15 FEBRUARY 2002 WANG 399 Atmospheric Circulation Cells Associated with the El NinÄo±Southern Oscillation CHUNZAI WANG Physical Oceanography Division, NOAA Atlantic Oceanographic and Meteorological Laboratory, Miami, Florida (Manuscript received 20 February 2001, in ®nal form 10 July 2001) ABSTRACT Atmospheric circulation cells associated with the El NinÄo±Southern Oscillation (ENSO) are described and examined using the NCEP±NCAR reanalysis ®eld and the NCEP sea surface temperature (SST) from January 1950 to December 1999. The divergent wind and pressure vertical velocity are employed for the identi®cation of atmospheric circulation cells. The warm phase of ENSO shows positive SST anomalies in the equatorial eastern Paci®c and along the east coast of Asia and the west coast of North America, and negative SST anomalies in the off-equatorial western Paci®c and in the central North Paci®c. Associated with this SST anomaly distribution are variations of atmospheric zonal and meridional circulation cells over the Paci®c. The equatorial zonal Walker circulation cell is weakened, consistent with previous schematic diagrams. The anomalous meridional Hadley circulation cell in the eastern Paci®c shows the air rising in the Tropics, ¯owing poleward in the upper troposphere, sinking in the subtropics, and returning back to the Tropics in the lower troposphere. The anomalous Hadley cell in the western Paci®c is opposite to that in the eastern Paci®c. The divergent wind and vertical velocity also show a midlatitude zonal cell (MZC) over the North Paci®c. The mean MZC is characterized by the air rising in the central North Paci®c, ¯owing westward and eastward in the upper troposphere, descending in the east coast of Asia and the west coast of North America, then returning back to the central North Paci®c in the lower troposphere. The anomalous MZC during the mature phase of El NinÄo shows an opposite rotation to the mean MZC, indicating a weakening of the MZC. 1. Introduction ing ENSO are well known (e.g., Webster and Chang 1988; McPhaden et al. 1998). However, how the Walker In a seminal paper, Bjerknes (1969) postulated an circulation cell evolves during ENSO from data has not atmospheric circulation cell in the zonal-vertical plane been well studied, probably because of a lack of ob- over the equatorial Paci®c, which he named the ``Walker servational data. The Walker circulation cell associated circulation'' since this circulation is part of the global with ENSO has been shown little quantitatively. Southern Oscillation phenomenon de®ned earlier by Sir The atmosphere also has meridional circulation cells: Gilbert Walker (1923, 1924, 1928). The Walker circu- lation cell is characterized as the air ascending in the the Hadley cell and the Ferrel cell (e.g., Trenberth et al. equatorial western Paci®c, ¯owing eastward in the upper 2000; and references therein). The Hadley circulation troposphere, sinking in the equatorial eastern Paci®c, cell is also thermally driven, located in the tropical and and returning toward the equatorial western Paci®c in subtropical regions. The heated tropical air rises and the lower troposphere. Bjerknes also visualized a close ¯ows aloft toward the subtropical region where it cools, relation among the Southern Oscillation, east±west sea sinks, and ¯ows back to the tropical region. The Ferrel surface temperature (SST) contrast in the equatorial Pa- cell is an extratropical meridional circulation cell char- ci®c Ocean, and the thermally driven Walker cell. Since acterized by the air ascending in the extratropical region then, the zonal Walker circulation cell has been rec- and descending in the subtropical region. Unlike the ognized to be associated with the interannual phenom- Walker and Hadley cells, the Ferrel cell is a thermally enon of the El NinÄo±Southern Oscillation (ENSO) that indirect cell. The Ferrel cell is forced mostly by transient has been intensively studied (e.g., Philander 1990; baroclinic eddy activity through associated poleward McCreary and Anderson 1991; Neelin et al. 1998). heat and momentum transports (e.g., Holton 1992). Lit- Schematic diagrams of the Walker circulation cell dur- tle is known about how these atmospheric meridional cells vary during the evolution of ENSO. Also, a strong zonal wind speed core, called the ``jet stream,'' is lo- Corresponding author address: Dr. Chunzai Wang, Physical cated just below the tropopause in the extratropics. Pre- Oceanography Division, NOAA Atlantic Oceanographic and Mete- vious studies (e.g., Horel and Wallace 1981) suggested orological Laboratory, 4301 Rickenbacker Causeway, Miami, FL 33149. that ENSO teleconnections link the jet stream with E-mail: [email protected] ENSO. 400 JOURNAL OF CLIMATE VOLUME 15 The recently available data of the National Centers gence is the divergent part of the wind although the ro- for Environmental Prediction±National Center for At- tational part is usually larger. In the search for evidence mospheric Research (NCEP±NCAR) reanalysis ®eld of atmospheric circulation cells, it is essential not only (Kalnay et al. 1996) provide an opportunity to study the to isolate the divergent part of the wind but also to as- atmospheric circulation patterns associated with ENSO. certain the continuity following the ¯ow between the The present paper uses the NCEP±NCAR reanalysis centers of upward and downward motion (e.g., Krish- ®eld and the NCEP SST data (Smith et al. 1996) to namurti et al. 1973; Hastenrath 2001). The analyses of describe and investigate how the Walker cell, the Hadley the divergent wind and vertical motion are prerequisites cell, the Ferrel cell, and the jet stream vary during the indispensable for the identi®cation of atmospheric cir- evolution of ENSO. Additionally, the combination of culation cells. Following these, we will mainly focus on atmospheric divergent wind and vertical motion data the distributions of atmospheric vertical motion and the shows an atmospheric zonal cell in the midlatitudes of divergent component of the wind when we discuss at- the North Paci®c. The rest of the paper is organized as mospheric circulation cells. Newman et al. (2000) com- follows. Section 2 introduces the data used in this paper. pared the 200-mb wind divergence ®elds from the Section 3 shows the annual variability of the atmo- NCEP±NCAR, the European Centre for Medium-Range spheric circulation patterns. Sections 4 and 5 show com- Weather Forecasts (ECMWF), and the National Aero- posites and temporal variations of the atmospheric cir- nautics and Space Administration (NASA) reanalyses. culation associated with ENSO, respectively. Section 6 Although the details of the divergence ®elds are different provides a discussion and summary. among these analyses, basic patterns are reasonably con- sistent (also Trenberth et al. 2000). Monthly SST data are also used in this study. SST 2. Data data are taken from the NCEP SST data set on a 28 lat The major data source in this study is the NCEP± 3 28 long grid from January 1950 to December 1999. NCAR reanalysis ®eld from January 1950 to December These SST ®elds were produced by using a spatial in- 1999. The NCEP±NCAR reanalysis ®eld uses a state- terpolation method employing empirical orthogonal of-the-art global data assimilation system on a 2.58 lat function analysis (see Smith et al. 1996 for the detailed 3 2.58 long grid (see Kalnay et al. 1996 for details). description). With all of these data, we ®rst calculate Variables used in this study are monthly atmospheric monthly climatologies based on the full record period horizontal wind velocity, vertical velocity, and velocity (1950±99) and then anomalies are obtained by subtract- potential at levels of 1000, 925, 850, 700, 600, 500, ing the monthly climatologies for each dataset from the 400, 300, 250, 200, 150, and 100 mb. The vertical com- data. ponent of wind ®eld in the NCEP±NCAR reanalysis ®eld is pressure vertical velocity. In our presentation, 3. Annual variability we multiply the pressure vertical velocity by 21, so positive values of the vertical velocity indicate an up- To better understand anomaly variations of atmo- ward movement of air parcels. Since we are interested spheric circulation cells, we ®rst consider annual vari- in tropical atmospheric circulation and northern mid- ability of tropospheric circulation patterns. Figure 1 latitude atmospheric circulation associated with the Pa- shows the boreal winter (January) climatologies of tro- ci®c ENSO, the analyses in this paper are shown from pospheric circulation. Centers of low (high) velocity 208S±608Nto1008E±808W. Although our analyses were potential are associated with divergent out¯ow (con- performed globally, our focus herein is on the Paci®c. vergent in¯ow) winds. Figures 1a±c show that diver- Results associated with Atlantic climate variability are gence (convergence) at the upper troposphere corre- reported in a companion paper (Wang 2002, manuscript sponds to convergence (divergence) at the lower tro- submitted to J. Climate). posphere, associated with upward (downward) vertical Horizontal wind velocity can be divided into a non- motion at the midtroposphere (three levels of 200, 500, divergent (or rotational) part and a divergent (or irrota- and 850 mb are chosen as representative of the upper, tional) part (e.g., Mancuso 1967; Krishnamurti 1971; mid-, and lower troposphere, respectively). Centers of Krishnamurti et al. 1973): v 5 vc 1 vf 5 k 3 =c 1 upper-tropospheric divergence and lower-tropospheric =f, where c is streamfunction and f is velocity potential. convergence in the equatorial region just west of the The ®rst part does not contribute to atmospheric divergent date line are characterized by strong upward velocity at ®elds associated with atmospheric vertical motion (it is the midtroposphere with maximum upward velocity in nondivergent). It is well known that the Walker and Had- the off-equatorial regions. The equatorial eastern Paci®c ley cells are thermally driven, associated with atmo- is associated with upper-tropospheric convergence, low- spheric convergence±divergence.