Tropical Response to the Atlantic Equatorial Mode: AGCM Multimodel Approach

Clim Dyn DOI 10.1007/s00382-009-0624-6

Tropical response to the Atlantic Equatorial mode: AGCM multimodel approach

T. Losada Æ B. Rodrı´guez-Fonseca Æ I. Polo Æ S. Janicot Æ S. Gervois Æ F. Chauvin Æ P. Ruti

Received: 23 December 2008 / Accepted: 25 June 2009 Ó Springer-Verlag 2009

Abstract On the frame of the AMMA-EU project, sen- the La Ninã conditions in the central Pacific. Although sitivity experiments for an Atlantic Equatorial mode ocean–atmosphere coupled experiments are required to test (AEM) which origin, development and damping resembles the latter hypothesis, the present studies shows how the the observed one during the last decades of the 20th cen- AEM is able to influence the rest of the tropics, a result tury, has been analysed in order to investigate the influence with important implications on ENSO seasonal on the anomalous summer West African rainfall. Recent predictability. studies raise the matter of the AEM influence on the next Pacific ENSO episodes and also on the Indian Monsoon. This paper evaluates the response of four different atmo- 1 Introduction spheric global circulation models, using the above-men- tioned AEM sensitivity experiments, to study the tropical The question of how the tropical oceans interact with each forcing associated with the Atlantic Ninõ mode. The results other, and what are the associated dynamical mechanisms show a remote signal in both the Pacific and Indian basins. explaining these teleconnections is, nowadays, a debated For a warm phase of the AEM the associated southward topic for the scientific community with a large amount of location of the ITCZ, with rising motions over the Equa- recent references about it (Chang et al. 2006; Wang 2006; torial Atlantic, leads to a global subsidence over the rest of Kug and Kang 2006; Kucharski et al. 2007, 2008, 2009; the tropics, weakening the Asian Monsoon and favouring Keenlyside and Latif 2007; Polo et al. 2008; Jansen et al. 2009). In particular, recent observational and model studies This paper is a contribution to the special issue on West African have emphasized the potential teleconnections between Climate, consisting of papers from the African Multidisciplinary the tropical Atlantic and the adjacent Tropical Indian Monsoon Analysis (AMMA) and West African Monsoon Modeling and Evaluation (WAMME) projects, and coordinated by Y. Xue and and Pacific Oceans (Wang 2006; Kucharski et al. 2007; P. M. Ruti. Keenlyside and Latif 2007; Kucharski et al. 2008, 2009; Polo et al. 2008). & T. Losada ( ) Á B. Rodrı´guez-Fonseca Á I. Polo It is known how the atmospheric response to a heating in Universidad Complutense de Madrid, Madrid, Spain e-mail: tldoval@fis.ucm.es the tropical Atlantic shows a Gill-type response (Vizy and Cook 2001; Kucharski et al. 2008) that leads to an increase S. Janicot Á S. Gervois in precipitation near the Gulf of Guinea (GG) and a LOCEAN/IPSL, CNRS, Universite´ Pierre et Marie Curie, decrease to the east. This Matsuno-Gill response to the Paris, France Equatorial mode, would produce the propagation of a F. Chauvin Kelvin wave pattern trapped in the tropics as a response to GAME/CNRM, Me´teó-France/CNRS, Toulouse, France the tropical heating, which would influence the Indian ocean basin (Kucharski et al. 2007, 2008, 2009). The works P. Ruti Progetto Speciale Clima Globale, Ente Nazionale per le of Kucharski and co-authors have described how, through NuoveTecnologie, l’Energia e l’Ambiente, Rome, Italy this mechanisms, the Equatorial mode could be modifiying 123 T. Losada et al.: Tropical response to the Atlantic Equatorial mode the ENSO–Indian Monsoon (IM) relation, according to indicated above, the ENSO–IM relationship also suffered a which a drier than normal monsoon season usually pre- weakening from the 70s which could be linked to the cedes peak El Ninõ conditions, and viceversa during la Atlantic variability (Kucharski et al. 2008). Ninã phase. In this way, this modification of the ENSO–IM The main purpose of this work is to elucidate how the could be realistically simulated during the period of 1950– interaction between the AEM and the West African con- 1999 only if changes in the Tropical Atlantic SSTs in vection influences the atmospheric circulation over the rest boreal summer were properly accounted for. of the tropical basins. To achieve this goal, we evaluate the How the tropical Atlantic (TA) variability affects the response of four different atmospheric general circulation tropical Pacific is a task that has been widely overlooked by models (AGCMs) that are running in the frame of the the scientific community. The indexes related to the AMMA-EU project; to the tropical forcing associated with Atlantic and Pacific Ninõs do not contemporaneously the Atlantic Ninõ mode. Thus, the analysis of the multi- correlate each other, although Wang (2006) has reported model sensitivity experiment will provide new insights on the existence of an Inter-basin SST gradient which induces the atmospheric global teleconnections associated with an a positive feedback in between the tropical oceanic basins. isolated Atlantic Ninõ pattern. ENSO influence on the Equatorial Atlantic is less clear, The AEM has been calculated taking into account its and it has been recently suggested that this is due to ENSO seasonal evolution and, maximizing its covariability with competing dynamical and thermodynamical influences the West African precipitation. The reason of the latter is there (Chang et al. 2006). Observations show that the TA that in the tropics, where horizontal differences in tem- can influence the tropical Pacific Ocean via the inter-basin perature and density are small, vertical motion, and thus SST gradient variability that is associated with the Atlantic convection, is the main process by which tropical heating is Walker circulation. Rainfall responds to the inter-basin balanced (Hagos and Cook 2005). SST gradient, regardless of which ocean is anomalously This study is divided into four sections. In ‘‘Sect. 2’’ we warm or cold (Wang et al. 2009). In fact, a feedback from describe the experiments and models used, as well as the the Atlantic on ENSO appears to exist, and slightly observational data. The results of the study are presented in improves the retrospective forecast skill of the conceptual ‘‘Sect. 3’’. The discussion and conclusions are displayed in model (Jansen et al. 2009). ‘‘Sect. 4’’. Several recent papers have point to an statistical relationship between the Atlantic and Pacific Ninõs from the 70s, in which the AEM would lead the Pacific ENSO 2 Methodology and data several months in advance (Jury et al. 2002; Melice and Servain 2003; Latif and Keenlyside 2008; Rodriguez- The models and simulations used in this work are ARPE- Fonseca et al. 2008; Polo et al. 2008), pointing to the GE (De´que´ et al. 1994), ECHAM-4 (Roeckner et al. 1996), possibility of increasing ENSO prediction due to the occur- LMDZ (Hourdin et al. 2006) and UCLA (Richter et al. rence of a feedback from the Atlantic. However, physical 2008). The main characteristics of their configuration have explanation about the dynamical mechanisms that could been described in Losada et al. (2009). explain the statistical link that the Equatorial mode, and its The methodology followed in this paper consists in the associated atmospheric response, has on the adjacent Pacific realization of different ensemble simulations, each com- basin during recent decades, has not been account yet. posed by ten members. The present study focuses on the tropical atmospheric The control simulation was run with climatological response to an AEM calculated for the 1979–2005 period, SSTs (Smith and Reynolds 2004) averaged from 1979 to after the Pacific ‘‘climate shift’’. Climate shift refers to the 2005. The EMP simulation was run with a positive SST alteration during 1976–1977 in the central North Pacific anomaly in the tropical Atlantic added to the climatology SST (Nitta and Yamada 1989; Trenberth 1990; Graham (Fig. 1). 1994; Miller et al. 1994). After this period, the propagation The calculation of the Equatorial mode pattern added to characteristics of ENSO changed (Wang 1995; An and the climatology in the sensitivity experiments starts with Wang 2000; Fedorov and Philander 2000; Wang and An the computation of the extended multiple covariance 2002; Latif and Keenlyside 2008) and also the onset phase, analysis (EMCA; Polo et al. 2008) between Atlantic SST being characterized by the appearance of SST anomalies in and West Africa precipitation with observational data. the Central Pacific that propagate eastward (Terray and Then different years are selected to define a composite of Dominiak 2005); in contrast with the canonical evolution years in which the covariance between the tropical Atlantic of an ENSO event, in which the SST anomalies first appear SST and the rainfall over West Africa is maximized. The in the South American coast and then propagate to the west years selected are 1984, 1987, 1988, 1989, 1998, 1999 for (Rasmusson and Carpenter 1982). Besides, and it has been positive SST anomalies; and 1982, 1983, 1992, 1997 for 123 T. Losada et al.: Tropical response to the Atlantic Equatorial mode

Fig. 1 MJ (left) and JAS (right) SST anomalies (°C) used as boundary conditions

negative SST anomalies. Details of the method of defini- to assess the significance of the response, we apply a test t tion of the anomalous SST pattern used are found in Los- of equal mean (von Storch and Zwiers 1999)at95%of ada et al. (2009). confidence level. The global tropical composite (Fig. 2, contours) for these years indicates the concomitant coexistence of a Pacific La Ninã, according with recent statitistical studies 3 Results: tropical teleconnections related that point to an statistical relationship between the Atlantic to the Equatorial mode and Pacific Ninõs from the 70s (Jury et al. 2002; Melice and Servain 2003; Latif and Keenlyside 2008; Rodriguez- Recent studies have shown how the atmospheric response Fonseca et al. 2008; Polo et al. 2008). to a heating in the Equatorial Atlantic is a Gill-Matsuno The anomalous response to the Equatorial mode will be type response with a Kelvin wave propagating to the Indian computed by subtracting the control to the EMP. In order region and an equatorial trapped Rossby wave propagating westward to Central America and eastern Pacific. These studies point, on the one hand, to a link between the African and Indian Monsoons through variations in the equatorial circulation (Kucharski et al. 2008, 2009); and, on the other hand, to a link between the Equatorial mode and the Pacific El Ninõ (Polo et al. 2008). This study revises the influence of the Equatorial mode on the adjacent tropical basins, trying to give a dynamical explanation to the teleconnections found. Figure 3 show the precipitation anomalies given by the different models over the Equatorial Africa in May–June (MJ) and July–August–September (JAS), with positive values over the Gulf of Guinea and West African coast, and negative values to the east. This pattern, which is weaker for ECHAM-4 and LMDZ, is present for all the models and months except for ARPEGE in MJ, and it is stronger after the onset (JAS, Fig. 3). Vizy and Cook (2001) and Kucharski et al. (2009) found a similar pattern in their Fig. 2 Composite maps for the precipitation anomalies (mm/day, experiments that can be explained as a Gill response to an shaded) and SST (°C, contours, dashed lines denote negative values, contour interval is 0.5°C) observed for the years used in the equatorial heating (Gill 1980), which produces an anoma- generation of the boundary conditions (see ‘‘Sect. 2’’) lous zonal flow from the east into the heating source 123 T. Losada et al.: Tropical response to the Atlantic Equatorial mode

(a Kelvin wave response), with anomalous divergence and (Fig. 3), is flanked with two positive anomalies to the east a decrease in the precipitation to the east of the heating. and west. In MJ, the positive centres are located over the The anomalous precipitation pattern in the Indian basin West Indian Ocean and Central America. From July, the is more or less significant depending of the model and negative lobe has already covered the Amazonian and month (Fig. 3), but a decrease of precipitation over west Caribbean region, and the positive centres have been dis- India that is significant from July is present in all the placed to the central Pacific (around 160°W) and the models. This weakness of the Indian monsoon under con- eastern Indian basin. The significance of the anomalies is ditions of warm SST anomalies in the Equatorial Atlantic is larger for JAS than for previous months. consistent with the statistical relation found by Kucharski The possibility of an influence of the east Atlantic- et al. (2008). In order to characterize the response in West African sector onto the Caribbean-South American our models, we look at the 200 hPa streamfunction area, through changes in the vertical motions has already and velocity potential anomalies (Figs. 4, 5), as indicative been reported by Hagos and Cook (2005), during boreal of the upper rotational and divergence circulation, winter (January). Our results show how, for May–Sep- respectively. tember season, the anomalous upper divergence over the The 200 hPa streamfunction (Fig. 4) response to the GG region, is stronger before the monsoon onset and Equatorial mode has the structure of a typical Gill-Matsuno leads to anomalous upper convergence over the Carib- response to equatorial heating, which is a quadrupole with bean, weakening the mean Walker circulation over the a pair of symmetric anticyclones at both sides of the Atlantic in the months previous to the northward shift of equator and to the west of the maximum heating, and a pair the ITCZ. After the onset, the maximum anomalous upper of cyclones to the east of the source of heating (Jin and troposphere divergence is located in the west of the Hoskins 1995). This response is baroclinic in the tropical Atlantic and the divergence anomalies cover the Carib- Atlantic and barotropic the extratropics (not shown). The bean and Central America regions; the upward branch of centres of the anomalies move to the west from May to the mean Walker circulation, that is located in this area, is September, following the evolution of the SST anomalies, being reinforced from July. At this stage, positive velocity and the amplitude and significance of the anomalies get potential anomalies become significant in the central stronger in time. Pacific for all the models except the ARPEGE (the sig- The westward migration of the anomalies over the nificance in JAS is clearer for ECHAM-4), and they get Equatorial Atlantic from May to September also appears in stronger in August and September. This anomalous pat- the 200-hPa velocity potential (VP) response (Fig. 5), also tern depicts a reinforcement of the Walker circulation cell following the evolution of the tropical convection (Fig. 3) of the Amazon-Central Pacific, in which both the associated with the surface westward Equatorial Atlantic ascending and subsiding branches of the cell are stronger warming (Fig. 1). All the models show a negative VP than in the climatological state, and the descending anomaly located over the equator and around 0° in MJ branch has spread a bit to the west. (except ECHAM-4, that presents the anomaly centred in To better illustrate the mechanism by which the ano- 20°W), that spreads to the west in time. This anomaly, malous convection associated with the AEM surface heating which coincides with the maximum anomalous rainfall is able to alter the Walker circulation, increasing the

Fig. 3 JAS global precipitation anomalies (mm day-1) for a ARPEGE, b ECHAM-4, c LMDZ and d UCLA EMP simulation. Contours delimites regions denote those areas where the anomalies exceed the 95% signiﬁcance level

123 T. Losada et al.: Tropical response to the Atlantic Equatorial mode

Fig. 4 Same as Fig. 3 but for 200 hPa streamfunction (106 m2 s-2)

Fig. 5 Same as Fig. 3 but for 200 hPa velocity potential (106 m2s-2)

subsidence over the Indian and Pacific basins, Fig. 6 depicts anomalous subsiding motions over the central Pacific, this anomalous zonal circulation given by the four analyzed around the date line (significant for all the models except models (longitude by height anomalous vertical velocity- ECHAM-4). This feature closely resembles the one obtained shaded-and zonal wind-contours- for JAS, averaged between by Jin and Hoskins (1995) for a resting stratified atmosphere 4°S and 4°N). All the simulations show strong anomalous forced by tropical heating, and it is consistent with the Gill upward motion over the Atlantic and Amazon region, (1980) response. In this way, the response to a warming in the accompanied with strong anomalous subsidence at both Equatorial Atlantic would produce a westward displacement sides of the maximum convection: one over East Africa of the maximums upward motions over the central Pacific. (around 40°E), which explains the dryness over that region; This displacement could explain the presence of anomalous and the other one from 50°W to the west, over the eastern upward motions in the eastern Pacific (around *120°W) for basin of the Pacific ocean. The latter is maybe a local most of the models. response to the fact that, in the Amazon region, the maximum Also, all the models show a tendency of a reinforced convection has been displaced eastward, towards the west upward branch of the Walker cell over the Indian Ocean, part of the Atlantic basin (the region of anomalous warming), UCLA model response is noisy, but the rest of the producing anomalous subsidence around 80°W. Also, all the models show a pretty clear signal in this way, although models show upper level divergence over the region of the ascendant motions are confined to the middle and heating and upper level convergence, very clear in the lower troposphere. These anomalous upward motions anomalous zonal wind response (Fig. 6, contours), and could be part of an anomalous local Hadley-type

123 T. Losada et al.: Tropical response to the Atlantic Equatorial mode

Fig. 6 JAS 4°S–4°N averaged vertical velocity (shaded; 100 Pa s-1) and zonal wind anomalies (m s-1, contour interval is 1.5 m s-1). Only 95% signiﬁcant areas are plotted

response to the upper level convergence over the Indian Wang (1995) and Terray and Dominiak (2005) among peninsula. others. In this way, the atmospheric response to a warming on The presence of warm SST anomalies in the Equatorial the tropical Atlantic would lead to an enhancement of the Atlantic would then be able to produce a response in the upward branch of the Walker circulation in the Amazon circulation of the Equatorial Pacific atmosphere that would region, anomalous upper divergence in that area, together be consistent with the forcing of a surface cooling in the with anomalous convergence and subsiding motions over central Equatorial Pacific ocean. The consistency of the the central Equatorial Pacific and over the Indian penin- four models response gives confidence to the results. sula. The anomalous subsidence over India would induce Whether or not the Pacific atmospheric anomalies can anomalous upward motions over the Equatorial Indian actually have an influence in the Pacific ocean SST is a Ocean that, in turn, produces a local reinforcement of the topic that cannot be assess by the kind of experiments upward branch of the Walker circulation over the region. performed in this work. Additional experiments, to be These anomalous subsiding motions in the central performed with ocean-atmosphere coupled models, should Pacific and anomalous upward motions over the Indian be designed in order to test the present hypotheses. Ocean would enhance the zonal pressure gradient between this two points, leading to an increase in the trade winds between them, surface divergence and a cooling of the 4 Conclusions SSTs over the central Equatorial Pacific. This situation would favour the development of an ENSO cold event In this study, we use four AGCM (ARPEGE, ECHAM-4, starting on the central Pacific, which would be in agree- LMDZ, UCLA) in order to analyze the impact of the AEM ment with the description of recent ENSO events made by on the rest of the tropical basins during the last decades of the 20th century.

123 T. Losada et al.: Tropical response to the Atlantic Equatorial mode

As a response to the equatorial heating of the Atlantic Pacific is related to a decrease of rainfall in the GG during basin, the atmosphere shows a Gill-Matsuno type quadru- the monsoon season (Janicot and Rodriguez-Fonseca polar circulation with an equatorial Rossby wave propa- 2007). In fact, the composite of the observed precipitation gating to the west and a Kelvin wave to the east. This anomalies for the years used in the computation of the response is reflected in the west-east dipole of anomalous Equatorial mode used in the experiments (Fig. 2), shows a precipitation found in the Equatorial Africa. The anoma- precipitation dipole associated with SST anomalies in the lous upward motions over the Atlantic Ocean are com- Equatorial Atlantic during MJ, with an decrease of rainfall pensated with anomalous subsiding motions over the over the Sahel, but it disappear for JAS, in which the central Pacific, leading to a westward displacement of the reported increase of precipitation over the Sahelian region maximum location of the downward branch of the Pacific associated with a cooling of the Pacific is observed. We Walker cell. Also, anomalous upper level convergence believe that the added influences of the Atlantic and Pacific appears over the Indian peninsula, causing the air to des- anomalies in the WAM are the responsible of this preci- cend and producing a inhibition of the IM convection and a pitation pattern evolution and further analysis is being done decrease of rainfall over India, supporting the findings of in this topic. We have raised here the idea of a possible Kucharski et al. (2007, 2008, 2009). The anomalous sub- influence of the Atlantic variability on the tropical Pacific sidence over India would induce an anomalous local atmospheric circulation. Nevertheless, the influence of this Hadley circulation that would produce a reinforcement of atmospheric anomalies in the Pacific ocean cannot be sta- the ascending motions over the Equatorial Indian Ocean, ted with this kind of experiments, and coupled modelling leading to the reinforcement of the upward motions in the studies are needed to state the existence or not of inter- lower troposphere. This feature can be also observed in the actions between basins, as well as the feedbacks between presence of positive precipitation anomalies in the Equa- Atlantic and Pacific basins in relation with the West torial Indian Ocean for all the simulations. The perturba- African monsoon. Also, additional AGCM sensitivity tion of the vertical motions in the Indo-Pacific basin would experiments for a tropical mode including the Atlantic and produce a gradient of SLP that would strengthen the trades the Indo-Pacific basins should be done in order to study the between these two points, with the appearance of surface associated mechanism in relation to the summer WA divergence over the Equatorial central Pacific that would anomalous rainfall. favour the conditions for a La Ninã event in the Pacific. Also, our results show how this possible impact appears Acknowledgments This study was supported by the EU-AMMA from July, when the deep convection associated to the project and the Spanish MEC project CGL2006-04471. Based on French initiative, AMMA was built by an international scientific Atlantic SST anomalies returns to the west and the maxi- group and is currently funded by a large number of agencies, espe- mum upper divergent flow is placed over the Amazon cially from France, UK, US and Africa. It has been the beneficiary of region. The influence of the Atlantic Equatorial mode on a major financial contribution from the European Community’s Sixth the Pacific would lead to a development of ENSO events Framework Research Programme. Detailed information on scientific coordination and funding is available on the AMMA International beginning in the central Pacific, which would be consistent website http://www.amma-international.org. The authors thank Javier with the work as reported by Wang (1995) and Terray and Garcıá-Serrano and Elsa Mohino for his very useful comments on the Dominiak (2005), that state that from the climate shift manuscript, and the two anonymous reviewers for their constructive the onset phase of an ENSO event is characterized by the suggestions and comments. appearance of SST anomalies in the Central Pacific that propagate eastward. According to Wang and An (2002), the References main factor affecting the observed changes in ENSO is the change in the basic state Equatorial winds and in the An S-I, Wang B (2000) Interdecadal change of the structure of the upwelling associated with it, pointing to the increased ENSO mode and its impact on the ENSO frequency. J Clim importance of the atmospheric bridge in the ENSO 13:2044–2055 Chang P, Fang Y, Saravanan R, Ji L, Seidel H (2006) The cause of the behaviour. Thus, previous results would support the idea of fragile relationship between the Pacific El Ninnõ and the Atlantic an influence of the Atlantic variability in the Pacific trop- Ninõ. Nature 443:324–328. doi:10.1038/nature05053 ical basin through changes in the atmospheric circulation. De´que´ M, Dreveton C, Braun A, Cariolle D (1994) The ARPEGE/IFS The observed negative link between Pacific and Atlantic atmosphere model: a contribution to the French community climate modeling. Clim Dyn 10:249–266 Ninõs could have an important influence on the summer Fedorov A, Philander SG (2000) Is El Ninõ changing? Science Sahelian rainfall. Although it has been documented that the 288:1997–2002 response to an isolated warm Pacific forcing is the decrease Giannini A, Saravannan R, Chang P (2003) Oceanic forcing of Sahel of the precipitation over Sahel (Rowell et al. 1995; Janicot rainfall on interannual to interdecadal time scales. Science 302:1027–1030 et al. 2001; Giannini et al. 2003), when focusing in recent Gill AE (1980) Some simple solutions for the heat induced tropical decades, the relationship varies and a warming in the circulation. Quart J Met Soc 106:447–462 123 T. Losada et al.: Tropical response to the Atlantic Equatorial mode

Graham NE (1994) Deacadal-scale climate variability in the tropical Nitta S, Yamada T (1989) Recent warming of tropical sea surface and North Pacific during the 1970s and 1980s: observations and temperature and its relationship to the Northern Hemisphere results. Clim Dyn 10:135–162. doi:10.1007/BF00210626 circulation. J Meteorol Soc Jpn 67:375–383 Hagos SM, Cook KH (2005) Influence of surface processes over Polo I, Rodrı´guez-Fonseca B, Losada T, Garcıá-Serrano J (2008) Africa on the Atlantic marine ITCZ and South American Tropical Atlantic variability modes (1979–2002). Part I: time- precipitation. J Clim 18:4993–5010 evolving SST modes related to West African rainfall. J Clim Hourdin F, Musat I, Bony S, Braconnot P, Codron F, Dufresne JL, 21:6457–6475. doi:10.1175/2008JCLI2607.1 Fairhead L, Filiberti MA, Friedlingstein P, Grandpeix JY, Rasmusson EM, Carpenter TH (1982) Variations in the tropical sea Krinner G, LeVan P, Li ZX, Lott F (2006) The LMDZ4 general surface temperature and surface wind fields associated with the circulation model: climate performance and sensitivity to Southern Oscillation/El Ninõ. Mon Weather Rev 110:354–384 parametrized physics with emphasis on tropical convection. Richter I, Mechoso CR, Robertson AW (2008) What determines the Clim Dyn 27:787–813 position and intensity of the South Atlantic Anticyclone in Janicot S, Rodriguez-Fonseca B (2007) Results on SST-forced signals austral winter? An AGCM Study. J Clim 21:214–229 in AMIP simulations. AMMA-EU project. Deliverable D1.1.1e Rodriguez-Fonseca B, Polo I, Garcia-Serrano J (2008) Recent trends Janicot S, Trzaska S, Poccard I (2001) Summer Sahel-ENSO in the Pacific-Atlantic connection. In: Proceedings of the ‘‘6a teleconnection and decadal time scale SST variations. Clim Asamblea Hispano Portuguesa de Geodesia y Geofisica’’, Dyn 18:303–320 Tomar, Portugal Jansen MF, Dommenget D, Keenlyside N (2009) Tropical atmo- Roeckner E, Arpe K, Bengtsson L, Christoph M, Claussen M, sphere–ocean interactions in a conceptual framework. J Clim Du¨menil L, Esch M, Giorgetta M, Schlese U, Schulzweida U 22:550–567 (1996) The atmospheric general circulation model ECHAM-4: Jin F, Hoskins BJ (1995) The direct response to tropical heating in a model descriptions and simulation of present-day climate. Max- baroclinic atmosphere. J Atmos Sci 52:307–319 Planck-Institute report 218, Hamburg, Germany, p 94 Jury MR, Enfield DB, Melice J-L (2002) Tropical monsoons around Rowell DP, Folland CK, Maskel K, Owen JA, Ward MN (1995) Africa: stability of ENSO associations and links with continental Variability of the summer rainfall over tropical North Africa rainfall. J Geophys Res C10(15):1–17 (1906–92): observations and modeling. Quat J Roy Meteorol Soc Keenlyside N, Latif M (2007) Understanding Equatorial Atlantic 121:669–704. doi:10.1002/qj.49712152311 interannual varaibility. J Clim 20:131–142 Smith TM, Reynolds RW (2004) Imporved extended reconstruction Kucharski F, Bracco A, Yoo JH, Molteni F (2007) Low-Frequency of SST (1854–1997). J Clim 17:2466–2477 variability of the Indian monsoon–ENSO relationship and the Terray P, Dominiak S (2005) Indian Ocean sea surface temperature tropical Atlantic: the ‘‘weakening’’ of the 1980s and 1990s. and El Ninõ–southern oscillation: a new perspective. J Clim J Clim 20:4255–4266 18:1351–1368 Kucharski F, Bracco A, Yoo JH, Molteni F (2008) Atlantic forced Trenberth KE (1990) Recent observed interdecadal climate changes component of the Indian monsoon interannual variability. in the northern hemisphere. Bull Am Meteorol Soc 71:988–993 Geophys Res Lett 35. doi:10.1029/2007GL033037 Vizy EK, Cook KH (2001) Mechanisms by which Gulf of Guinea and Kucharski F, Bracco A, Yoo JH, Tompkins A, Feudale L, Ruti P, Eastern North Atlantic Sea surface temperature anomalies can Dell’Aquila A (2009) A Gill-Matsun-type mechanism explains influence African rainfall. J Clim 14:795–821 the Tropical Atlantic influence on African and Indian Monsoon von Storch H, Zwiers FW (1999) Statistical analysis in climate rainfall. Quart J R Met Soc 135:569–579 research. Cambridge University Press, New York. ISBN Kug J-S, Kang I-S (2006) Interactive feedback between ENSO and 0521450713 the Indian Ocean. J Clim 19:1784–1801 Wang B (1995) Interdecadal changes in El Ninõ onset in the last four Latif M, Keenlyside NS (2008) El Ninõ/southern oscillation response decades. J Clim 8:267–285 to global warming. PNAS. doi:10.1073/pnas.0710860105 Wang C (2006) An overlooked feature of tropical climate: Inter- Losada T, Rodriguez-Fonseca B, Janicot S, Gervois S, Chauvin F, Pacific-Atlantic variability. Geophys Res Lett 33:L12702. doi: Ruti P (2009) A multimodel approach to the Atlantic Equatorial 10.1029/2006GL026324 mode. Impact on the West African monsoon. Clim Dyn. doi: Wang B, An S (2002) A GCM-based assessment of the global 10.1007/s00382-009-0625-5 moisture budget and the role of land-surface moisture reservoirs Melice JL, Servain J (2003) The tropical Atlantic meridional SST in processing precipitation. Clim Dyn 20:13–29. doi:10.1007/ gradient index and its relationships with the SOI, NAO and s00382-002-0274-4 Southern Ocean. Clim Dyn 20:447–464. doi:10.1007/s00382- Wang C, Kucharski F, Barimalala R, Bracco A (2009) Teleconnec- 002-0289-x tions of the Tropical Atlantic to the Tropical Indian and Pacific Miller AJ, Cayan DR, Barnett TP, Graham NE, Oberhuber JM (1994) Oceans: a review of recent findings. Spec Issue Meteorol The 1976–77 climate shift of the Pacific Ocean. Oceanography Zeitschrift 18. doi:10.1127/0941-2948/2009/0394 7:21–26

123