Based on four independent datasets, this study presents observational evidence of a shallow meridional circulation cell in the eastern tropical Pacific. In this shallow meridional circulation cell the northerly cross-equatorial return flow from the ITCZ into the Southern Hemisphere is found immediately above the atmospheric boundary layer, in contrast to the classic concept of the Hadley-type deep meridional circulation whose northerly return flow resides in the upper troposphere. The strength and depth of this shallow meridional circulation undergo a distinct annual cycle. Possible causes and climatic implications of this shallow meridional circulation are discussed.
The large-scale meridional circulation in the tropical eastern Pacific, based on conventional wisdom, is similar to the zonal mean Hadley cell. This classic meridional circulation consists of ascending motions at the latitude of the intertropical convergence zone (ITCZ), descending motions south of the equator, a southerly inflow into the ITCZ near the surface, and a northerly return flow in the upper troposphere (e.g., see Fig. 6e in Wang and Enfield 2003). This deep meridional circulation is illustrated by the dashed arrows in Fig. 1, which extends vertically through the entire troposphere. Recent scrutiny of in situ observations provides evidence indicating that this classic conceptual model for the large-scale meridional circulation in the tropical eastern Pacific is subject to modification. As discussed in section 2, independent data of rawinsonde, dropsonde, and wind profiler measurements from different locations in the tropical eastern Pacific and different periods all show a return (northerly) flow from the ITCZ in the lower troposphere atop the atmospheric boundary layer (ABL). This low-level return flow, together with the southerly trades in the ABL, forms a shallow meridional circulation cell (solid arrows in Fig. 1), in contrast to the conventional deep circulation pattern. Similar shallow meridional circulation cells in the divergent part of the winds have been observed previously (see section 3). But this is the first time it is documented using the total wind field. Possible causes and implications of this shallow meridional circulation cell are discussed in section 3.
The observational evidence for the shallow meridional circulation pattern is based on four independent data sources.
a. TAO ship soundings
Upper-air soundings have been launched from research vessels tending the eastern portion of the Tropical Atmosphere–Ocean (TAO) mooring array since 1995. The data collection has been relatively intensive since 1999 as part of monitoring being carried out for the Eastern Pacific Investigation of Climate processes (EPIC) program (Cronin et al. 2002, p. 205). This sampling consists of soundings collected four times a day during the servicing of the longitudinal lines along 95° and 110°W each boreal spring and fall. The vertical resolutions of the soundings vary from about 10 hPa near the surface to 1 hPa near the tropopause. Measured variables include wind speed and direction, temperature, relative humidity, and pressure. Transects along individual lines (normally between 8°S and 8°N) require 5–10 days to complete. Because the ship speeds vary, the soundings are spaced irregularly in latitude.
An all-time (1995–2002) average of meridional wind (υ) and relative humidity (RH) measured by the TAO ship soundings along 95° and 110°W during August–December is plotted in Fig. 2a. The ABL is clearly shown by the high values of RH near the surface and its sharp vertical gradient near the height of 1 km south of the equator. North of the equator, the boundary layer is not as well defined in RH, especially near the latitudes of the ITCZ (6°–8°N) where high values of RH penetrate into the midtroposphere, presumably due to the vertical transport of moisture by atmospheric deep convection. South of the ITCZ, the mean southerly inflow is confined to the ABL. Across the top of the ABL, the meridional wind abruptly changes its direction. Weaker but evident northerly flows are seen in the lower troposphere (1–5 km). The northerly flows, whose strength may vary in latitude and height, extend from the ITCZ latitudes into the Southern Hemisphere. They are what we have referred to as the low-level return flow. They form, together with the boundary layer southerly flow and ascending/descending motions (not shown), a shallow meridional circulation cell.
The low-level return flow and the shallow meridional circulation pattern tend to be more evident in some individual sections, as from the soundings taken during a TAO cruise along 95°W between 1 and 11 November 2000 (Fig. 2b). This particular section illustrates an interesting feature of the low-level return flow, namely, the relatively high moisture content at the same levels. This suggests advection of water vapor from the ITCZ by the low-level return flow, whose implications are discussed in section 3. Also noticed in this region is the much stronger upper-level return flow, which is associated with moisture contents higher than at the levels below in the midtroposphere where southerlies prevail. This indicates southward moisture advection from the ITCZ latitudes by the upper-level return flow. The two return flows simultaneously form a deep and a shallow meridional circulation cell, respectively, as illustrated in Fig. 1.
b. EPIC2001 dropsondes
Atmospheric profiles of temperature, humidity, pressure, and winds were measured by dropsondes launched from research aircraft during a field campaign of EPIC in 2001 (EPIC2001). The aircraft repeated the same flight pattern eight times along 95°W between 0° and 12°N during 7 September–10 October 2001. The atmospheric dropsondes were launched from an altitude of 5.5 km at intervals of 1° latitude. The vertical resolution of the dropsondes is 5 s, equivalent roughly to 0.5 hPa. The eight flight dates were chosen mostly for logistical rather than scientific reasons. Each flight can thus be considered a random snapshot in time spanning through a peak month of the ITCZ. The large-scale background for those flights and detailed descriptions of the dropsondes are given in McGauley et al. (2003, manuscript submitted to J. Climate, hereafter MZB).
Meridional–vertical cross sections of υ and RH are shown in Fig. 3a for the eight flight average and in Fig. 3b for the single flight of 2 October 2001. The low-level return flow is clearly shown in both, confirming the existence of the shallow meridional circulation cell as observed from the TAO ship soundings. The vertical shear in the meridional wind between the southerlies in the ABL and the low-level return flows above is extraordinary in both the TAO ship and EPIC2001 observations. The implication of this shear due to the low-level return flow is discussed in section 3.
c. FGGE dropsondes
During the First Global Atmospheric Research Program (GARP) Global Experiment (FGGE; January–February and May–June 1979), aircraft dropsondes were launched in the eastern Pacific in a region bounded by 15°N–15°S, 90°W–175°E (Kloesel and Albrecht 1989). A zonal mean of υ measured by the dropsonde data shown by Yin and Albrecht (2000) is reproduced in Fig. 4. Northerly flows between 800 and 600 hPa (2–6 km) is clearly shown atop the southerly wind in the ABL. In this case, the low-level return flow is particularly strong south of the equator and extends to a higher level than observed from the EPIC2001 and TAO ship data.
d. Wind profilers
The last observational dataset comes from 915-MHz wind profilers at Christmas Island (2°N, 157.4°W) and San Cristóbal, Galápagos (0.9°S, 89.6°W) since 1995. In the low height mode, the profilers provide wind speeds and directions with vertical resolution of 105 m and vertical extent of 5200 m, which are used here. Hourly averaged horizontal winds are generally accepted as accurate to better than 1 m s−1. Detailed descriptions of the 915-MHz profilers are given by Gage et al. (1994).
Probability distribution functions (PDFs) of daily mean υ measured by the profilers at San Cristóbal and Christmas Island are shown in Fig. 5 to illustrate the most observable vertical profiles of υ. At San Cristóbal (Fig. 5a), strong, persistent ABL southerlies are capped by a layer of substantial vertical shear with a tendency for northerlies above 1 km. Such northerlies atop the boundary layer are also observed by Hartten and Gage (2000) in the seasonal mean. The mean low-level return flow (maximum PDF) reaches its peak at 1.5 km from the surface but it extends all way up to the midtroposphere. At Christmas Island (Fig. 5b) the ABL southerly flow is much weaker than at San Cristóbal and so is the vertical shear, whereas the low-level return flow is apparently stronger. The amplitude of the mean low-level return flow appears to reach the maximum (∼3 m s−1) near 2.5 km. At both locations, large variability exists in υ, meaning that the return flow may not exist all the time. But through most of the lower troposphere above the ABL, the mean, indicated by the maximum PDF, is negative (northerlies).
The low-level meridional return flow measured at the single point at San Cristóbal is representative of a large-scale meridional circulation feature in the region. In Fig. 6, its amplitude is compared to that measured simultaneously at the TAO service ship located through a wide range of latitudes (8°S–8°N) at 95°W during two cruises in 1999 and 2000. Even though the amplitudes and the heights of the low-level return flows measured at different latitudes may not always be the same, there is not a single case in which the low-level return flow is detected at San Cristóbal by the wind profiler but not at other latitudes by ship soundings at the same time, or vice versa. Therefore, mean meridional winds from the point measurement of the wind profilers can be used as an index of the large-scale shallow meridional circulation in the eastern Pacific.
The wind profiler data show a strong seasonal cycle in both the strength and vertical extent of the shallow meridional circulation. At San Cristóbal, its strength and depth are at maximum in December and at minimum in March (Fig. 7a). At Christmas Island (Fig. 7b), the meridional circulation is the strongest and deepest in September, the shallowest in March, and the weakest (in terms of the low-level return flow) in May. The lack of a prominent return flow during the boreal spring is consistent with other aspects of the annual cycle in the region (Mitchell and Wallace 1992). Notably, the asymmetry about the equator is the least this time of year, as evidenced by the double ITCZ in precipitation in the eastern Pacific visible from satellite data during neutral years of the ENSO cycle (Zhang 2001). The maximum amplitude of the low-level return flow is less than a quarter of the maximum amplitude of the ABL southerly trades at San Cristóbal while the two are comparable at Christmas Island.
Data from four independent sources all show the existence of a mean low-level northerly flow atop the atmospheric boundary layer (ABL) in the eastern Pacific. This low-level northerly flow, together with ABL southerlies, ascending motions in the ITCZ, and descending motions south of the equator (as implied by the dearth of clouds above the ABL), form a shallow meridional circulation cell. This shallow meridional circulation contrasts with the Hadley-type deep meridional cell, which has long been viewed as the only dominant meridional circulation pattern in the region. The observational results presented here suggest that the conventional view of the tropical large-scale circulation needs to be modified for the tropical eastern Pacific to include the shallow meridional circulation cell.
The limitation in the data used here may raise concerns regarding the conclusion on the existence of a shallow meridional circulation. The TAO ship soundings suffer from aliasing (it takes 5–10 days to complete a meridional transect). The FGGE data were collected during several flights over a period of about a month (May–June 1979); the flight patterns were not restrictively at fixed longitudes. The only true instantaneous (relatively to the large-scale circulation) measurements of the meridional–vertical cross section are from the eight EPIC2001 flights, which cover only north of the equator. The only continuous time series are point measurements by the island wind profilers.
These data limitations notwithstanding, the main conclusion of this study on the existence of the shallow meridional circulation has been made out of two considerations. First, the low-level return flow is repeatedly revealed by four independent datasets without exception. These datasets are almost all in situ observations currently available in the eastern Pacific region that can be used to depict the large-scale meridional circulation. They cover different longitudes and time periods. Second, the observations are consistent to diagnoses of global model analyses. Trenberth et al. (2000) found that, in two global model analysis products, the first leading vertical EOF mode of the tropical large-scale divergent circulations represents the traditional Hadley–Walker-type deep circulations; the second leading mode represents shallow circulations confined to the lower troposphere. They questioned whether this second mode is realistic or not because there was no observational support for that mode. The results presented here provide observational evidence for their shallow mode.
Important questions pertaining to the shallow meridional circulation can be raised. Does it also exist in other part of the Tropics? What is its cause? What are its roles in tropical climate? The existing observations are insufficient to fully answer these questions. Brief speculative discussions are given here to motivate further studies.
There is no reason to exclude the possibility that similar shallow meridional circulations also exist in other tropical regions, even though no direct observational evidence outside the eastern Pacific is yet available. This possibility is supported by the diagnoses of Trenberth et al. (2000). They found from an analysis of divergent flow in a global analysis dataset shallow meridional circulation cells in many other tropical regions, such as Africa, the Atlantic Ocean, and the Americas (see their Fig. 17). Such a shallow meridional circulation pattern in the divergent flow in monthly global model reanalysis is quite common (e.g., Fig. 12 in Tomas and Webster 1997). This is the first time that the shallow meridional circulation is documented using the total wind field. Preliminary results from a comparison between our observations and some global model simulations and reanalysis indicate that the low-level return flow is significantly underrepresented in the total wind fields from the models (not shown). The fact that the shallow meridional circulation can be reproduced but is underestimated by global models suggests that its primary cause is related to large-scale dynamics, which should be well represented by the models; but its strength might depend on physical processes that have to be parameterized in global models, often unreliably (such as cumulus convection). Tomas and Webster (1997) related the shallow meridional circulation in the divergent wind to advection of absolute vorticity. Wu (2003) suggested that a shallow overturning circulation consisting of flows in the boundary layer and the lower troposphere exists only in the absence of deep convection. Murakami et al. (1992) speculated that a shallow meridional circulation might exist in the eastern Pacific because of its lack of abundant deep convection. Recent satellite observations have shown that precipitating systems in the ITCZ are often indeed relatively shallow (e.g., Nesbitt et al. 2000). It is conceivable that if shallow convection in the ITCZ is essential to the observed strength of the shallow meridional circulation, then the underestimate of the strength implies an underestimate of shallow convection in the ITCZ in the models. We notice, however, that in one of our observations (Fig. 2b), the shallow and deep circulations can coexist. In any case, the dynamics of the shallow meridional circulation at large remains unknown.
The low-level return flow may have significant climatic implications in the tropical eastern Pacific through its effects on ABL winds and clouds. The southerly flow of the ABL in the eastern Pacific is mainly driven by the meridional gradient in sea surface temperature (SST) (Lindzen and Nigam 1987), but other factors such as momentum entrainment at the top of the ABL are also important (Stevens et al. 2002; MZB). The existence of the low-level return flow tends to reduce the ABL southerlies. The northerlies above the ABL enhance the entrainment due to increased shear production of turbulence at the top of the ABL, and represent a source of air with meridional momentum of the opposite sense. By influencing the ABL and surface winds, the low-level return flow indirectly participates in air–sea interaction, which is deemed crucial to the climate variability in the eastern Pacific (e.g., Xie and Philander 1994). Previous studies (e.g., Schneider and Lindzen 1977; Wu 2003) suggested that the response of the large-scale meridional circulation to deep convective heating in the ITCZ is a deep overturning cell confined to the free troposphere with its lower branch southerly flow atop the ABL; to shallow convective heating, it is a shallow overturning cell with its upper branch northerly flow atop the ABL, as observed in this study. Based on these, we propose a mechanism for deep convection in the ITCZ to play a role in air–sea interaction in the eastern Pacific. Deep convective anomalies may enhance ABL and surface southerlies by weakening the low-level return flow of the shallow meridional circulation and thereby reducing the vertical wind shear and northerly momentum entrainment across the top of the ABL.
Through its moisture advection (e.g., Fig. 2b), the low-level return flow is also liable to be important to ABL clouds, which play a key role in the climate system south of the equator but are often poorly simulated by current global models (Yu and Mechoso 1999). Enhanced moisture atop the ABL south of the equator in association with the low-level return flow might enhance ABL clouds by lessening the drying effects of entrainment or limit ABL clouds by reducing radiative cooling at the top of the clouds. It has yet to find out which effect dominates. The shallow meridional circulation as conceptualized in Fig. 1 may not represent the actual trajectory of air parcels. Strong zonal winds exist in the area (Hartten and Gage 2000). At the location and altitude of the low-level return flow, the zonal wind is typically easterly at 2–8 m s−1. These and other matters merit further study on the shallow meridional circulation. Existing observational data are, however, very limited in tropical oceanic regions. Numerical models offer promise, but their capabilities of reproducing the shallow meridional circulation must first be quantified.
The authors would like to thank Ken Gage for making the wind profiler data available to this study. This research was supported by NSF through Grant ATM0002363 (CZ, MM) and by the Joint Institute for the Study of the Atmosphere and Ocean (JISAO) under NOAA Cooperative Agreement No. NA17RJ1232 (NAB).
Corresponding author address: Dr. Chidong Zhang, RSMAS, University of Miami, 4600 Rickenbacker Causeway, Miami, FL 33149-1098. Email: email@example.com
National Oceanic and Atmospheric Administration Contribution Number 995 and Pacific Marine Environmental Laboratory Contribution Number 2558.