/1520-
0469(1976)033<2350:NSOTSJ>2.0.CO;2.
Marullo, S., B. B. Nardelli, M. Guarracino, and R. Santoleri (2007),
Observing the Mediterranean Sea from space: 21 years of Pathfinder‐
AVHRR sea surface temperatures (1985 to 2005): Re‐
analysis and val-
idation, Ocean Sci., 3, 299–310, doi:10.5194/os-3-299-2007.
Nykjaer, L. (2009), Mediterranean Sea surface warming 1985–2006, Clim.
Res., 39,11–17, doi:10.3354/cr00794.
Patzert, W. C. (1974), Wind‐induced reversal in Red Sea circulation, Deep
Sea Res. Oceanogr. Abstr., 21,109–121, do i:10.1 016/001 1-74 71(7 4)
90068-0.
Raitsos, D. E., S. J. Lavender, Y. Pradhan, T. Tyrrell, P. C. Reid, and
M. Edwards (2006), Coccolithophore bloom size variation in response
to the regional environment of the subarctic North Atlantic, Limnol.
Oceanogr., 51, 2122–2130, doi:10.4319/lo.2006.51.5.2122.
Raitsos, D. E., G. Beaugrand, D. Georgopoulos, A. Zenetos, M. A. Pancucci‐
Papadopoulou, A. Theocharis, and E. Papathanassiou (2010), G lobal
climate change amplifies the entry of tropical species into the eastern
Mediterranea n Sea, Limnol. Oceanogr., 55,1478–1484, doi:10.4319 /
lo.2010.55.4.1478.
Raitsos, D. E., S. J. Lavender, C. D. Maravelias, J. Haralabous,
A. McQuatters‐Gollop, M. Edwards, and P. C. Reid (2011), Macro-
scale factors affecting diatom abundance: A synergistic use of Continuous
Plankton Recorder and satellite remote sensing data, Int. J. Remote Sens.,
32, 2081–2094, doi:10.1080/01431161003645832.
Richardson, A. J., and D. S. Schoeman (2004), Climate impact on plankton
ecosystems in the northeast Atlantic, Science, 305, 1609– 1612,
doi:10.1126/science.1100958.
Rodionov, S. N. (2004), A sequential algorithm for testing climate regime
shifts, Geophys. Res. Lett., 31, L09204, doi:10.1029/2004GL019448.
Saha, S., et al. (2006), The NCEP Climate Forecast System, J. Clim., 19,
3483–3517, doi:10.1175/JCLI3812.1.
Wang, B. (2006), The Asian Monsoon, 844 pp., Springer, Chichester, U. K.
Y. Abualnaja, I. Hoteit, P. K. Prihartato, and D. E. Raitsos, Red Sea
Research Centre , King Abdullah University for Science and Technology,
Thuwal 23955‐6900, Saudi Arabia. (yasser.abualnaja@kaust.edu.sa;
ibrahim.hoteit@ka ust.edu.sa; perdana.kar im@kaust.edu.s a; dionysios.
raitsos@kaust.edu.sa)
T. Chronis and G. Triantafyllou, Hellenic Centre for Marine Research,
Institute of Oceanography, GR ‐190 13 Anavyssos, Greece. (themis@ath.
hcmr.gr; gt@ath.hcmr.gr)
RAITSOS ET AL.: RED SEA CLIMATE DRIVEN WARMING L14601L14601
5of5