Climatology of the Tropical Atmosphere and Upper Ocean

k k

1

Climatology of the Tropical Atmosphere and Upper Ocean

This brief introduction identifies a number of phe- nomena that characterize the spatial and temporal variability of the tropical climate. Planetary scale cir- culations include Hadley-like circulations, the more zonal Walker systems, westerly ducts (WDs), tropical upper tropospheric troughs (TUTTs), quasi-biennial lower stratospheric oscillations, and the monsoon sys- tems. Each of these features is “planetary” in scale and oscillates on interannual time scales. Within these slowly varying features of the tropical atmosphere and ocean, there is a myriad of circulation features with higher-frequency variability. These include the Madden–Julian Oscillation (MJO), quasi-biweekly variability, easterly waves, tropical cyclones and depres- sions, and a high-amplitude diurnal cycle. These higher-frequency scales of variability are not indepen- dent and are modulated by lower-frequency events such as the MJO, which, in turn, is modulated by the annual cycle and interannual variability. In subsequent chapters we will clarify the underlying mechanisms of these circulation features and their roles in weather and climate.

1.1 The Growth of Tropical Meteorology

Prior to the 1970s, the abiding concern in meteorol- ogy was the development of numerical models for the extratropical weather prediction. In the early twentieth century V. Bjerknes (1904) had identified the neces- sary physical laws to describe atmospheric motion and established the concept of prediction as an initial value problem. He also realized that the equations could not be solved analytically. Richardson (1922),^Ithough, showed that solutions could be attained numerically using a finite difference approach for solving the nonlin- ear differential equations that, eventually, would form the basis of weather and climate modeling. However, Richardson’s first attempts at weather prediction, cal- culated for a small region by hand, possessed extremely large errors. It would be 30 years later when numerical

prediction proved fruitful with the development of the electronic computer and the identification of approx- imations that would reduce Richardson’s numerical errors.

In the decades that followed, a deeper understanding of fundamental modes of the midlatitudes and their instabilities developed. Rossby (1940), for example, had explained that waves in the extratropical westerlies resulted from the conservation of potential vorticity.

Charney (1947) and Eady (1949) revealed that these waves arose from instabilities of the westerly wind regime growing at the expense of the energy of the background flow and formed efficient agents for energy and momentum transfer. With the emergence of the electronic computer, and an increasing network of atmospheric data, extratropical numerical weather prediction became a reality. A very useful history of numerical weather prediction is given by Harper et al.

(2007), marking the 50th anniversary of the first US operational numerical weather predictions made by the Joint Numerical Weather Prediction Unit in July 1954.

Whereas the focus of early numerical weather predic- tion was the northern hemisphere (NH) middle latitudes, the need to monitor and understand tropical phenom- ena grew rapidly. It was soon realized that to forecast events in an extratropical region for even short forecast horizons, a global model would be required. For such a model to work, global initial data were necessary. It became clear that errors arising from the neglect of the tropics or the southern hemisphere (SH) led to a rapid degradation of forecast skill in the NH. Thus, for very practical reasons, both an improved tropical database and a keener understanding of tropical phenomenology were required.

Prior to the 1960s, atmospheric data were especially scarce in the tropics. Nevertheless, some key climatolog- ical features were well known. Fields of sea-surface tem- perature (SST), obtained mostly along shipping routes, varied slowly in space and time compared to higher lat- itudes. Furthermore, temperature and pressure in both the atmosphere and the ocean possessed smaller hori- zontal gradients than at higher latitudes. The trade wind

k k regimes of both hemispheres were known to be steadier

than any other wind system on the planet.

Monsoons, and their distinct seasonality, had been described, at least grossly, for over 300 years and were thought to be explained largely in terms of differential heating of the ocean and the land. Near the equator, the equatorial atmosphere was characterized by a state of relative inactivity and stagnation called, initially,the dol- drums.^IIWithin these regions, there existed considerable

“unsteadiness” with squalls and propagating convective disturbances. Also, it was realized that the tropics were the source region of the most violent storms on the planet, tropical cyclones, that impacted the tropics, subtropics, and extratropics alike.

Interestingly, Halley (1686) had identified many of these large-scale characteristics of the tropics and pro- duced an atlas of surface winds based on ships’ logs that had been painstakingly collected over the years. These observations were of such widespread interest and economic value that the compilations were published in 1690 by the Royal Society. These climatologies were being used by mariners setting out on commercial ven- tures around the world. Scientists attempted to explain the physics behind the patterns of observed wind and weather.¹

The numerical weather models being developed in the 1960s required initial data that were far more detailed than a mere climatological description of the tropics.

Driven by this growing need for data, the World Mete- orological Organization (WMO) and the International Council of Scientific Unions (ICSU) in 1967 launched a 15-year project called the Global Atmospheric Research Program (GARP). Besides improving the atmospheric data stream, it organized a system for global data col- lection and spawned several important field experiments including GARP Atlantic Tropical Experiment (GATE, boreal summer 1974: Houze and Betts (1981)). GATE is still the largest and most complex international scientific experiment ever undertaken, with 10 nations – Brazil, Canada, France, Federal Republic of Germany, German Democratic Republic, Mexico, Netherlands, USA, UK, and USSR – working in close collaboration contributed 39 specially equipped ships, 13 large research aircraft, several meteorological satellites, and some 5000 person- nel to an intensive three-month study of weather systems in the tropical eastern Atlantic Ocean. A further 50 countries in Africa and South America participated by making special land surface and upper-air observations.

A subsequent major international field program, under the auspices of GARP, gathered data on the Asian monsoon. This was the Monsoon Experiment

1 These early efforts to explain the trade winds and monsoons are discussed extensively in Chapters 10 and 14.

(MONEX), with a field phase occurring in 1978–1979.

There were a number of subprograms contained within MONEX: the summer and winter monsoon experi- ments (Summer MONEX and Winter MONEX). There were three major emphases: an Arabian Sea component (June–July 1979), the Arabian desert campaign and the Bay of Bengal component (BOB: August 1979), and Winter MONEX (December–February (DJF) 1978–1979). Each phase of MONEX also had an oceanographic component.

By the 1970s, a number of influential studies had high- lighted the importance of the tropics in global weather and climate. At the same time, research began to unravel some of the basic physics of low-latitude meteorology and oceanography. For example:

(i) Matsuno (1966)^IIIshowed that there were classes or families of atmospheric planetary scale waves that were trapped about the equator and propagated both with and against the prevailing winds of the tropics. Here “trapped” has a special connotation.

Trapped waves have a peculiar structure, essentially decreasing in amplitude away from the equator.

Matsuno’s waves turned out to be subsets of global modes that, depending on the parameter range chosen, become increasingly trapped about the equator, depending on the equivalent depth (see Chapter 6) of the fluid and the Doppler-shifted frequency of the mode. These waves were quickly connected to the propagating convection across the tropics. Further, it was shown that similar families of wind-driven planetary scale waves existed in the ocean, also with maximum variance at low latitudes. We will explore the nature of these waves in some detail, including the impact of a varying basic state on their structure, in Chapters 6 and 7.

(ii) In the 1920s, Sir Gilbert Walker^IV had shown that there was a planetary scale 2–5-year oscillation in surface pressure across the tropics between the eastern and western hemispheres.² He referred to this mode as the “Southern Oscillation” or SO. However, its physical nature remained unex- plained, and largely unexplored, for the next 40 years. Troup (1965) and Bjerknes (1969)^V had the inspiration to suggest that Walker’s oscillation had roots in both the atmosphere and the ocean.

In essence, these studies enabled the conjoining of the low-frequency variability of the atmosphere and the ocean by explaining the basic physics of a phenomenon that has been shown to alter the weather and climate of vast swathes of the planet.

The oceanic part of the phenomenon was El Niño and the coupled phenomena became known as El

2 Walker (1924a,b, 1928).

k k Niño–Southern Oscillation (ENSO). Thus emerged

the concept of coupled ocean–atmospheric modes where the scale, amplitude, and period of the phenomena are determined jointly by both the dynamics of the atmosphere and the ocean through a series of feedbacks. The need for data to clarify joint connections between the atmosphere and ocean led to the multi-year Tropical Ocean–Global Atmosphere (TOGA) and the TOGA Coupled Ocean–Atmosphere Response Experiment (TOGA COARE).

(iii) A third major development came with the obser- vation that the tropics contain distinctly unique subseasonal phenomena not seen in the extra- tropics. Madden and Julian (1971), against the challenges of scare data, identified a large-scale eastward propagating entity, evident in both pres- sure fields and convection, appearing to form in the Indian Ocean and to propagate eastward across the equatorial ocean. Now named the MJO,³ this phenomenon possessed strong variance in the 30–60 day period band. Like ENSO, it too seemed to have strong teleconnections influencing remote parts of the tropics and subtropics.

Recently, Li et al. (2018) reported on an earlier sys- tematic study by Xie et al. (1963), where radiosonde stations in the warm pool region of the Pacific Ocean were analyzed. A robust signal was uncovered with a 40–50 day oscillation. The oscillation commenced with an acceleration of low-level westerlies over South India and South Asia, followed by an eastern propagation of low-frequency wave-like phenomena. Xie et al. noted a strong association of typhoon genesis and that this newly discovered phenomena may have predictive util- ity. This early work complements for a smaller region and over a more limited time span the more global results of Madden and Julian (1971). We also believe that it is important to acknowledge what Li et al. (2018) refer to as “hidden gems” that exist in journals that, hitherto, were not globally available.

Many field campaigns were designed to clarify these new ideas. The TOGA program gathered simultane- ous data from the atmosphere and the ocean to test Bjerknes’s hypothesis with the hope that it would lead to the prediction of the ENSO phenomena. The initial field experiment has grown into the Global Tropical Monitoring Array,^VI of which there are three compo- nents: Tropical Atmosphere Ocean (TAO) array in the equatorial Pacific Ocean, Research Moored Array for African–Asian–Australia Monsoon Analysis and Pre- diction (RAMA) in the Indian Ocean, and Prediction

3 Madden and Julian (1971, 1972).

and Research Array in the Atlantic (PIRATA) in the Atlantic.

Further field campaigns were designed specifically to understand the role of moist atmospheric processes in the tropics. Prior to GATE, early thinking centered around the vision that tropical convection was made up almost exclusively of convective towers with an incidental stratiform component. Following GATE, it was realized that the stratiform layer extended some hundreds of kilometers out from the convective clouds and accounted for about 50% of oceanic precipitation.⁴ However, more data of the coupling of dynamics and moist convection were required to understand the nature of the connections. This need led to additional field cam- paigns such as the Equatorial Mesoscale Experiment (EMEX),⁵ which was conducted in Northern Aus- tralia during January–February 1987 together with the Australian Monsoon Experiment (AMEX).⁶ The South China Sea Monsoon Experiment (SCSMEX)⁷ was carried out in the May–August period in 1988.

The TOGA Coupled Ocean–Atmosphere Response Experiment (TOGA COARE), perhaps the largest international field program since GATE, was conducted in the western equatorial Pacific. The Intensive Observ- ing Period (IOP) occurred from 1 November through 28 February 1993. During these four months, nearly 1200 people from more than 20 nations conducted more than 700 days of ship operations, released nearly 12 000 raw- insondes, completed 125 aircraft flights, and maintained continuous operation of 30 moored instrument systems.

Details of the experimental plans can be found in Web- ster and Lukas (1992). A central scientific objective was the detailed determination ocean–atmosphere fluxes during disturbed and undisturbed conditions over the warm pool. A preliminary summary of the results of the experiment appears in Godfrey et al. (1998).

A follow-up program (Joint Air–Sea Monsoon Experiment: JASMINE⁸) took place in the BoB in the summer of 1998 to seek similarities and differ- ences between the Pacific and Indian Ocean warm pools. Later, the “Dynamics of the MJO” experiment (DYNAMO) was designed and implemented in the equatorial Indian Ocean and Indonesian archipelago during late 2011–early 2012.⁹

During the last few decades, the tropical database has improved immensely. First, a multitude of satellites, carrying diverse instrumentation, provide observations at an unprecedented frequency and resolution. Now,

4 E.g. Schumacher and Houze (2003).

5 Webster and Houze (1991).

6 Holland et al. (1986).

7 Lau et al. (2000).

8 Webster et al. (2002a, b, c).

9 Gottschalck et al. (2013).

k k the total tropical data set consists of an amalgam of

satellite data, together with observations from more traditional sources (e.g. surface meteorological stations, upper air soundings, commercial aircraft observations, and moored and floating and profiling ocean buoys).

However, the resultant body of data is inhomogeneous in space and time and is, thus, difficult to ingest and analyze. The mix of data led to the creation of the

“reanalysis system”aimed at rendering the vast fields of data into systematic gridded data. This is accomplished by using a weather prediction or a climate model as a data assimilator. Every 6–12 hours spatially and tempo- rally non-homogeneous data is ingested into a model that performs spatial and temporal averaging. Out of this system comes a gridded data stream advancing through the years that is dynamically and thermo- dynamically consistent. Such data sets are routinely produced by a number of international agencies.

The initial concept of reanalysis was created by Bengtsson and Shukla (1988). A comprehensive overview of reanalysis products can be found in the Uni- versity Corporation of Atmospheric Research (UCAR)/

National Center for Atmospheric Research (NCAR) Climate Data Guide located at https://climatedataguide .ucar.edu/climate-data/atmospheric-reanalysis-overview- comparison-tables. The original NCAR/National Cen- ter for Environmental Prediction (NCEP) data set was described in Kalnay et al. (1996). The ERA-40 reanalysis data set is described in Uppala et al. (2005).

In the early days of numerical weather prediction, there was an expectation that if the initial state of the atmosphere and the ocean was better described, and if the forecast model was increasingly representative of the physical system, that the quality of the forecast would improve. This is true in one respect as an egregiously poor initial data or model will produce an egregiously bad forecast. Yet, the discovery of nonlinear error growth (Lorenz 1969) changed the philosophy of numer- ical weather and climate prediction irrevocably. Rather than seeking a deterministic outcome (i.e. the weather will bexat dayy) one seeks a probabilistic forecast from an ensemble of forecasts (the weather at a particular location willxat dayybut at a certain probability: e.g.

Leutbecher and Palmer 2008). The European Centre for Medium Range Weather Forecasts (ECMWF) conducts forecasts with time scales of weeks some 100 times each day with slightly perturbed initial conditions. In essence, this form of forecasting places an even greater premium on initial data. The poor documentation of a remote system in the tropics, for example, can introduce errors and very rapidly influence the probability distribution of weather and climate globally.

Probabilistic forecasts allow reasoned decision mak- ing (e.g. Palmer 2002^VII) through the simple relationship

that risk of an event occurring (R) is given by the prod- uct of the probability of the occurrence (P) of the event times the potential cost of the event (C). That is, R=P×C. Such forecasts allow a user to decide whether or not to take action, whether it is to determine the risk of a cholera outbreak (Thompson et al. 2006) or the possibility of a major broad-scale flooding (Webster et al. 2010; Hopson and Webster 2010). All of these factors place additional demand on describing and understanding tropical phenomena.

This chapter commences the study of tropical phe- nomena with a description of low-latitude climatology and the associated variability using the reanalysis database. In subsequent chapters we will attempt to understand these phenomena from a fundamental dynamical and thermodynamical perspective.

1.2 Seasonal Characteristics

To analyze the zonally axisymmetric nature of the tropi- cal climate it is convenient to define zonal and time aver- ages of an arbitrary variable𝛼as

[𝛼(𝜑,z,t)] =

∫

2𝜋 0

𝛼(𝜆, 𝜑,z,t)d𝜆 and 𝛼(𝜆, 𝜑,z) =

∫

t₂

t₁

𝛼(𝜆, 𝜑,z,t)dt (1.1a)

where𝜆,𝜑,z, andtare longitude, latitude, height, and time, respectively. Collectively, we can write:

[𝛼(𝜑,z)] =

∫

t₂ t₁ ∫

2𝜋 0

𝛼(𝜆, 𝜑,z,t)d𝜆dt (1.1b) to define a quantity averaged both in longitude and time.

30

25 20

15 10

5 40°S 30°S 20°S 10°S 0 10°N 20°N 30°N 40°N DJF

JJA Annual

°C

latitude T_SST (φ)

Figure 1.1 Latitudinal distribution of the zonally averaged sea-surface temperature (SST:[T_SST(𝜑)]) for the two solstitial seasons (DJF and JJA) and the annual average (Annual). In all seasons and in the annual mean, maximum SST occurs in the summer hemisphere and not at the equator. Units:∘C.

Source: Data from NCEP reanalysis.

k k Figure 1.2 Zonally averaged vertical

atmospheric temperature[T(𝜑,z)]

(∘C: color bar) and specific humidity [q_v(𝜑,z)](10⁻³kg kg⁻¹contours) for (a) JJA and (b) DJF. Isotherms<−60∘C are highlighted by white contours.

Source: Data from NCEP reanalysis, that possesses no moisture data above 300 hPa.

–80 –70 –60 –50 –40 –30 –20 –10 0 10 20 30 100

200

400

600

800

1000 100 200

400

600

800

1000

°C

p (hPa)p (hPa)

Temperature and specific humidity

90°S 60°S 30°S EQ 30°N 60°N 90°N

90°S 60°S 30°S EQ

(a) JJA

(b) DJF

30°N 60°N 90°N

10

10 8

6 4 2

12

8 6 4 2 2

1.2.1 Zonal Variability 1.2.1.1 Sea–Surface Temperature

Figure 1.1 displays the zonally averaged mean SST [T_SST(𝜑)] for the two solstitial seasons June–August (JJA) and DJF in addition to the annual average (ANN).

In all three profiles, a broad and relatively flat maximum spans the low latitudes. The mean maximum SST is not located at the equator but in the summer hemisphere some 5–7∘ from the equator. In the boreal summer the maximum is near 5∘N whilst during the boreal winter, two smaller maxima occur located, respectively, at 7–8∘S and 5∘N. Even the annual average (black curve) possesses an off-equator SST maximum. SST decreases with latitude across the subtropics and more rapidly in higher latitudes.

1.2.1.2 Temperature and Humidity

The vertical distribution of the zonally average seasonal mean air temperature ([T]∘C) and specific humidity

([q_v]g kg⁻¹)¹⁰for the two solstitial seasons are plotted in Figure 1.2. The vertical distribution of specific humidity (contours) shows a rapid decrease with height and tends to follow the latitudinal structure of temperature.

The warmest tropospheric temperatures and the high- est specific humidity occur in the summer hemisphere although the asymmetry between the two hemispheres is more pronounced during JJA when warm tempera- tures extend further poleward than during the austral summer. This poleward extension is likely to be associ- ated with the northward location of the Asian Monsoon precipitation.

Perhaps the most striking feature of the zonal temper- ature sections are the extreme low values in the equato- rial upper troposphere. This tropical “cold dome” is the coldest part of the global atmosphere with the exception, perhaps, of the winter stratosphere over Antarctica.

10 Here, the NCEP/NCAR reanalysis products were used. However, these data sets do not include humidity data above 300 hPa.

k k

MAMJJA SONDJF MAMJJA

SON DJF

MAMJJA SON DJF

MAMJJA SONDJF MAMJJA

SON DJF

9 8 7 6 5 4 3 2 1

0 80S 60S 40S 20S 0 20N 40N 60N 80N

latitude rainfall (1010 m3/day)

9 8 7 6 5 4 3 2 1

0 80S 60S 40S 20S 0 20N 40N 60N 80N

latitude rainfall (1010 m3/day)

9 8 7 6 5 4 3 2 1

0 80S 60S 40S 20S 0 20N 40N 60N 80N

latitude rainfall (1010 m3/day)

9 8 7 6 5 4 3 2 1

0 80S 60S 40S 20S 0 20N 40N 60N 80N

latitude rainfall (1010 m3/day)

9 8 7 6 5 4 3 2 1

0 80S 60S 40S 20S 0 20N 40N 60N 80N

latitude rainfall (1010 m3/day)

9 8 7 6 5 4 3 2 1

0 80S 60S 40S 20S 0 20N 40N 60N 80N

latitude rainfall (1010 m3/day)

80S 60S 40S 20S 0 20N 40N 60N 80N

latitude

80S 60S 40S 20S 0 20N 40N 60N 80N

latitude

80S 60S 40S 20S 0 20N 40N 60N 80N

latitude

MAMJJA SON DJF

9 8 7 6 5 4 3 2 1 0 10

rainfall rate (mm/day)

9 8 7 6 5 4 3 2 1 0 10

rainfall rate (mm/day)

9 8 7 6 5 4 3 2 1 0 10

rainfall rate (mm/day)

(a) Rainfall rate (mm/day) (b) Volume (10¹⁰m³/day) (c) Annual volume (10¹⁰m³/day)

(i) Land + ocean (i) Land + ocean (i) Land + ocean

(ii) Ocean (ii) Ocean (ii) Ocean

(iii) Land (iii) Land (iii) Land

6.88×10¹¹ 6.68×10¹¹

5.64×10¹¹ 4.82×10¹¹

1.24×10¹¹ 1.86×10¹¹

Figure 1.3 Zonally averaged precipitation[P(𝜑)]between 90∘S and 90∘N for JJA, SON, DJF, and MAM in terms of (a) average rain rate (mm day^–1) and (b) total precipitation by volume (m³day^–1) that takes into account the contraction of area with increasing latitude.

Distributions are shown for the annual average, JJA, SON, DJF, and MAM. (c) Annual precipitation by volume. Total seasonal precipitation (land plus ocean) is shown in panels (i) as well as the contributions over the ocean (panels ii) and over the land areas (panels iii). Results are summarized in Table 2.1. The precipitation data set used in these calculations was the Global Precipitation Climatology Project (GPCP, Adler et al. 2003) data set. Calculations using the more latitudinally constrained CMORPH data set yielded very similar results.

Source: Data sets from the NOAA Climate Prediction Center MORPHing technique (CMORPH, Joyce et al. 2004).

Jordan (1954) thought that the explanation of the high tropospheric temperatures presented a considerable challenge for tropical meteorology.

1.2.1.3 Precipitation

Figure 1.3 plots the zonally averaged seasonal rainfall rate distributions [P(𝜑)]between 90∘S and 90∘N. Two different forms of rainfall rate are used: the zonally aver- aged rainfall rate (mm day⁻¹: Figure 1.4a) and the rain- fall rate in terms of volume (10¹⁰m³day⁻¹: Figure 1.4b).

The latter form takes into account the geometric con- traction of area of both land and ocean with increasing latitude. For example, if the zonally average rainfall rate at 30∘and 60∘latitude are the same, thetotalrainfall will be greater at 30∘because of its greater circumference.^VIII

Rainfall rates of both kinds are separated into three cat- egories: total rain rates (land plus ocean: top row), ocean rain rates (middle row), and land rates (bottom row).

The annual distributions of precipitation rate by volume appear in Figure 1.3c.

Overall, the largest precipitation rates (mm day⁻¹) occur in the tropics (Figure 1.3a(i)) irrespective of the metric used. The JJA profile has a strong maximum near 10–15∘N. This distribution is in contrast to DJF, where two near-equatorial maxima occur on both sides of the equator: a minor maximum just north of the equator and a second, the major, extending to almost 20∘S. The two solstitial distributions possess maxima in the NH with the September, October, November season (SON) rainfall rates being quite similar to those of JJA. The

k k

1000 800 600 400 200

1000 800 600 400 200

(i) Zonal wind (m s^–1)

latitude (c) DJF

1000 800 600 400 200

1000 800 600 400 200

(i) Zonal wind (m s^–1)

latitude (d) MAM

1000 800 600 400 200

90°S 60°S 30°S EQU 30°N 60°N 90°N

(ii) Meridional wind (m s^–1), 500 hPa vertical velocity (i) Zonal wind (m s^–1)

1000 800 600 400 200

(a) JJA (b) SON (i) Zonal wind (m s^–1)

–3 –2 –1 0 1 2 3

10–4 hPa s–1

–3 –2–1 0 1 23

10–4 hPa s–1

–3 –2–1 01 23

10–4 hPa s–1

–3 –2 –1 0 1 2 3

10–4 hPa s–1

pressure (hPa)pressure (hPa)

(ii) Meridional wind (m s^–1), 500 hPa vertical velocity

(ii) Meridional wind (m s^–1), 500 hPa vertical velocity (ii) Meridional wind (m s^–1), 500 hPa vertical velocity

pressure (hPa)pressure (hPa)

latitude latitude

90°S 60°S 30°S EQU 30°N 60°N 90°N

90°S 60°S 30°S EQU 30°N 60°N 90°N 90°S 60°S 30°S EQU 30°N 60°N 90°N

90°S 60°S 30°S EQU 30°N 60°N 90°N

90°S 60°S 30°S EQU 30°N 60°N 90°N 90°S 60°S 30°S EQU 30°N 60°N 90°N

90°S 60°S 30°S EQU 30°N 60°N 90°N

200 400 600 800 1000

200 400 600 800 1000

Figure 1.4 Seasonal distribution of the zonally averaged zonal wind ([U(z, 𝜑)]m s^–1, upper panels) and the meridional wind ([V(z)], m s^–1, lower panels) for (a) JJA, (b) SON, (c) DJF, and (d) MAM, respectively. Light blue shading in the upper panels denotes zonally averaged easterly winds. Red arrows denote the direction of the meridional wind. The solid red curve in the lower panels shows the 500 hPa zonally averaged vertical velocity (10⁻⁴hPa s^–1). Dark shading at the base of the panels shows the zonally averaged surface elevation.

Source: Data from NCEP reanalysis.

March, April, May season (MAM) zonally averaged precipitation maximum is closer to the equator (5∘N) but there is a secondary maximum rate near 10∘S. Over- all, the seasonal sequence of precipitation portrays a migration following the seasonal “march of insolation.”

There is an exception, however. A near-equatorial maximum persists in the NH in all seasons except DJF.

The lower two panels of Figure 1.3a differentiate between land and ocean rainfall rates. In general, the rain rates (mm day⁻¹) over land are greater than the rain rates over the ocean. They, too, possess a seasonal variability following the sun, but with much larger amplitudes.

Consider now the rain rates byvolumeas a function of latitude (m³day⁻¹), where large differences occur between land and ocean areas. Due to the much smaller area of the land compared to the ocean area, especially in the SH, the total rainfall by volume over land is far

less than the total over the ocean (cf. the bottom two panels of Figure 1.3b). Finally, Figure 1.3c shows the latitudinal distribution of the total zonally and annu- ally averaged rainfall by volume. The numbers in the panels refer to the summed hemispheric rainfall volume (10¹¹m³day⁻¹). The lower two panels show the volume of rainfall over the land and ocean. It is interesting to note that despite the vast differences in the geography of the NH and SH, annually (right-hand column) the total hemispheric rainfall by volume in each hemisphere is almost identical. The rainfall by volume is summarized in Table 2.1.

1.2.1.4 Wind Fields

The mean seasonal zonal averaged wind components ([U(𝜑,z)] and [V(𝜑,z)]) appear in the top and bottom panels of Figure 1.4a–d for all four seasons. There are a number of easily identifiable regimes. Easterlies

k k (shaded blue) occupy most of the lower tropospheric

subtropics. These represent the trade winds that have equatorial meridional and easterly components in both hemispheres. Broad-scale westerlies jets, with maxi- mum strength in the midlatitude dominate the upper troposphere near 300–200 hPa. In the solstitial seasons the stronger of the two westerly jet streams is located in the winter hemisphere. During the equinoctial seasons the midlatitude westerly jets are fairly symmetrical about the equator. The maximum westerlies are in the middle latitudes and the westerly winds increase with height (i.e.𝜕[U]∕𝜕z>0). Also, maximum winds in the winter hemisphere occur near where𝜕[T]∕𝜕𝜑is a maximum.

The magnitude of the zonal mean meridional wind is far weaker than the zonal wind, especially outside the tropics. During JJA, the low-latitude surface meridional wind is northward crossing the equator, with a south- ward return flow in the upper troposphere. During DJF, the circulation is reversed. The meridional flow is also confined vertically with maximum values in the lower and upper troposphere. All these features are part of the zonal mean Hadley Circulation, a direct circulation rais- ing warmer air and subsiding cooler air in the subtropics.

The zonally averaged vertical velocity is also plotted in the bottom panels of Figures 1.4d. The strongest fields occur in the two solsticial seasons¹¹between the lower level convergence of the meridional wind and the upper level divergence.

1.2.2 Spatial Variability in the Tropics

Whereas the latitudinal structure offers only weak latitu- dinal gradients at low latitudes, much richer fields occur in longitude.

1.2.2.1 Surface Temperature

The distribution of surface temperature, T_s(𝜑, 𝜆)_, between 45∘N and 45∘S for all four seasons is displayed in Figure 1.5 over both the land and ocean. The heavy black contour over the oceans encloses SST ≥28 ∘C an area often called the ocean warm pool (OWP).¹² The black contour over the land encloses a mean temperature≥36∘C.

The longitudinal gradients of SST in the tropical ocean basins are much weaker than at higher latitudes.

At all times of the year, the warmest SSTs reside in the western parts of the Pacific and Atlantic basins. The gradients are reversed in the Indian Ocean. The com- bination of the warm water of the Pacific and Indian

11 The wordsolsticecomes from the Latin for sunsolandstilor stareorto standreflecting the most poleward apparent migration of the sun. Equinox is also from the Latin indicating equal nightnox and day.

12 The 28∘C delimiter of the warm pool is somewhat arbitrary.

Oceans form the conglomerate “Indo-Pacific warm pool” that straddles the Indonesian Archipelago. There is a seasonal shift of the warm pool moving slightly about the equator into the summer hemisphere roughly following the insolation maximum.

During the boreal spring (MAM), the Indian Ocean SST possesses the warmest ocean temperatures on the planet. At this time, the near-equatorial zonal surface pressure difference across the Indian Ocean is near zero compared to about 2–3 hPa during the other seasons.

Changes have occurred in the equatorial Pacific Ocean as well. The longitudinal pressure difference along the equator is <2 hPa, compared to a maximum of 4 hPa during DJF. These seemingly subtle differences will be shown to be rather important in understanding the evolution of the monsoon, the longitudinal circulations such as the Walker Circulation, and also the interannual variability of the Pacific Ocean.

The magnitudes of the land surface temperatures¹³ are clearly more extreme than ocean temperatures at similar latitudes at all times of the year. Maxima are found in the deserts of North Africa and West Asia during all seasons except DJF when only North and Central Australia have mean temperatures that exceed 36∘C. The principal reason for the greater extremes of land temperatures is the inability of the land surface to store heat at depth, unlike the ocean, which has the capability to transport heat downwards through the upper ocean wind-induced turbulence. Consequently, heat may accumulate below the surface, and on annual time scales introduces a lag between the phase of solar forcing and maximum SST (Figure 1.5). Land temper- atures, on the other hand, especially away from coasts, tend to follow the seasonal migration of solar radiation.

However, in regions of significant precipitation, land temperatures are cooler than in more arid regions at similar latitudes.

Large ocean–land surface temperature contrasts occur in many regions of the planet, especially in the summer hemispheres. These differences prompted Hal- ley’s differential buoyancy theory of the monsoons, noted earlier, whereby air parcels over heated land are more buoyant than their counterparts over the cooler ocean. However, it should be pointed out here that while ocean–land temperature differences may be important they may not be a sufficient condition to create a large-scale monsoon circulation. For example, the temperature differences in the western Arabian Sea

13 Based on the Berkeley Earth Surface Temperature Project data (Rohde et al. 2013).

k k Figure 1.5 Mean seasonal surface

temperature(T_S(𝜑, 𝜆))for (a) JJA, (b) SON, (c) DJF, and (d) MAM between 45∘S and 45∘N. Bold contour over oceans enclose SSTs≥28∘C. The bold line over land signifies surface temperatures≥36∘C. Data from NCEP reanalysis based on Reynolds (1988, 2002). The contours over land represent the mean daily surface air temperatures averaged by season based on the Berkeley Earth Surface Temperature Project reconstructions (Rohde et al. 2013).

40°N 20°N Equ 20°S 40°S

40°N 20°N Equ 20°S 40°S

40°N 20°N Equ 20°S 40°S

40°N 20°N Equ 20°S 40°S

0° 60°E 120°E 180° 120°W 60°W 0°

0° 60°E 120°E 180° 120°W 60°W 0°

0° 60°E 120°E 180° 120°W 60°W 0°

0° 60°E 120°E 180° 120°W 60°W 0°

–30° –20° –10° 0° 10° 20° 30° 40° 50°

28° 36°

(a) JJA

(b) SON

(c) DJF

(d) MAM

Surface ocean and land temperatures

°C

and the Arabian Peninsula are close to the largest on the planet yet the land region remains very dry in summer and the winds are not on-shore, except for local sea breezes.

1.2.2.2 Precipitation

The mean seasonal distributions of rainfall rates (mm day⁻¹) are displayed in Figure 1.6a to d. The blue contour encloses a rain rate>8 mm day⁻1, correspond- ing to an annual rate of just under 3 m. There are some persistent features that appear throughout the year.

For example, there is a broad maximum oriented from northwest to southeast across the South Pacific (the South Pacific Convergence Zone: SPCZ) and, although

of lesser intensity, a similarly oriented band extend- ing across the South Atlantic from Brazil. The latter band is referred to as the South Atlantic Convergence Zone (SACZ). Satellite data available in the early 1970s allowed these bands to be identified for the first time as coherent climatological features that also migrated eastward during an El Niño.¹⁴

Another prominent feature is the narrow band of rainfall extending across the entire Pacific just north of the equator, often referred to as the Intertropical Convergence Zone (ITCZ). A similar feature, which also remains within the NH throughout the year, occurs in the tropical Atlantic.

14 E.g. Streten (1973).

k k

mm/day (a) JJA

(b) SON

Seasonal Precipitation Rate (mm/day)

40°N 20°N EQU 20°S 40°S

(c) DJF

(d) MAM

2 4 6 8 10 12 14 16 18 20 30 40

40°N 20°N EQU 20°S 40°S

40°N 20°N EQU 20°S 40°S

40°N 20°N EQU 20°S 40°S

0° 60°E 120°E 180°E 120°W 60°W 0°

0° 60°E 120°E 180°E 120°W 60°W 0°

0° 60°E 120°E 180°E 120°W 60°W 0°

0° 60°E 120°E 180°E 120°W 60°W 0°

Figure 1.6 Spatial distributions of the precipitation rate,P(𝜆, 𝜑), for (a) JJA, (b) SON, (c) DJF, and (d) MAM fields (mm/day). Blue contour encloses precipitation rate>8 mm day⁻¹. Dashed line represents the equator. Source: Data sets from the NOAA Climate Prediction Center MORPHing technique (CMORPH, Joyce et al. 2004).

There are several interesting seasonal rainfall distributions:

1. Boreal summer JJA (Figure 1.6a): The NH rain- fall rates are at their most intense. There are high rainfall rates in the warm pool region of the west- ern Pacific but the largest occurs in the Northern BoB (16–18 mm day⁻¹), totaling nearly 2 m in JJA.

Whereas this center of precipitation is associated with the South Asian Monsoon, the maximum rain- fall is not over land; in fact, it decreases in intensity northwest along the Ganges Valley. Heavy rainfall

can be found in Southeast Asia coastal regions but with decreasing intensity through Northern China. A secondary maximum occurs to the west of Sumatra in the near-equatorial SH. There is also substantial rainfall across Central and West Africa, the latter associated with the West African Monsoon when the Pacific and the Atlantic ITCZ are at their maximum intensity. In the eastern Pacific the ITCZ is located at its most poleward position.

2. Boreal autumn SON (Figure 1.6b): Rainfall rate maxima can be found in equatorial Africa and South America associated with the first of the two

k k equinoctial rainfall periods. The Pacific Ocean

ITCZ remains strong in the NH, similar to the location of the Atlantic ITCZ. These NH features are sufficient to explain the location of the zonally averaged maximum apparent in Figure 1.3. Dur- ing SON, especially at the latter end of the period, the Indian Ocean coast of Africa receives substan- tial rainfall. These are referred to as the “short rains.”

3. Austral summer DJF(Figure 1.6c): The locus of max- imum rainfall has moved south of the equator with extensive rainfall over southern Africa, North Aus- tralia, and Brazil. The SPCZ remains a prominent precipitation feature throughout the year but during DJF it attains its most intense state. The SACZ also reaches its annual intensity maximum. The eastern Pacific ITCZ, and its weaker Atlantic counterpart, still remain in the NH.

4. Austral autumn MAM (Figure 1.6d): The second equinoctial maximum occurs over equatorial Africa, Indonesia, and equatorial South America. Rainfall also occurs along the Indian Ocean coast of equa- torial East Africa. These are the “long rains.” The Pacific ITCZ still remains in the NH, although at its weakest state of the year but just strong enough to help explain the NH relative maximum in the MAM zonally averaged rainfall rate (Figure 1.3). There is an indication of a very weak ITCZ just south of the equator in the Eastern Pacific Ocean, suggesting a double ITCZ. For future reference, this weak band occurs at the time of not only the warmest eastern Pacific SSTs but also when the cross-equatorial SST gradient is smallest.

Finally, it is worth reiterating the observation of near interhemispheric symmetry of the total rainfall both annually and seasonally appearing in Figure 1.3.

Considering the very different geography of the two hemispheres and the range of types of rainfall that occur, it is a most surprising result. The rainfall results are reminiscent of the findings of Stephens et al. (2015), which showed that the difference between hemispheres of the annual integrated net radiation at the top of the atmosphere (TOA) is near zero.

1.2.2.3 Surface Pressure

The distributions of mean sea-level pressure (MSLP hPa: colored background) for each of the four seasons are plotted in Figure 1.7a–d. In the simplest sense, the tropics are a zone of low pressure between the sub- tropical high-pressure systems. The minimum pressure meanders through the summer tropics. This is often called the equatorialornear-equatorial trough, or, over South Asia, the monsoon trough. This low-pressure

band, though, is not always linked to precipitation.

The surface trough passes through the great subtrop- ical desert regions where at the time of minimum MSLP and maximum temperature there is little, if any, precipitation. The desert troughs are made up of thermally induced low-pressure systems. In summary, the near-equatorial oceanic trough is collocated with either a maximum in SST or surface land temperatures but not necessarily with maximum wind convergence or precipitation. The statistics of the relative locations of MSLP, SST, precipitation, and wind convergence are explored below.

The bands of subtropical anticyclones, located at roughly 30∘of latitude on either side of the equator, are quasi-permanent features. These high-pressure zones appear near the region of descent of the Hadley cells (Figure 1.4), altered somewhat by the distribution of land and ocean and time of year. During the boreal summer (Figure 1.7a), the location of the high-pressure centers is determined by the relatively colder oceanic temperatures and by the relative warmth of the land- masses. In fact, the entire Eurasian landmass is marked by anomalously low pressure, promoting onshore mon- soon flow. This is referred to as the Asiatic Low. During the boreal autumn (Figure 1.7b), pressures tend to rise over Eurasia as insolation decreases. At the same time, surface pressures start to fall over equatorial Africa and South America, creating convergence and rainfall (Figure 1.7b). The surface pressure over North Australia has started to fall and reaches a minimum in the boreal winter (Figure 1.7c), promoting the north Australian monsoon.

During the SON and DJF (Figure 1.7b and c), NH SSTs are generally warmer than the land surface temper- ature of the land. The strength of the ocean anticyclones decreases and shifts closer to or over land and may be replaced, as in the North Pacific, by a low-pressure sys- tem. We note, too, that the high-pressure system in the North Atlantic (often called the Bermuda High) shifts eastward to the vicinity of the Azores Islands during the boreal winter (Figure 1.7c), where it assumes a new sea- sonal name, theAzores High. The winter high-pressure system extending over the northeastern part of Eurasia (theSiberian High) becomes the strongest of all conti- nental high-pressure systems and is the result of intense radiational cooling to space. The winds associated with the eastern branch of the Siberian High flow south- ward across the South China Sea to form the winter monsoon.

In general, the locations of the SH subtropical high-pressure systems remain much the same through- out the year, varying little in longitude and latitude. The southeast trade winds on their northeast flanks of the

k k

990 994 998 1002 1006 1010 1014 1020 1024 1028 1032 1034 1038 hPa

(c) DJF

(d) MAM (a) JJA

(b) SON

Mean sea-level pressure and surface winds

60°N 45°N 30°N

Equ 15°N

15°S 30°S 60°N 45°N 30°N

Equ 15°N

15°S 30°S

60°N 45°N 30°N

Equ 15°N

15°S 30°S 60°N 45°N 30°N

Equ 15°N

15°S

0° 60°E 120°E 180° 120°W 60°W 0°

30°S

10 m s^–1

Figure 1.7 (a) Mean sea level pressure (MSLP: shading relative to color bar: hPa) between 60∘N and 40∘S for (a) JJA, (b) SON, (c) DJF, and (d) MAM. Warm colors denote relative high pressure while cool colors denote relative low pressure.

The vectors represent the 925 hPa wind field relative to the vector scale.

Source: Data from NCEP reanalysis.

anticyclones continue throughout the year. The differ- ence in the steadiness of the pressure regions between hemispheres is perhaps determined by the smaller per- centage area of landmasses in the SH such that the positions of the high-pressure systems are determined mostly by the SST distribution.

1.2.2.4 Wind Fields

The vectors in Figure 1.7 represent the climatological seasonal near-surface wind fields. Equatorward of the NH anticyclones are the northeast trades that converge toward the oceanic ITCZ. Similarly, equatorward of the SH high-pressure systems are the southeast trade winds.

k k The trade winds¹⁵ are the steadiest surface winds on

the planet and during their trek toward lower latitudes gather water vapor through evaporation, which will be eventually condensed through ascent nearer to the equator. In effect, the trade winds act as vast “solar collectors.” As the SST warms toward the equator, the saturated vapor pressure rises, allowing increasing water vapor transport as the trade winds converge.

Equatorward of the trades are relatively calm regions known as the doldrums. The trades merge together into the equatorial trough region – a region marked by storminess and strong convective activity, but also by heat, humidity, and uncertain wind.

Winds also converge into the tropical landmasses of the summer hemispheres, producing the monsoon circulations of South and East Asia, West Africa, North Australia, and South America. This conver- gence is clearly evident in JJA in the Indian Ocean, where a strong cross-equatorial gyre “feeds” the Asian monsoon. During DJF, similar cross-equatorial flows, although not as strong, occur in the Atlantic Ocean into South America and across Indonesia into North Australia. A more detailed view of the regional monsoons is presented below. The trade winds and monsoon flow are important components of the cou- pled ocean–atmosphere system. These tropical wind systems drive ocean currents that are also significant transporters of heat.

Substantial spatial variability in winds also occurs in the upper troposphere. The 200 hPa horizontal winds are displayed in Figure 1.8a to d. The blue contour separates easterlies from westerlies, thus marking the U(𝜆, 𝜑) =0 contour. Compared to the zonally aver- aged wind plotted in Figure 1.4 these fields are much more complex. There are a number of notable features.

Westerly jetstreams (marked as “W”) dominate the subtropics and extratropics. During SON and DJF the jetstreams tend to move equatorward and poleward during MAM and JJA. Downstream of the jets are regions of westerlies that extend between the extrat- ropics and the equatorial regions. These are referred to as “westerly” ducts(WDs). During JJA, a band of easterlies extends spasmodically around the planet. The strongest easterlies occur to the south of Asia. This is referred to as the monsoonal easterly jet (E), which is part of the monsoon gyre forced by the monsoonal heating over South Asia. The easterly jet discovered by

15 It is generally thought that the name “trade winds” had a commercial origin because of their use by early traders crossing the Atlantic Ocean for the Americas. However, Philander (1996) suggests that the name has a nautical origin reflecting the steadiness of the winds coming from the word “tread,” which refers to the steady path of a ship’s progress, probably emerging from the Middle English word “trade” meaning “path” or “track.”

Koteswaram (1958) is a quasi-geostrophic easterly wind maximum jet stream overlying southern Asia in the high troposphere (∼100 hPa) with a core near 15∘N. Looking downstream along the jet, temperatures decrease from right to left across the current. During SON, the easterly jet decreases in magnitude and distinct zones of west- erlies appear over the equatorial Pacific and Atlantic oceans. These regions of westerlies continue through DJF and MAM.

The upper tropospheric equatorial westerlies are par- ticularly strong over the central and eastern Pacific dur- ing the boreal winter. This is a region referred to as the WD, a zone where high-amplitude extratropical distur- bances appear to propagate into the tropics.¹⁶In addi- tion, principally in the boreal summer, the westerly duct is replaced by the Tropical Upper-Tropospheric Trough (or TUTT), marked by “TT” in Figure 1.9. The term TUTT was introduced in Sadler (1976), a dynamic fea- ture he hypothesized to be associated with the genesis of typhoons. Hanley et al. (2001) provide an extensive study of the general association of tropical cyclones and TUTTs.

1.2.2.5 Moisture Flux

The horizontal moisture transport in an atmospheric column across the tropics and subtropics is character- ized by the vertically integrated moisture transportB_q, defined as:

B_q(x,y) =

∫

TOA

0

q_v(𝜑, 𝜆,z)V(𝜑, 𝜆,̃ z)dz (1.2) where q_v(𝜑, 𝜆, z) and V(𝜑, 𝜆,̃ z) represent the distri- butions of the time-averaged specific humidity and horizontal velocity within a column, respectively.

“TOA” refers to the“top of the atmosphere”or where p → 0.B_q is strongly weighted toward the lower tro- posphere, as specific humidity decreases rapidly with height, as clearly shown in Figure 1.2. Both the boreal summer and winter distributions of the moisture flux vector are depicted with colored shading denoting mag- nitude. A comparison with Figure 1.6 shows that the divergence of water vapor flux is associated with regions of(P–E)>0 and convergence with(P–E)<0, whereP is precipitation andEis surface evaporation. The largest values of water vapor flux appear in the northern Indian Ocean near South Asia during the boreal summer. This arises from the accumulation of water vapor, collected by the southeast trades south of the equator and across the Arabian Sea and then condensing over India and the BoB. A divergent water vapor flux (P>E) exists to the east of Indonesia, feeding the warm pool precipitation maximum. In the Atlantic Ocean, theB_emaximum, far

16 Webster and Holton (1982), Tomas and Webster (1994), and Tomas et al. (1999).

k k 60°N

45°N 30°N

Equ 15°N 15°S 30°S 60°N 45°N 30°N Equ 15°N

15°S 30°S

60°N 45°N 30°N

Equ 15°N

15°S 30°S 60°N 45°N 30°N

Equ 15°N 15°S

0° 60°E 120°E 180° 120°W 60°W 0°

30°S

200 hPa wind field

(a) JJA

(b) SON

(c) DJF

(d) MAM

0 10 20 30 40 50

m s^–1 W

W W

W

W TT

WD

WD

W

W E

W

WD

Figure 1.8 Same as Figure 1.8 but for 200 hPa horizontal wind fields for (a) JJA, (b) SON, (c) DJF, and (d) MAM.

Bold blue line separates easterlies from westerlies. “W”, “WD”, and “TT”

denote westerly maxima, westerly ducts, and TUTTs, respectively. “E”

refers to the location of the boreal summer easterly jet stream to the south of India. Magnitude of the wind field is color coded relative to the bar. Source: Data from NCEP reanalysis (Kalnay et al. 1996).

weaker in both the boreal summer and winter, appears to be associated with the precipitation maxima in the Carribean and nothern South America, respectively.

1.2.3 Variability Along the Equator

Here, we now examine the seasonal structure in the longitudinal-height plane between 5∘N and 5∘S.

1.2.3.1 Temperature and Moisture

Figure 1.10a and b plots the equatorial sections of T(𝜆,z) and q_v(𝜆,z) for the boreal summer and win- ter, respectively. Both fields are remarkably flat. The

“flatness” of theT(𝜆,z)field, and other thermodynamic fields, created considerable discussion in the 1960s and early 1970s.

k k

100°W 50°W 0° 50°E 100°E 150°E 160°W

40°S 20°S EQU 20°N 40°N ¹⁰⁰

100

100 100

10 0

100

100 100

100 100

100

100 300

300

100°W 50°W 0° 50°E 100°E 150°E 160°W

40°S 20°S EQU 20°N

40°N 100 100

100 100

100

100 100

100

100

100

100

100 100 100

300 100

(b) DJF (a) JJA

10³ kg m s^–1

kg m s^–1

0 100 200 300 400 500 600

Vertically Integrated Moisture Flux B_e (λ,φ)

Figure 1.9 Distribution of mean vertically integrated moisture transport,B_q(𝜆, 𝜑), from Eq. (1.2) for the period (a) June–September (JJA) and (b) December–February (DJF) relative to the vector key. Viewed in the context of moisture transport, the Asian–Australian monsoon system appears in both the boreal summer and the boreal winter as strong interhemispheric systems with moisture sources clearly defined in the trade wind regimes of the winter hemisphere. The African summer and winter monsoons are less clearly defined in terms of moisture transport and are similar in magnitude to the North Australian summer monsoon. Weak moisture fluxes into northwest Africa may be seen, for example, but the region is dominated by strong westward moisture fluxes associated with the trade wind across the Atlantic and into the Americas. Source: Adapted from Webster and Fasullo (2002).

Figure 1.10 does offer a surprise, though. The specific humidity fields have a much more longitudinal struc- ture than the temperature fields. Gradients of specific humidity can only be explained if there are dynamic circulations in the longitude–height plane. In fact, the maxima inq_v(𝜆,z)correspond to the ascending regions of the Walker Circulations and the minima to regions of descent.

1.2.3.2 Wind Fields

The structure of the zonal and meridional wind fields along the equator (Figure 1.11a and b) is far more complex than the corresponding temperature fields of Figure 1.4. In particular, they do not resemble the zonally averaged wind fields appearing in Figure 1.4 at all. For example, Figure 1.4a(i) shows that the zon- ally averaged JJA upper-tropospheric zonal wind field is easterly with magnitudes of 5–10 m s⁻¹. The correspond- ing fields in DJF have magnitudes<5 m s⁻¹. In contrast, however, the JJAU(𝜆,z)-field (Figure 1.11a(i)) possesses a region of strong upper tropospheric easterlies with

magnitudes of about 10–15 m s⁻¹ that dominate the eastern hemisphere. In the western hemisphere, though, there are moderate westerly winds (5 m s⁻¹). Together, they sum to the weak easterly zonally averaged observed winds of Figure 1.4a.

During DJF, the overall pattern along the equator is similar to the summer although the intensity of the upper-tropospheric westerlies over the tropical Pacific, and also over the Atlantic, have larger amplitudes>10 m s⁻¹. The zonal average of these winds results in weak upper tropospheric easterlies (Figure 1.4c). During both seasons the lower tropospheric zonal winds are out of phase with those in the upper troposphere. During both JJA and DJF westerlies extend through the lower and middle troposphere of the eastern hemisphere, reversing in sign in the western hemisphere.

Figure 1.11b indicates that, generally, the longitudi- nal structure of meridional winds (V(𝜆,z)) follows the mean distribution portrayed in Figure 1.4a and c. Dur- ing the boreal summer, the meridional wind component is toward the south in the upper troposphere. In the

k k 1000

800 600 400 200 100

300

500

700

900

p(hPa)

1000 800 600 400 200 100

300

500

700

900

p(hPa)

(a) JJA

T(^oC)

0 60E 120E 180 120W 90W 0

0 60E 120E 180 120W 90W 0

(b) DJF

Temperature and specific humidity (5°N-5°S)

longitude

–80 –70 –60 –50 –40 –30 –20 –10 0 10 20 30

Figure 1.10 Longitudinal-height section of temperatureT(𝜆,p)(∘C, color bar) and specific humidity q_v(𝜆,p)(10⁻³kg kg^–1contours) averaged between 5∘S and 5∘N for (a) JJA and (b) DJF. Near the surface, the isotherms tend to follow the SST closely. Higher in the troposphere, the longitudinal temperature gradient is almost flat whereas the specific humidity fields show greater longitudinal variation. Source: Data from NCEP reanalysis but we note that there is no moisture data available above 300 hPa.

boreal winter there is a distinct reversal in sign. During both seasons, the upper tropospheric meridional winds are out-of-phase with the lower troposphere. Yet there is still considerable variability in longitude to suggest that the mean zonally averaged meridional cells are a composite of a number of smaller scale meridional cells. These cells may be associated with equatorial orographic features. For example, maxima (northward in JJA and southward in DJF) are collocated with the East African Highlands 40∘E–60∘E, the islands of the Indonesian Archipelago 90∘E–130∘E, and the Andes 120∘W.

1.3 Macro-Scale Circulations

The circulations of the tropical atmosphere and ocean are characterized by a wide range of spatial scales pos- sessing different periods of variability. Here we describe

seasonal characteristics of the largest of these phenom- ena: the Hadley, the Walker, and the monsoon circu- lations, as well as major climatological features of the tropical oceans.

1.3.1 Hadley’s Circulation

A meteorologicalcause celebre during the seventeenth and eighteenth centuries was how to explain why the trade winds moved from east to west, opposite to the rotation of the planet. This problem occupied some of the great minds of the time including Johannes Kepler and Galileo Galilei. Halley (1686) argued that the apparent westward movement of the solar heating, as Earth rotated about its axis, created a westward moving heat source that induced buoyancy and rising motion at the location of the greatest solar heating. The apparent westward movement of the differential buoyancy was thought to produce westward winds as the sun moved to