How ENSO impacts precipitation in southwest central Asia

GEOPHYSICAL RESEARCH LETTERS, VOL. 34, L16706, doi:10.1029/2007GL030078, 2007

How ENSO impacts precipitation in southwest central Asia Annarita Mariotti1,2 Received 19 March 2007; revised 25 June 2007; accepted 19 July 2007; published 18 August 2007.

[1] A linkage between ENSO and the hydroclimatic variability of the southwest central Asia region (SWCA) is established through observational analysis of precipitation, moisture flux and sea level pressure data, with further support from an atmospheric model of intermediate complexity. Enhanced precipitation in SWCA during warm ENSO events results from an anomalous southwesterly moisture flux coming from the Arabian Sea and tropical Africa, which is generated along the northwestern flank of the high pressure anomaly over the Indian and western Pacific Oceans, part of the canonical ENSO sea-saw pressure anomalies. The ENSO impact on SWCA precipitation is found to be greatest in the transition seasons of autumn and spring, but the dynamical impact on pressure and circulation persists throughout the year. This connection was particularly strong in recent decades. Model sensitivity experiments further show that this is driven primarily by tropical Pacific SST anomalies and associated large-scale sea-level pressure changes, while the Indian Ocean SST has opposite effects. Citation: Mariotti, A. (2007), How ENSO impacts precipitation in southwest central Asia, Geophys. Res. Lett., 34, L16706, doi:10.1029/2007GL030078.

tion) events was also suggested to explain the large fluctuations in Caspian Sea level during the 20th century [Arpe et al., 2000]. More broadly, observational analyses indicate a relationship between ENSO events and interannual precipitation variability in parts of SWCA in autumn [Kiladis and Diaz, 1989; Mason and Goddard, 2001; Nazemosadat and Ghasemi, 2004; Mariotti et al., 2005] and to a lesser extent, in winter [Tippett et al., 2005; Nazemosadat and Ghasemi, 2004; Syed et al., 2006]. On the other hand the influence of the North Atlantic Oscillation (NAO) on the region is small [Mariotti et al., 2005; Mariotti and Arkin, 2006; Syed et al., 2006]. [4] In spite of these highly relevant studies, the SWCA region has been largely ‘neglected’ by the climate community, partly because of the complexity of the hydroclimatology in the region, and partly because of its distance from the major known modes of climate variability such as ENSO and the NAO. The goal of this work is to advance the understanding of how ENSO events impact climate variability in SWCA by analyzing the spatial and temporal variability of precipitation and atmospheric circulation, and to explore the underlying mechanisms by means of model experiments.

1. Introduction

2. Data and Model

[2] The southwest central Asia region (SWCA; for convenience the region of interest is defined by a box, as in Figure 2 of section 3, including countries such as Iraq, Iran, Afghanistan, Turkmenistan, Uzbekistan, Kazakhstan, as well as parts of eastern Europe) is characterized by arid highlands, deserts and vast steppes, with complex topography formed by the mountain ranges of the Zagros, Alborz, Hindu-Kush, Pamir and Tian Shan. Precipitation is largely determined by mid-latitude westerly cyclones that move in during the colder seasons and interact with the orography [Martyn, 1992; Rodionov, 1994; Trigo et al., 2002; Evans et al., 2004]. [3] As a result, rain is concentrated in the winter-time, with also significant amounts in the transition seasons of spring and autumn (see Figures S1– S5).1 Hydroclimatological variability in this region is large. For instance, a very severe drought was experienced in parts of southwest Asia during the period 1998– 2002 [Waple et al., 2002] and a link with the La Nin˜a-like conditions during this period has been suggested [Barlow et al., 2002; Hoerling and Kumar, 2003]. The influence of ENSO (El Nin˜o Southern Oscilla-

[5] Land precipitation data used in this study include the high resolution (0.5°  0.5°) gauge-only CRU dataset for the period 1948 – 2000 [Mitchell and Jones, 2005] and the NCEP/CPC PRECL data for the period 1948 – 2003 [Chen et al., 2004]. As data sparsity is a problem for SWCA, blended satellite-rain gauge products are also analyzed for the period since 1979: GPCP data [Adler et al., 2003]; CMAP data [Xie and Arkin, 1997]; CAMS-OPI, gauge-only over land [Janowiak and Xie, 1999]. All are available monthly on a 2.5°  2.5° grid up to 2006. Re-analyses monthly precipitation (from 6-hr forecasts) is also used: NCEP/NCAR data ([Kalnay et al., 1996]; since 1948, roughly at 1.9° resolution) and ECMWF ERA-40 data ([Uppala et al., 2004]; 1957 – 2002, at 2.5°  2.5° resolution). Although in reanalyses precipitation is model-derived, this computation is based on atmospheric quantities which are nudged to observations. NCEP sea-level pressure (SLP) and vertically integrated moisture flux are analyzed to describe more comprehensively the atmospheric anomalies. [6] We use the model QTCMg [Zeng et al., 2004], the global version of the Quasi-Equilibrium Tropical Circulation Model [Neelin and Zeng, 2000; Zeng et al., 2000]. The model is based on the primitive equations and includes a relatively complete set of physical parameterizations. Vertical structure is simplified and most accurate near strongly convecting zones, further away it is comparable

1 Earth System Science Interdisciplinary Center, University of Maryland, College Park, Maryland, USA. 2 Ente per le Nuove Tecnologie, l’Energia e l’Ambiente (ENEA), Rome, Italy.

Copyright 2007 by the American Geophysical Union. 0094-8276/07/2007GL030078

1 Auxiliary materials are available in the HTML. doi:10.1029/ 2007GL030078.

L16706

1 of 5

L16706

MARIOTTI: ENSO IMPACT ON SOUTHWEST CENTRAL ASIA

L16706

series [Ebisuzaki, 1997] or the standard Student’s t-testing technique.

3. Characteristics of the ENSO Impact

Figure 1. Correlation of southwest central Asia precipitation index and the NINO3.4 index for the period 1948 – 2000. (a) Seasonal correlation using various datasets (see legend; for ERA-40 the correlations are for the period 1958 – 2000). Full symbols are for 95% significant values. (b) Timeseries of September – May southwest central Asia precipitation index (black) and NINO3.4 (green). to a two-layer model. Resolution is 5.6°  3.7°. The model was forced by observed SST from 1870 – 2002 (GISST; N. A. Rayner et al., Version 2.2 of the Global Sea-Ice and Sea Surface Temperature data set, 1903 – 1994, Climate Research Technical Note, 74, 1997, available from the Hadley Centre, UKMO, Bracknell, Berks). Sensitivity experiments were conducted using SSTs from different basins, in particular the tropical Pacific and Indian Oceans, while climatological SSTs were used elsewhere. All are 9member ensemble runs and the ensemble means are analyzed here (anomalies are departures from the run with global climatological SSTs). [7] A SWCA index is defined as area-averaged precipitation for the region 40°E –73°E and 30°N – 47°N. ENSO activity is represented by the NINO3.4 index [Barnston et al., 1997]. Seasonal anomalies are detrended using a leastsquares linear regression fit. Linear correlation with the NINO3.4 index is computed for two intercomparison periods: 1948– 2000 and 1979 – 2006. For the longer period, ‘‘warm’’ (El Nin˜o) minus ‘‘cold’’ (La Nin˜a) ENSO composite anomalies are computed based on the upper and lower quartiles of the NINO3.4 timeseries (27 events altogether). Significance for the various analyses is estimated using a Monte Carlo simulation using surrogate time series with the same power spectra as the original time

[8] The correlation between the SWCA precipitation index and NINO3.4 over the period 1948– 2000 is positive throughout the rainy season (Figure 1). Values are highest and most consistent among datasets in autumn (ASO-OND; roughly 0.5– 0.6). In spring (MAM-AMJ) values are generally lower (roughly 0.3 – 0.4) but still significant. In winter, only the observational datasets give significant results (about 0.3). In summer there is no signal as this is the dry season for the region. For the September to May means the correlation is about 0.5. The correspondence between ENSO events and SWCA precipitation anomalies is particularly good since the late 1970s (see Figure 1b). An ENSO composite analysis for the period 1948 – 2000 (not shown) indicates that the autumn and spring precipitation anomalies occur respectively before and after the mature phase of ENSO events, typically in winter (autumn anomalies represent 10– 30% of annual precipitation). [9] Figure 2 shows the ENSO precipitation and circulation patterns for the September – May mean anomalies as well as for individual seasons. For precipitation, the September – May correlation is positive in a large region East of the Caspian Sea. The signal is particularly strong in SON: precipitation correlation is positive in a large area South and East of the Caspian Sea with maxima (over 0.45) around the major mountain ranges. In MAM and DJF, the patterns have smaller extent and are mostly confined to the eastern part of the region of interest (further South in MAM compared to DJF). [10] A southwesterly moisture flux is associated with the September – May precipitation pattern during warm ENSO events (correlation is about 0.4 South of the Caspian Sea). This circulates on the northwestern flank of a large-scale positive SLP anomaly, part of the western pole of the ‘‘seasaw’’ pattern typical of ENSO events [Trenberth and Caron, 2000] (see Figures S1– S5) also affected, via Rossby wave response, by anomalous atmospheric heating in the eastern Indian Ocean [Annamalai et al., 2005]. The southwesterly flux brings moisture to SWCA across the Arabian Peninsula mainly from the Arabian Sea and tropical Africa-ITCZ region. The SON and MAM moisture flux and SLP patterns are similar to the September – May mean, although differences exist: in SON, the southwesterly moisture flux is reinforced in virtue of its stronger Arabian Sea component, apparently a result of the stronger meridional SLP anomaly gradient over the eastern Indian Ocean; in MAM, the fluxes from the Arabian Sea and tropical Africa are reduced and stay separate, as is also the case for the precipitation anomalies downstream. Compared to other seasons, in DJF these fluxes converge further South, the resulting flux is more zonal and is connected to SWCA precipitation anomalies only through a secondary meridional flux; SLP anomalies are weaker over the Arabian Sea and stronger over tropical Africa, as a result the SLP pattern is also more zonal and further South. Interestingly, similar pressure and flow anomalies also exist in the summer season (not shown), but there is little rainfall anomaly because of the climatological lack of cyclonicity during this season.

2 of 5

L16706

MARIOTTI: ENSO IMPACT ON SOUTHWEST CENTRAL ASIA

L16706

Figure 2. Seasonal ENSO patterns for the period 1948 – 2000. Shown are the correlation with NINO3.4 for precipitation (CRU; 95% significant values are shaded) and SLP (NCEP; contours), and the regression with the vertically integrated moisture flux (vectors, in Kg m-1s-1). SWCA focus region is boxed (red). (a) September – May means. (b) DJF means. (c) MAM means. (d) SON means. [11] Additional analyses focusing on the period 1979 – 2006 (see Figures S1 –S5) indicate that the ENSO impact over SWCA has been particularly strong in recent decades (as also suggested by Figure 1b). Seasonal correlation values for the SWCA index are high (between 0.5– 0.7) throughout the rainy season (slightly lower in JFM) and very consistent among the greater number of observational datasets available for this shorter period. Compared to the longer period (see Figure 2a), the 1979– 2000 September – May ENSO precipitation pattern is stronger South of the Caspian Sea and weaker to its East. As for the longer period, this occurs in connection with a southwesterly moisture flux. [12] In summary, the analyses above suggest that the enhanced southwesterly moisture flux from the Arabic Sea and Africa is the culprit of increased SWCA precipitation during El Nin˜o. In contrast with the ENSO teleconnection mechanisms in many areas (i.e. the Amazon or Indonesia), where enhanced precipitation anomalies directly correspond to lower SLP, over SWCA the SLP anomaly is small. In fact for this region, the positive rainfall anomalies correspond to even positive (albeit not-significant) pressure anomalies (Figure 2). The southwesterly moisture flux anomaly flows along the northwestern flank of the expansive high SLP anomaly over the Indian Ocean and western Pacific, broadly, the opposite pole, in the Southern Oscillation sea-saw pattern, to that over the eastern Pacific.

4. Pacific Versus Indian SST [13] However the mechanism responsible for this flux and in particular the roles played by SST anomalies in the

Pacific and Indian Oceans are not clear. On one hand, the major role of the Pacific Ocean in ENSOs suggests its potential importance. On the other hand, Indian Ocean SST undergoes robust changes during ENSOs, and given its proximity to the SWCA region may also play an important role. [14] To explore these issues, ensemble model experiments were conducted with forcing from observed SST. The seasonal correlation of the SWCA index from the ensemble mean of simulations forced with global SST anomalies gives results that are similar to observations, with positive correlation values throughout the rainy season peaking in autumn (see Figures S1 –S5). Analyses below focus on the autumn season (SON) when the ENSO signal over SWCA is strongest in the observations. The SON ENSO composite anomalies of precipitation and moisture flux from observations and the model experiments are displayed in Figure 3 (see Figures S1 – S5 for observational SST and SLP composites). The model simulates the broad ENSO teleconnection pattern such as wetting in the eastern Equatorial Pacific and drying in the western Pacific, but the precipitation anomalies over the Indian Ocean and Africa are somewhat shifted to the East. The degree of agreement with observations is comparable to a typical atmospheric GCM [see also Zeng et al., 2000]. In the SWCA region, the model simulates a southwest-northeast oriented wet anomaly (El Nin˜o) that is similar to observations, but of smaller amplitude. The modeled moisture flux anomalies also show a broadly similar pattern, with major inflow from the Arabian Sea, though the flow over Africa differs somewhat from observations. A remote teleconnection, such as that to

3 of 5

L16706

MARIOTTI: ENSO IMPACT ON SOUTHWEST CENTRAL ASIA

L16706

Figure 3. Precipitation and moisture flux ENSO composite anomalies for autumn (1948– 2000) (a) from observations: land precipitation from CRU, oceanic precipitation and moisture flux from NCEP (b) from the ensemble of simulations with global SST anomalies (c) from the ensemble with SST anomalies in the Pacific Ocean-only (d) from the ensemble with SST anomalies in Indian Ocean-only. Precipitation is mm/d (shaded; note color bars differ for observations and model results). Moisture flux vector length is rescaled to display the relatively small anomalies away from the main ENSO forcing (a.u). For precipitation, only 95% significant anomalies are plotted.

SWCA away from the ENSO centers of action, is often difficult to capture with a model. Thus, this general agreement lends confidence in further sensitivity experiments to identify the relative role of SSTs from different oceanic basins. [15] When Pacific-only SST anomalies are used to force the model, the response over SWCA is similar to the global SST simulations, but the signal is significantly stronger in both precipitation and moisture flux. In fact, the main differences compared to these simulations are the enhanced easterly moisture fluxes and decreased precipitation over the Indian Ocean, leading to greater precipitation amounts downstream over eastern tropical Africa and SWCA. In contrast, in the Indian SST-only runs the pattern is one of moisture convergence and enhanced precipitation over the western Indian Ocean. As a result, drier than usual conditions are seen in upstream surrounding regions including SWCA. Thus, the role of Indian Ocean SST anomalies is to counter the Pacific influence which would otherwise have an even stronger impact on SWCA (similar competing mechanisms have been found by Goddard and Graham [1999]). Overall, the model results support the mechanism suggested by the observational analysis that wetter (drier) SWCA during El Nin˜o (La Nin˜a) is caused by enhanced (reduced) moisture flux flowing around the northwest flank of the high (low) pressure anomaly over the Indian Ocean.

5. Conclusion

autumn. The associated circulation pattern during El Nin˜o (La Nin˜a) involves a southwesterly (northeasterly) moisture flux that brings more (less) moisture into this region. This flux flows along the northwestern flank of the large-scale high pressure anomaly over the Indian and western Pacific Oceans, broadly, the western pole of the Southern Oscillation sea-saw pattern. Unlike many ENSO teleconnections in which lower pressure and weaker subsidence leads to more precipitation, this mechanism does not require a change in the local pressure, but rather involves a change in the tropical moisture supply to subtropical-midlatitude storms; a mechanism that may potentially be important more broadly for tropical-midlatitude interactions [Branstator, 2002; Selten et al., 2004]. This mechanism, complements that suggested by Barlow et al. [2002] for ENSOs with especially strong western Pacific warm pool anomalies, involving Rossby wave propagation and a change of subsidence over the region. [17] Results from an intermediate atmospheric model forced by observed SSTs support such a dynamical link between SWCA and ENSO (autumn). Model sensitivity experiments further show that the tropical Pacific plays a primary role. In contrast, warming of the western Indian Ocean during El Nin˜o tends to counter the effect of the Pacific Ocean by inducing moisture convergence over itself at the expense of precipitation in neighboring regions including SWCA. This link and its seasonal manifestations deserve futher investigation in more realistic GCM experiments.

[16] Observational analyses suggest that ENSO significantly affects precipitation in SWCA. The influence persists throughout the rainy season, and is particularly strong in

[18] Acknowledgments. The author is grateful to all those who provided data for this work, to Jin-Ho Yoon for providing the model simulations and Ning Zeng for the insightful discussions. The manuscript

4 of 5

L16706

MARIOTTI: ENSO IMPACT ON SOUTHWEST CENTRAL ASIA

benefited from the constructive comments of H. Annamalai and an anonymous reviewer. This work was partially funded by the NASA program ‘‘Research Opportunities for Precipitation Measurement Missions’’ (NRA-02-OES-05).

References Adler, R. F., et al. (2003), The version-2 global precipitation climatology project (GPCP) monthly precipitation analysis (1979 – present), J. Hydrometeorol., 4(6), 1147 – 1167. Annamalai, H., S. P. Xie, and J. P. McCreary (2005), Impact of Indian Ocean sea surface temperature on developing El Nin˜o, J. Clim., 18, 302 – 319. Arpe, K., L. Bengtsson, G. S. Golitsyn, I. I. Mokhov, V. A. Semenov, and P. V. Sporyshev (2000), Connection between Caspian Sea level variability and ENSO, Geophys. Res. Lett., 27(17), 2693 – 2696. Barlow, M., H. Cullen, and B. Lyon (2002), Drought in central and southwest Asia: La Nin˜a, the Warm Pool, and Indian Ocean precipitation, J. Clim., 15, 697 – 700. Barnston, A. G., M. Chelliah, and S. B. Goldenberg (1997), Documentation of a highly ENSO-related SST region in the equatorial Pacific, Atmos. Ocean, 35, 367 – 383. Branstator, G. (2002), Circumglobal teleconnections, the jet stream waveguide, and the North Atlantic Oscillation, J. Clim., 15, 1893 – 1910. Chen, M., P. Xie, J. E. Janowiak, P. A. Arkin, and T. M. Smith (2004), Verifying the reanalysis and climate models outputs using a 56-year data set of reconstructed global precipitation, paper presented at the 84th AMS annual meeting, Am. Meteorol. Soc., Seattle, 11 – 15 January. Ebisuzaki, W. (1997), A method to estimate the statistical significance of a correlation when the data are serially correlated, J. Clim., 10, 2147 – 2153. Evans, J. P., R. B. Smith, and R. J. Oglesby (2004), Middle East climate simulation and dominant precipitation processes, Int. J. Climatol., 24, 1671 – 1694. Goddard, L., and N. E. Graham (1999), Importance of the Indian Ocean for simulating rainfall anomalies over eastern and southern Africa, J. Geophys. Res., 104(D16), 19,099 – 19,116. Hoerling, M., and A. Kumar (2003), The perfect ocean for drought, Science, 299, 691 – 694. Janowiak, J. E., and P. Xie (1999), CAMS-OPI: A global satellite-rain gauge merged product for real-time precipitation monitoring applications, J. Clim., 12(11), 3335 – 3342. Kalnay, E., et al. (1996), The NCEP/NCAR 40-year reanalysis project, Bull. Am. Meteorol. Soc., 77, 437 – 471. Kiladis, G. N., and H. F. Diaz (1989), Global climatic anomalies associated with extremes in the Southern Oscillation, J. Clim., 2, 1069 – 1090. Mariotti, A., and P. Arkin (2006), The North Atlantic Oscillation and oceanic precipitation variability, Clim. Dyn., 28(1), 35 – 51, doi:10.1007/ s00382-006-0170-4.

L16706

Mariotti, A., J. Ballabrera-Poy, and N. Zeng (2005), Tropical influence on Euro-Asian autumn rainfall, Clim. Dyn., 24(5), 511 – 521, doi:10.1007/ s00382-004-0498-6. Martyn, D. (1992), Climates of the World, 436 pp., Elsevier, New York. Mason, S. J., and L. Goddard (2001), Probabilistic precipitation anomalies associated with ENSO, Bull. Am. Meteorol. Soc., 82, 619 – 638. Mitchell, T. D., and P. D. Jones (2005), An improved method of constructing a database of monthly climate observations and associated highresolution grids, Int. J. Climatol., 25(6), 693 – 712. Nazemosadat, M. J., and A. R. Ghasemi (2004), Quantifying the ENSOrelated shifts in the intensity and probability of drought and wet periods in Iran, J. Clim., 17, 4005 – 4018. Neelin, J. D., and N. Zeng (2000), The first quasi-equilibrium tropical circulation model-formulation, J. Atmos. Sci., 57, 1741 – 1766. Rodionov, S. N. (1994), Global and Regional Climate Interaction: The Caspian Sea Experience, 241 pp., Kluwer Acad., Baton Rouge. Selten, F. M., G. Branstator, M. Kliphuis, and H. A. Dijkstra (2004), Tropical origins for recent and future Northern Hemisphere climate change, Geophys. Res. Lett., 31, L21205, doi:10.1029/2004GL020739. Syed, F. S., F. Giorgi, J. S. Pal, and M. P. King (2006), Effect of remote forcings on the winter precipitation of central southwest Asia part 1: Observations, Theor. Appl. Climatol., 86, 147 – 160, doi:10.1007/ s00704-005-0217-1. Tippett, M. K., L. Goddard, and A. G. Barnston (2005), Statistical-dynamical seasonal forecasts of central-southwest Asian winter precipitation, J. Clim., 18(11), 1831 – 1843. Trenberth, K. E., and J. Caron (2000), The Southern Oscillation revisited: Sea level pressures, surface temperatures, and precipitation, J. Clim., 13, 4358 – 4365. Trigo, I. F., G. R. Bigg, and T. D. Davies (2002), Climatology of cyclogenesis mechanisms in the Mediterranean, Mon. Weather Rev., 130, 549 – 569. Uppala, S., P. Kallberg, A. Hernandez, S. Saarinen, M. Fiorino, X. Li, K. Onogi, N. Sokka, U. Andrae, and V. Da Costa Bechtold (2004), ERA-40: ECMWF 45-year reanalysis of the global atmosphere and surface conditions 1957 – 2002, ECMWF Newsl., 101, 2 – 21. Waple, A. M., et al. (2002), Climate Assessment for 2001, Bull. Am. Meteorol. Soc., 83, 938 – 938. Xie, P. P., and P. A. Arkin (1997), Global precipitation: A 17-year monthly analysis based on gauge observations, satellite estimates, and numerical model outputs, Bull. Am. Meteorol. Soc., 78(11), 2539 – 2558. Zeng, N., J. D. Neelin, and C. Chou (2000), The first quasi-equilibrium tropical circulation model-implementation and simulation, J. Atmos. Sci., 57, 1767 – 1796. Zeng, N., H. Qian, E. Munoz, and R. Iacono (2004), How strong is carbon cycle-climate feedback under global warming?, Geophys. Res. Lett., 31, L20203, doi:10.1029/2004GL020904. A. Mariotti, Earth System Science Interdisciplinary Center, 2207 Computer and Space Sciences Building, University of Maryland, College Park, MD 20742-2425, USA. ([email protected])

5 of 5