1. INTRODUCTION

Of all the atmospheric circulation systems, monsoons exhibit the most significant seasonal variations. They are a central component of the global climate system and are large enough to influence it (WCRP 2009). Monsoon anomalies may cause droughts, floods, and other extreme weather or climate events. Monsoon circulations are also critical to the global transport of atmospheric energy and water vapor. Monsoon variability has profound societal and economic influences on the regions they affect, where more than 70% of the world's population lives. Thus, monsoon research is essential not only for understanding global atmospheric circulation and climate change, but also for preventing and mitigating disaster and achieving sustainable development.

The concept of the monsoon is a long-standing one in climatology, and both the awareness and the observation of monsoons have a long history. In English, the term “monsoon” is derived from an Arabic word with several possible spellings (e.g., mausim, mausam, mausem, mawsim, mausin) (Dash 2005), or from the Malayan monsin (Pédelaborde 1963); all these words mean “season.” The existence of the monsoon was recognized long before it was given this name. In the twenty-third or twenty-second century bc, the Chinese emperor Shun wrote a poem entitled “Southerly Wind”:

This is the earliest written record that implies the main features of the East Asian summer monsoon and its importance for the people's livelihood (Zeng 2005). In addition, a poem entitled “Northerly Wind” in the earliest Chinese poetry anthology, Shi Jing (English translation: Book of Odes), reads,

This is the earliest text that describes the typical features of the East Asian winter monsoon. Thus, as early as 3,000 years ago, the Chinese were already aware of the East Asian summer and winter monsoons.

In the Middle Ages, people began to document their observations of the monsoon. Arab navigators found alternate prevailing southwesterlies from April to September and northeasterlies from October to March in the oceans between India and East Africa. In 1554, an Arab author, Sidi-Ali, recorded the dates of monsoon onset and retreat at fifty stations in the Indian Ocean (Pédelaborde 1963).

Halley (1686) was the first to propose a theoretical explanation of the Asian monsoon, in terms of surface thermal mechanics. This classic view regards the monsoon as an enormous sea breeze. Hadley (1735) modified Halley's monsoon model by considering the Coriolis force. Voyeikov (1879) linked seasonal precipitation variations to the monsoon. Subsequently, others set out similar views. Particularly, Monsoons, edited by Fein & Stephens (1987), provided an overview that included a summary of modern monsoon theory, by Webster; a historical review, by Singh; a review of monsoon physics, by G. Kutzbach; and an assessment of the possible causative role of the northern summer perihelion, by J. Kutzbach. The paleomonsoon has also become a subject of active interest (Fairbridge 1986). Paleomonsoon variability has been studied extensively using marine and terrestrial records as well as numerical modeling—these studies cover, for example, the Indian monsoon (Kutzbach & Otto-Bliesner 1982, Prell 1984, Kutzbach et al. 1989, Prell & Kutzbach 1992, Clemens et al. 1996, Fleitmann et al. 2003, An et al. 2011), the East Asian monsoon (An et al. 1990, 1991a, 2000, 2001; Liu & Ding 1998; Wang et al. 2008c), and African paleoclimate change (Kutzbach 1980, 1981; Sarnthein et al. 1981; Rossignol-Strick 1983; deMenocal 1995; Kutzbach & Liu 1997). In 2007, the Past Global Changes (PAGES) project established a global monsoon working group to bring the modern global monsoon concept to the paleoclimate community (Wang et al. 2012).

The traditional concept of monsoon dynamics is that the difference in the heat capacities of the land and the upper ocean leads to a seasonal wind reversal caused by greatly amplified temperature change over land compared with over ocean. The traditional monsoon concept is regional, and it refers to certain areas in the tropics that exhibit significant seasonal variation in circulation and precipitation. Recently, however, this concept has been extended to the global scale.

Sankar-Rao (1966, 1970) was the first to use the term “global monsoon”; he explored the impact of land-sea thermal contrast and topography on global monsoon circulation. Charney (1969) noted that the monsoon is driven by a seasonal migration of the intertropical convergence zone (ITCZ). Sikka & Gadgil (1980), Chao & Chen (2001), Gadgil (2003), and Wang (2009) also supported this hypothesis. Hoskins & Rodwell (1995) investigated the Asian summer monsoon in a global context, as a dominant feature of global circulation. Trenberth et al. (2000) defined the global monsoon in terms of a persistent global-scale seasonal overturning of atmospheric circulations throughout the tropics and subtropics. Qian (2000) discussed alternating dry/wet states and their relation to a global-scale monsoon between 40°S and 40°N. Wang & Ding (2006) discussed the global monsoon precipitation domain and defined the global monsoon as a dominant mode of annual variation in the tropics (Wang & Ding 2008, Chang et al. 2011), and Nie et al. (2010) investigated monsoon variations by employing a convective quasi-equilibrium for the diabatic effects of moist convection as well as the regional differences. However, their work did not include extratropical monsoons.

Flohn (1951) proposed a new interpretation of the monsoon as a normal shift in planetary circulation zones. In this framework, some tropical regions (e.g., India, Indo-China, and South China), some subtropical regions (e.g., California, Maghreb, and South Africa), and some frigid regions (e.g., the Siberian coastlands, Alaska, northern Canada, Iceland, and northern Norway) can be considered to be monsoonal regions. However, the frigid monsoon has not yet been fully recognized.

Since 1970s, a series of studies has been published on the dynamics of the axisymmetric Hadley circulation (Schneider & Lindzen 1977, Held & Hou 1980, Schneider 1987). The results of these studies demonstrated that in response to axisymmetric diabatic heating, the atmospheric circulation adopts two distinct regimes: thermal equilibrium and angular momentum conservation. In the tropics, where the planetary vorticity is small and the Rossby radius of deformation (Pierrehumbert et al. 2011) is large, a weak forcing may overcome the in situ planetary vorticity and produce a meridional circulation, forming a nonlinear angular momentum conservation regime. However, in the extratropics, where opposite conditions appear, a weak forcing may not generate meridional circulation, and the temperature is in local thermal equilibrium. Lindzen & Hou (1988) demonstrated that if the heating maximum is shifted a few degrees from the equator, the forced Hadley circulation becomes asymmetric about the equator, causing the dramatic development of a rising arm in the summer hemisphere and a sinking arm in the winter hemisphere. This can be explained by the inertial unstable development of the ITCZ on the summer side of the equator (Tomas & Webster 1997, Tomas et al. 1999, Pierrehumbert 2000). Plumb & Hou (1992) further studied the axisymmetric atmospheric response to an off-equator external forcing centered at 25°N and quantified the theory by identifying a threshold forcing for the transition from the thermal equilibrium to the angular momentum conservation regime. They demonstrated that for a forcing smaller than the threshold, the relative vorticity is weak, the absolute vorticity is determined by the planetary part, and the atmospheric response takes the regular linear thermal equilibrium regime. Conversely, for a forcing stronger than the threshold, the relative vorticity is strong, the absolute vorticity is determined by the relative vorticity part, and the atmospheric response takes the angular momentum conservation regime. Because the threshold is not large, the authors suggested that the angular momentum conservation regime could be relevant to real thermally forced tropical flows, most obviously tropical monsoonal flows.

Li & Zeng (2002) proposed a dynamical normalized seasonality (DNS) index to describe the characteristics of global monsoons and presented a three-dimensional distribution. This index is based on the intensity of the normalized seasonality of the wind field and can be used to depict the seasonal cycle and interannual variability of monsoons over different areas; the papers by Li & Zeng (2002) and Li et al. (2010) provide further details about the physical definition. Global monsoons include global tropical and subtropical monsoons at the surface, and global tropospheric and stratospheric monsoons in the vertical. The DNS index is applicable to any general monsoon circulation (Ellis et al. 2004). Li & Zeng (2005) specified that the global tropical monsoon lies between the boreal summer and winter locations of the ITCZ.

The paleoclimate community has also synthesized global monsoon variability from a geological perspective (An 2000, Wang 2009, Clemens et al. 2010, Ziegler et al. 2010b, Caley et al. 2011a, Cheng et al. 2012). Modeling and geological evidence suggest that a megamonsoon likely occurred during the Phanerozoic as a consequence of the Pangaean megacontinent (Kutzbach et al. 1989, Loope et al. 2001). Paleomonsoon evolution during the Cenozoic has been inferred from early Paleocene tropical rainforest fossils in Colorado (Johnson & Ellis 2002) and from the Eocene monsoon forest in central Australia (Greenwood 1996). The paleomonsoon system in East Asia was probably established during the Eocene and became persistent during the late Oligocene to early Miocene, with further development thereafter (Parrish et al. 1982, Zhou 1984, Gu & Renaut 1994, Sun & Wang 2005, Guo et al. 2008, Qiang et al. 2011, An et al. 2014). The paleomonsoon developed later in Africa and India (Kroon et al. 1991, Hilgen et al. 1995, Sepulchre et al. 2006). Wang et al. (2006) and Cheng et al. (2012) have used high-resolution oxygen isotope records from globally distributed speleothems to analyze orbital to millennial paleomonsoon variability on a global scale. In addition, Liu et al. (2003b, 2004) have studied the evolution of global monsoons in the Holocene using a coupled ocean-atmosphere model.

The study of monsoons has thus been enlarged from the local and regional scale to the global scale, from tropical regions to subtropical regions, and from the troposphere to the stratosphere. It has been extended to overlapping multiscale temporal interactions, and from a land-sea thermal contrast mechanism to a land-air-sea mechanism that incorporates interactions linked to the annual solar radiation cycle. Essentially, monsoon circulation is associated with the variation of large-scale pressure gradients in various monsoon regions. We have noticed that surface monsoons in different regions are a global phenomenon closely connected to the human environment. We refer here to the global monsoon, or global monsoons, as an integration of various regional monsoons (Wang et al. 2012) within a global context in terms of similar dynamics and behaviors. Therefore, we propose the following definition of the global monsoon:

The article is organized as follows: In Section 2, we outline the distribution and features of the global monsoon. In Section 3, we discuss multiscale monsoon variability from the orbital, millennial, and centennial to the interdecadal, interannual, and intraseasonal timescales. In Section 4, we emphasize the importance of the Tibetan Plateau for the formation and evolution of Asian monsoons. In Section 5, we summarize and provide perspective on the study of the global monsoon.

2. DISTRIBUTION AND FEATURES OF THE GLOBAL MONSOON

Here we adopt the DNS index of Li & Zeng (2002) to outline the distribution of the global monsoon (Figure 1) and to subdivide the global monsoon according to geographic regions. A large-scale monsoon index can be defined as the averaged DNS values over a monsoon domain. As shown in Figure 1, the tropical monsoon essentially lies between the seasonal migration boundaries of the ITCZ (Li & Zeng 2005, Wang 2009). Under the annual cycle of solar radiation, the seasonal migration of the ITCZ, caused by cross-equatorial pressure gradients, produces a strong tropical monsoon (Tomas & Webster 1997, Zeng & Li 2002, Webster & Fasullo 2003). The global tropical monsoon is distributed primarily across the following regions: tropical Asia, Australia, and Africa (Khromov 1957, Ramage 1971); South America (Zhou & Lau 1998); and the tropical eastern Pacific (as well as Central America) (Lau 2003, Li & Zeng 2003). The tropical Asian-Australian monsoon is the archetypal monsoon system and consists of the following main subsystems: the Indian or South Asian monsoon (Goswami et al. 2003), the South China Sea monsoon (Ding & Liu 2001, Feng & Li 2009), the Indochina Peninsula and Western North Pacific monsoon (Tao & Chen 1987, Wang et al. 2008a), the tropical Australian monsoon, and the Maritime Continent monsoon (Webster et al. 1998).

Subtropical monsoons in the Southern and Northern Hemispheres are caused by the seasonal shift of the subtropical high and land-sea distribution and are closely related to large-scale topography, the Rossby radius of deformation, the jet stream, and the interaction between the jet stream and large-scale topography (Molnar et al. 2010). The Northern Hemisphere subtropical monsoon consists of the East Asian (Tao & Chen 1987), North American (Douglas et al. 1993), North African (Khromov 1957, Pédelaborde 1963), and Tibetan Plateau monsoons (Tang & Reiter 1984), as well as the subtropical North Atlantic and North Pacific monsoons (Li & Zeng 2003). The Southern Hemisphere subtropical monsoon includes the Southern Australian (Feng et al. 2010), South African (Khromov 1957, Pédelaborde 1963), and subtropical South Pacific monsoons (Li & Zeng 2003).

Global monsoon systems show a distinct three-dimensional vertical structure that has not been widely considered in the literature. The vertical structure is characterized by significant baroclinicity as well as an obvious asymmetry between the Northern and Southern Hemispheres due to differences in land-sea distribution. In the lower troposphere, the global monsoons include the tropical monsoon and the Southern and Northern Hemisphere subtropical monsoons. In the middle and upper troposphere, only the Southern and Northern Hemisphere subtropical monsoon systems exist, and their extent within the seasonal shift of the subtropical high ridges increases with altitude (see figures 2–4 in Li & Zeng 2000 and figures 3–5 in Li & Zeng 2005). A stratospheric monsoon exists in both the Northern and Southern Hemisphere extratropical regions (Li & Zeng 2000).

Figure 2 illustrates the distributions of semipermanent pressure systems and prevailing circulations in winter and summer, as well as lower tropospheric monsoonal and nonmonsoonal regions. The southwesterly and southeasterly monsoons converge in the Asian-Australian monsoon region (0°–20°N, 140°E–20°W) within the Northern Hemisphere in June, July, and August (JJA) and become a northeasterly wind in December, January, and February (DJF), with prevailing northwesterlies in the Southern Hemisphere (0°–10°S, 20°E–160°E). During JJA, the prevailing southeasterlies from Africa, the South Indian Ocean, and Australia (20°S–0°, 20°W–140°E) turn right after crossing the equator and become southwesterlies in tropical North Africa and South Asia. During DJF, the prevailing winds in the above-listed Northern and Southern Hemisphere monsoonal areas reverse. The low-level winds in the subtropical and frigid regions also experience seasonal reversals (Figure 2). This feature implies that either the natures of the semipermanent pressure systems associated with these monsoon systems reverse (i.e., a high-pressure system turns into a low-pressure system, and vice versa), or their positions shift considerably.

The seasonal variation in rainfall over the monsoon regions is related to variations in monsoon circulation. To clarify this relationship, we define a normalized seasonal dry-wet index (SDWI), as follows:

where R_W and R_D indicate the amounts of precipitation in wet and dry seasons, respectively. To simplify, R_W (R_D) here is taken as the seasonal averaged rainfall in JJA (DJF) in the Northern Hemisphere and DJF (JJA) in the Southern Hemisphere. The SDWI, unlike the monsoon precipitation index defined by Wang & Ding (2008), extends the viewpoint to the global and emphasizes the seasonality of rainfall over the monsoonal regions. Domains with positive SDWI occur when the wet season rainfall is at least three times the dry season rainfall, emphasizing the significant seasonal variation of precipitation between dry and wet conditions. This index thus highlights strong seasonality and rainfall concentration. Positive SDWI values in JJA produce patterns that follow most of the Northern Hemisphere monsoon regions and Southern Hemisphere subtropical regions; positive SDWI values in DJF produce patterns that follow most of the Southern Hemisphere tropical monsoon regions and parts of the Northern Hemisphere subtropical regions (Figure 2). Overall, the regions with significant seasonal rainfall variation correspond well with the monsoon regions shown in Figure 1. Water vapor delivered by monsoon circulation originates primarily from tropical oceans, providing a potential global monsoon link.

3. MULTISCALE MONSOON VARIABILITY

There is a wide range of timescales in monsoon variability. Here we cover monsoon variability on timescales ranging from orbital (10⁵∼10⁴ yr) to intraseasonal scales and address the common features among different monsoon regimes. Orbital to centennial (10² yr) variability is mainly inferred from proxies related to monsoonal wind and precipitation in geobiological archives, whereas interdecadal to intraseasonal oscillations are mainly derived from instrumental records. Because monsoons are traditionally described according to geographic associations (e.g., the Indian monsoon, the East Asian monsoon) in historical and published materials, we review monsoon variability from different geographic regions, with an emphasis on Asian and African monsoon areas.

3.1. Orbital-Scale Monsoon Variability Since the Pliocene

Orbital-scale monsoon variability has been identified from deep-sea, eolian, lacustrine, and speleothem records. Integrating terrestrial and marine monsoon records with modeling results can provide valuable insights into the dynamics of orbital-scale variability.

Plio-Pleistocene Indian summer monsoon variations have been reconstructed using biogenic and lithogenic indices from the Arabian Sea (Clemens et al. 1996). A long-term decrease in the lithogenic grain size fraction suggests that the persistent growth of the Northern Hemisphere ice sheets since 3.5 Ma has weakened the intensity of the Indian summer monsoon (Figure 3). Since 2.6 Ma, the timing of strong monsoons lagged behind the maximum global ice volume at the precession and obliquity bands by 83º and 124º, respectively. The variability of multiple proxies and associated internal phase relations indicates that the strength of the Indian summer monsoon is sensitive to orbital forcing and the extent of Northern Hemisphere glaciation (Clemens et al. 1996). The phases and timing of multiple monsoon proxies since 350 ka suggest that three mechanisms influence the timing of strong Indian summer monsoons within the precession and obliquity bands: (a) sensible heating of the Asian topography, (b) glacial boundary conditions, and (c) latent heat export from the southern subtropical Indian Ocean (Clemens & Prell 2003).

The phase of the Arabian Sea signals at the precession band was reevaluated by comparing multiple high-resolution productivity and oxygen minimal intensity records from the Arabian Sea with results from a transient climate modeling experiment (Clemens et al. 2010, Ziegler et al. 2010a, Caley et al. 2011b). Three independent proxies (bromine, foraminiferal assemblages, and lithogenic grain size) retrieved in the northern Arabian Sea demonstrated that the productivity and wind strength signals in the northern Arabian Sea were mainly controlled by Indian summer monsoon processes (Figure 4). A synthesis of the Indian summer monsoon variability demonstrated similar periodicities and phase lags in the precession and obliquity bands, implying that internal climate forcing might strongly modulate the timing of strong Indian summer monsoons within both the precession and obliquity cycles (Clemens et al. 2010, Caley et al. 2011b).

A high-resolution continental record from the Heqing paleolake indicates the importance of interhemispheric forcing in driving Indian summer monsoon variability at the glacial-interglacial timescale (An et al. 2011). The spectrum of the Heqing proxy records displays primary orbital periods associated with eccentricity (100 kyr) and precession (23 and 19 kyr) (Table 1). On the basis of phase and amplitude relationships between the Indian summer monsoon, benthic δ¹⁸O, and North Atlantic ice-rafted debris records, Indian summer monsoon variation can be divided into three intervals: 0.92–0.13 Ma, 1.82–0.92 Ma, and 2.60–1.82 Ma (Figure 3). During the older and younger intervals, the variations are closely linked to interactions between Northern and Southern Hemisphere dynamic processes, whereas during the middle interval, the northern low is dominant. Change in the Indian summer monsoon at glacial-interglacial timescales is driven by the relative dominance of both the northern low-pressure and southern high-pressure systems, which is necessary for understanding global monsoon dynamics (An et al. 2011, Caley et al. 2013).

East Asian summer monsoon variations have been reconstructed using land- and marine-based proxies (Figure 3) (e.g., An et al. 1990, 1991a,b; Liu & Ding 1998; An 2000; Wang et al. 2005b; Clemens et al. 2008). On the basis of magnetic susceptibility and carbonate content variations in Chinese loess–red clay sequences, the evolutionary process of the East Asian summer monsoon can be divided into three stages: (a) an initiation before 3.6 Ma; (b) strengthening between approximately 3.6 and 2.7 Ma; and (c) large-amplitude fluctuations since 2.7 Ma, followed by two large shifts at approximately 1.2 and 0.5 Ma (An et al. 1990, Xiao & An 1999, Sun et al. 2006b). Interpretation of the Pleistocene South China Sea proxies is complicated by the influences of both monsoon circulations and sea level change on sedimentary geochemistry (Wang et al. 2005a). However, numerous proxies have been employed to infer monsoon-induced oceanic changes (e.g., Wang et al. 1999a, 2003, 2005b). The results suggest a strong East Asian summer monsoon prior to 2.7 Ma, small-amplitude monsoon oscillations between 2.7 and 1.2 Ma, and large-amplitude fluctuations after 1.2 Ma (Clemens et al. 2008).

The climate shift at approximately 2.7 Ma was a global phenomenon, and it was particularly dramatic in the monsoonal regions. Both the phase relation and amplitude between different monsoon proxies shifted significantly around that time (An 2000, Ding et al. 2000, Clemens et al. 2008, Sun et al. 2010). At approximately 1.2 and 0.6 Ma, the East Asian summer monsoon was characterized by a two-stage strengthening, with a distinct shift of the dominant 41-kyr periodicity before 1.2 Ma to a dominant 100-kyr periodicity after 0.6 Ma (Liu et al. 1999). Because the East Asian summer monsoon climate is determined by the interactions between the global atmosphere, land, ocean, and ice systems within the specific geographic setting of East Asia, the mid-Pleistocene monsoon shift is likely linked to changing lower boundary conditions—for example, regional tectonic uplift (An et al. 1990, Xiao & An 1999), ice expansion in the high-latitude Northern Hemisphere (Clark et al. 2006, Raymo et al. 2006), and the strong asymmetry of hemispheric climates (Guo et al. 2009). Since 800 ka, a good correlation between loess proxies (magnetic susceptibility and grain size) and benthic δ¹⁸O records suggests that the East Asian monsoon has been strongly coupled to ice volume changes (Ding et al. 1995, Liu et al. 1999).

In contrast, absolute-dated oxygen isotope records from cave speleothems reveal a dominant 23-kyr cycle that lags summer insolation at 65ºN by 43º(Wang et al. 2008c, Cheng et al. 2009). The differing expression of orbital-scale monsoon variability between loess magnetic susceptibility and marine records relative to speleothem δ¹⁸O records likely reflects the sensitivity of the different archives to East Asian summer monsoon variation (Figure 4) (Clemens et al. 2010, Cheng et al. 2012). Whether loess magnetic susceptibility can record monsoon variability on a precession timescale largely depends on sedimentation rate and postdepositional processes that can attenuate the variability of the loess proxies (Feng et al. 2004, Sun et al. 2006a). Similarly, the effects of summer and winter precipitation, moisture source, and monsoonal wind on speleothem δ¹⁸O variability require further investigation (Wang et al. 2001b, Yuan et al. 2004, Clemens et al. 2010, Pausata et al. 2011, Cheng et al. 2012). Because of these spectral differences—in particular the lack of a distinct precessional signal in Chinese loess and of the 100- and 41-kyr cycles in speleothems—investigators must explore new proxies from high-resolution terrestrial and marine records and their sensitivities to East Asian summer monsoon intensity.

A characteristic sedimentary expression of changes in African monsoon intensity is the occurrence of cyclic organic-rich layers (sapropels) in the Eastern Mediterranean Sea (Rossignol-Strick 1983, Hilgen et al. 1995). Sapropels S1 to S10 are clearly identifiable in color reflectance records (i.e., percent reflectance at 540 nm) and elemental geochemistry (e.g., Wehausen & Brumsack 2000, Calvert & Fontugne 2001, Lourens et al. 2001, Ziegler et al. 2010b). Wind-borne dust records from deep-sea sediments of the subtropical Atlantic and Mediterranean Sea also provide evidence for monsoon-related Plio-Pleistocene African climate change (e.g., Tiedemann et al. 1994, deMenocal 1995, Larrasoaña et al. 2003).

Marine-based eolian records exhibit significant shifts in flux variability near 2.8, 1.7, and 1 Ma (Figure 3). Dust record spectra indicate that before 2.8 Ma, the subtropical African climate varied primarily at precessional periodicity. After 2.8 Ma, marked 41-kyr variability persisted until 1 Ma. The increase at 2.8 Ma in 41-kyr variability coincides with the onset of Northern Hemisphere glaciation, suggesting that the African monsoon was sensitive to remote changes in high-latitude climate (deMenocal & Rind 1993). This sensitivity became further amplified after 1.0 Ma and became characterized by large fluctuations in duration and amplitude of the eolian cycles at a 100-kyr period (deMenocal 1995, Trauth et al. 2009). High-resolution records of bulk sediment geochemistry from the eastern Mediterranean also show distinct 100- and 23-kyr periodicities (Figure 3, Table 1), implying combined influences of high- and low-latitude processes on African climate (deMenocal & Rind 1993, Ziegler et al. 2010b).

Phase and amplitude spectra of monsoonal proxies reveal that the East African monsoon follows a quasi-direct response to Northern Hemisphere summer insolation with dominant 100- and 23-kyr periodicities (Table 1), but the West African monsoon demonstrates distinct lags of several thousand years relative to precession minima and obliquity maxima, with similar weights of precession and obliquity signals (Weldeab et al. 2007a, Caley et al. 2011a). However, the South African monsoon variability inferred from a crater lake is out of phase with the North African monsoon change (Partridge et al. 1997). Transient modeling results that relied only on insolation conditions and on both insolation and ice conditions exhibited an unexpected highly linear response to insolation forcing (Tuenter et al. 2005, Kutzbach et al. 2008, Ziegler et al. 2010b). These simulations indicate that inclusion of variable global ice volume can affect the amplitude of African monsoon strength but has only a small impact in the precession phase (Ziegler et al. 2010b). Different time lags are likely related to the occurrence of precession-paced North Atlantic cold events, which delayed the onset timing of the West and East African monsoons to varying degrees (Weldeab et al. 2007a, Ziegler et al. 2010b).

The dynamics of orbital-scale monsoon variability can be attributed to changes in orbital parameters (eccentricity, obliquity, and precession), surface boundary conditions, and atmospheric CO₂ concentration (e.g., Kutzbach & Otto-Bliesner 1982; Prell & Kutzbach 1987, 1992). The orbital parameters can lead to seasonal changes in solar insolation received at Earth's surface (Berger 1978). The effects of orbital forcing on monsoon strength are due to changes in the amplitude of seasonal insolation and the resulting land-sea thermal and pressure contrasts (Kutzbach & Guetter 1986, Prell & Kutzbach 1987). Obliquity, which varies from 22.1° to 24.5° with a periodicity of approximately 41 kyr, can affect the seasonal variation of incoming insolation in both hemispheres, with a greater effect at high latitudes than at low latitudes (Milanković 1941). Precession, which reflects seasonal changes in the aphelion and perihelion with dominant periods of approximately 23 and 19 kyr, has opposite impacts on insolation changes between the Northern and Southern Hemispheres, with a more significant influence at low latitudes than at high latitudes, resulting in a significant interhemispheric contrast. Changes in net annual insolation can further influence high-latitude sea ice and tropical sea surface temperature (Kutzbach & Gallimore 1988). Therefore, changing obliquity and precession can affect monsoon variability through direct insolation forcing and indirect oceanic feedbacks (Liu et al. 2003b, 2004), as evidenced by distinct obliquity and precession cycles in geological records (Table 1). Although mean solar irradiation varies only slightly in response to changes in eccentricity, eccentricity can strongly modulate the precessional amplitude and therefore produce seasonal insolation changes. The eccentricity cycle may have been transmitted into paleomonsoon systems through the nonlinear response of low-latitude carbon cycles (Wang et al. 2003).

Surface boundary conditions and CO₂ changes also affect glacial-interglacial monsoon variability, especially at midlatitudes (Kutzbach & Guetter 1986). The Northern and Southern Hemisphere ice sheets act mainly through pressure and temperature systems, particularly through the cross-equatorial pressure and temperature gradients, to influence the strength of monsoon circulations (Tomas & Webster 1997, An et al. 2011). Variations in greenhouse gases (particularly CO₂) are strongly coupled with global temperature change on orbital timescales and can also influence monsoon variability by affecting temperature and pressure gradients (Felzer et al. 1998, Petit et al. 1999, An et al. 2011). Meanwhile, sea level changes associated with the changing land-ocean distribution and the waxing and waning of ice sheets can lead to shifts in the coastline and marine dynamical conditions. These factors can further influence monsoon variability by changing the land-ocean temperature/pressure contrast and heat/moisture transport from the ocean to the land. Our syntheses indicate that regional monsoons commonly have distinct responses to solar insolation, but that the impacts of obliquity and glacial cycles differ across various locations.

3.2. Millennial-Scale Monsoon Variability Since the Last Glaciation

Suborbital/millennial climate variability is characterized by abrupt changes within decades, followed by stable periods that persist for hundreds to thousands of years (Broecker et al. 1985, Cronin 2009, and references therein). Examples include the Younger Dryas event and Dansgaard–Oeschger cycles identified in the Greenland ice core records (Dansgaard et al. 1993), as well as ice-rafted debris events in the North Atlantic (Heinrich 1988). Impacts of these abrupt events on monsoon variability have been found in both hemispheres (Porter & An 1995, Schulz et al. 1998, Wang et al. 2001b, Altabet et al. 2002, Voelker 2002, Overpeck & Cole 2006, Pisias et al. 2010), with a near-simultaneous pattern in the Northern Hemisphere and opposite changes in the Southern Hemisphere (Augustin et al. 2004, Wolff et al. 2010).

Millennial Indian summer monsoon variability has been observed from multiple proxies studied in Arabian Sea sediments (Sirocko et al. 1993, Overpeck et al. 1996, Schulz et al. 1998, Altabet et al. 2002). These proxies reflect Indian summer monsoon–induced processes such as upwelling, surface productivity, and ventilation within the intermediate waters of intense oxygen-minimum zones. They are well correlated with Dansgaard–Oeschger cycles in Greenland ice cores, where weak monsoon events are associated with cold events in the Northern Hemisphere during the last glaciation (Figure 5). On the continent, numerous speleothem δ¹⁸O records show that dry conditions correlate with cold periods in Greenland (Neff et al. 2001, Burns et al. 2003, Fleitmann et al. 2003, Sinha et al. 2005, Cai et al. 2006). These marine and terrestrial records together yield a coherent picture of Indian summer monsoon behavior on millennial timescales during the late Pleistocene.

In East Asia, a strong teleconnection between the East Asian winter monsoon and the North Atlantic climate has been identified by correlation of loess grain size oscillations and ice-rafting events in the North Atlantic (Porter & An 1995, An & Porter 1997), implying that the northern westerlies play an important role in transmitting this temperature signal from the North Atlantic to the Asian monsoon regions. Subsequently, similar correlations have been established based on numerous loess-based monsoonal proxies (Xiao et al. 1995, Guo et al. 1996, Chen et al. 1997, Zhang et al. 1997, Ding et al. 1998, Fang et al. 1999, Wu et al. 2006). Millennial-scale summer monsoon variability has also been inferred from South China Sea sediment records (e.g., Wang et al. 1999b, Oppo & Sun 2005). Recently, integration of loess grain size and speleothem δ¹⁸O records with modeling results has suggested that the Atlantic Meridional Overturning Circulation (AMOC) is likely a driver of abrupt change in East Asian monsoon systems (Sun et al. 2012).

The well-dated speleothem δ¹⁸O record from Hulu cave, eastern China, provides a robust correlation of abrupt monsoon changes with Dansgaard–Oeschger cycles and Heinrich events in high-latitude regions during the last glaciation (Wang et al. 2001b), yielding a steady mechanical connection between North Atlantic events and Asian summer monsoon changes. This correlation—that is, weak East Asian summer monsoons coincident with cold periods in the North Atlantic, and vice versa—has been confirmed by other speleothem records from central and southern China (Figure 5) (Yuan et al. 2004, Dykoski et al. 2005, Cheng et al. 2006, Kelly et al. 2006, Wang et al. 2008c). Moreover, lake records reveal an antiphase relationship between the Asian summer monsoon and westerlies on millennial timescales, indicating the interplay of these two systems at Lake Qinghai (An et al. 2012).

The interpretation of the speleothem δ¹⁸O in China is still under debate. A few researchers have suggested that the cave δ¹⁸O in eastern China reflects changes in moisture transport and precipitation upstream over the Indian Ocean and the Indian summer monsoon region (Maher 2008, Dayem et al. 2010, Pausata et al. 2011), whereas more recent modeling results indicate that it does provide a proxy for East Asian summer monsoon intensity (Liu et al. 2014). It is not currently feasible to elucidate the relative contributions of local rainfall, various moisture sources, and distant fractionation and to what extent the temperature influences the speleothem δ¹⁸O. New records from key regions and sophisticated isotope-enabled global circulation model simulations will be necessary to inform further deliberations on this issue.

In South American monsoon regions, speleothem records also document millennial-scale changes, which are correlated with abrupt events in the North Atlantic but exhibit an antiphase relationship with speleothem δ¹⁸O records from Asian monsoon regions (Figure 5) (Wang et al. 2004, 2006; Cruz et al. 2005; Kanner et al. 2012; Cheng et al. 2013). This antiphase relationship indicates a close interaction between the Northern and Southern Hemispheres and suggests that a shift in the ITCZ may be responsible for the mirroring across hemispheres of abrupt events (Wang et al. 2004, 2006). In southwestern North America, millennial-scale climate changes during the late Pleistocene have been identified from speleothem δ¹⁸O records (Figure 5) (Asmerom et al. 2007, 2010; Wagner et al. 2010). Speleothem δ¹⁸O values decreased during cold stadials and increased during warm interstadials, however, due to the changing fraction of moisture from the Gulfs of Mexico and California (heavier δ¹⁸O during the summer monsoon season) relative to that from the North Pacific (lighter δ¹⁸O during winter) (Asmerom et al. 2010, Wagner et al. 2010), a pattern contrary to that in the Asian monsoon region.

Marine and terrestrial records have shown that in the northern Africa monsoon domain, monsoonal rainfall decreased during the Younger Dryas event, Heinrich event 1, and the intervals from 8.4 to 8 ka and from 4.2 to 4 ka, indicating the response of the northern African monsoon climate to temperature changes in the North Atlantic (deMenocal et al. 2000, Gasse 2000). Subsequently, a growing body of records has confirmed that the northern African monsoon tends to weaken at intervals corresponding to Dansgaard–Oeschger stadials and Holocene cold spells, resulting in drier and dustier conditions in the Sahel region (Figure 5) (Talbot et al. 2007; Weldeab et al. 2007a,b; Itambi et al. 2009; Niedermeyer et al. 2010; Zarriess & Mackensen 2010). In contrast, South Africa was wet, with significant seasonal variation during the stadials. This anticorrelation may be associated with a southward shift of the rainfall belt due to ITCZ migration during North Atlantic cold events, which in turn is linked with the bipolar seesaw response of the sea surface temperature to the AMOC (Garcin et al. 2007, Moernaut et al. 2010).

Because millennial-scale monsoon variability in various monsoon regions indicates significant coherence, and is well correlated with abrupt events in the North Atlantic region, the ocean thermohaline circulation hypothesis is widely accepted as the mechanism responsible for the teleconnection (Broecker et al. 1992, Alley et al. 1999, Clement & Peterson 2008, and references therein). Both observational and modeling studies suggest that freshwater input and/or a greater extent of sea ice in the North Atlantic may lead to significant changes in North Atlantic Deep Water and the AMOC, resulting in decreased temperatures across the northern high latitudes and a southward shift of the ITCZ (e.g., Zhang & Delworth 2005, Broccoli et al. 2006, Menviel et al. 2008, Zhang et al. 2010). In addition, sea ice changes over the North Atlantic regions may also play an important role in amplifying and transmitting the signal, and the resulting southward shift of the ITCZ, which in turn lead to weakening of the Asian summer monsoon and increased convective precipitation over Brazil and South Africa (Chiang & Bitz 2005). In Asian monsoon regions, it is also plausible that decreased temperatures across the northern high latitudes may lead to a strengthened East Asian winter monsoon and increased snow cover over the Tibetan Plateau (Barnett et al. 1989, Overpeck et al. 1996), which may in turn weaken the Asian summer monsoon in a coupled response.

Although the AMOC hypothesis can explain abrupt millennial-scale climate change and its global linkages, the recurrence of Dansgaard–Oeschger events approximately every 1,500 yr on average seems to be linked with changes in external insolation and internal interactions within the ocean-ice-climate system (Ghil et al. 1987, Maasch & Saltzman 1990, Mayewski et al. 1997). Modeling results suggest that Dansgaard–Oeschger cycles may initially be driven by periodic freshwater inputs into the North Atlantic Ocean induced by external solar activity (Braun et al. 2005, Clemens 2005, Li et al. 2005). Recently, Petersen et al. (2013) hypothesized that the abrupt warming at the onset of a Dansgaard–Oeschger cycle is caused by the rapid retreat of sea ice after the collapse of an ice shelf in the Nordic Sea, whereas gradual cooling during the subsequent interstadial phase is determined by the timescale of ice-shelf regrowth. In any case, the AMOC in the North Atlantic is most likely at the crux of these abrupt events, and land-air-sea interaction probably plays an important role in amplifying the impact of solar activity and modulating monsoon changes on millennial timescales.

Millennial-scale events may have differing characteristics within different monsoon regions. For example, Zhou et al. (2001) found that the Younger Dryas event in East Asia manifested globally in different ways, suggesting that El Niño–Southern Oscillation (ENSO)-like variability and other tropical factors may have been superimposed on this cold event. An (2000) proposed that strengthening of the southern trade winds and associated Southern Oscillation may increase the precipitation in East Asia through cross-equatorial flow. Speleothem δ¹⁸O records from southwestern China demonstrate that climate change in the Southern Hemisphere has an important influence on Asian summer monsoon variations on millennial timescales (Cai et al. 2006).

During the Holocene, the Asian summer monsoon has exhibited substantial oscillations on millennial to decadal timescales as well (Figure 6) (Neff et al. 2001; Fleitmann et al. 2003; Gupta et al. 2003, 2005; Dykoski et al. 2005; Wang et al. 2005b; Cai et al. 2012), with weak monsoons corresponding to cold events in the North Atlantic (Bond et al. 1997). An episodic slowdown of the AMOC, and the consequent strengthening of Northern Hemisphere circulations, is considered a likely cause for a southward ITCZ shift that results in changes in low-latitude monsoonal precipitation patterns (Barber et al. 1999, Murton et al. 2010, Yu et al. 2010). On the one hand, a multiproxy, annual-layered stalagmite from central China recently showed that the duration and evolution of the precipitation during the 8.2-ka event are indistinguishable from those observed in the Greenland ice cores, suggesting an effective and rapid atmospheric teleconnection exists between the North Atlantic and East Asian summer monsoon regions (Liu et al. 2013). On the other hand, the periodicities of solar activity and monsoon records are highly correlated with Δ¹⁴C records, suggesting that solar output changes are partly responsible for fine-scale monsoon variability in the Holocene (Figure 6) (Shindell et al. 2001; Fleitmann et al. 2003; Gupta et al. 2003, 2005; Wang et al. 2005b; Asmerom et al. 2007; An et al. 2012; Cai et al. 2012). However, other feedback processes may be involved to amplify solar output effect (Kodera 2004), as solar irradiation changes that occur on millennial to decadal timescales are very small (Vieira et al. 2011).

3.3. Centennial-Scale Monsoon Variability over the Past Millennium

The primary features of global climate during the past millennium include the most recent warm period (the Medieval Climate Anomaly, 800–1300), a cold period (the Little Ice Age, 1400–1850), and the subsequent warming (1850–present). Centennial and multidecadal variability of the global monsoon during the past millennium has been investigated extensively to ascertain characteristic patterns of natural versus anthropogenic forcings (Kumar et al. 1999, Verschuren et al. 2000, Russell & Johnson 2005, Stager et al. 2005, Newton et al. 2006, Tan et al. 2008, Liu et al. 2009, Sachs et al. 2009).

Many archives, including stalagmites, lake sediments, marine sediments, and historic documents, have been studied to understand centennial-scale hydrological changes in the Asian-Australian monsoon area over the past millennium (Figure 7). A large body of evidence suggests that the area located at the northern limit of the Asian-Australian summer monsoon region, including northern China (Tan et al. 2008, Zhang et al. 2008, Liu et al. 2011) and India (Sinha et al. 2011), experienced a drier climate during the Little Ice Age than during the Medieval Climate Anomaly. In contrast, areas in the southern part of the Asian-Australian monsoon region, including Indonesia (Newton et al. 2006, Oppo et al. 2009, Tierney et al. 2010), entered the Little Ice Age with a wetter climate. This antiphased variation between the Asian and Australian summer monsoons during the past millennium has been attributed to the southward migration of the ITCZ (Newton et al. 2006, Sachs et al. 2009, Tierney et al. 2010). However, some hydrological records from southern China (Chu et al. 2002, Tan et al. 2009, Yan et al. 2011, Zeng et al. 2011) (Figure 7) indicating a wetter Little Ice Age, along with the record from northern Australia (Wasson & Bayliss 2010) (Figure 7) indicating a drier Little Ice Age, are difficult to explain by the southward ITCZ migration and might be caused by other factors (e.g., Yan et al. 2011).

Recent studies have also highlighted multidecadal oscillations of the Asian summer monsoon during the last millennium (Zhang et al. 2008, Tan et al. 2011). Power spectrum analyses indicate that the speleothem δ¹⁸O time series show consistent solar cycles, with periodicities at 80–120, 27–35, ∼20, and ∼11 yr, suggesting that solar activity may influence the variability of the Asian summer monsoon on multidecadal timescales. Solar activity may affect the thermal contrast between Asia and the North Pacific, and thus the intensity of the East Asian summer monsoon (Zhao et al. 2007b, Tan et al. 2011). Modeling results suggest that volcanic activity may also have contributed to multidecadal global monsoon variability during the past millennium (Liu et al. 2009). Other factors, such as the ENSO and sea surface temperature in other tropical oceans, may also affect multidecadal variations of the Asian-Australian summer monsoon (Kumar et al. 1999, Oppo et al. 2009). In addition, a stalagmite δ¹⁸O record from Wanxiang Cave, China, shows similarities to the North Atlantic ice-rafted debris record and the North Atlantic Oscillation record during the past 1,800 yr, suggesting a potential link between the East Asian summer monsoon and the North Atlantic Oscillation (Zhang et al. 2008).

Compared with those in the Asian-Australian monsoon area, hydrological variations in African monsoon regions over the past millennium are less well established, and high-resolution climatic records are concentrated in the lakes of eastern Africa (Verschuren et al. 2000, Johnson et al. 2001, Russell & Johnson 2005, Stager et al. 2005, Wolff et al. 2011, Tierney et al. 2013). Lake level and lake sediment records do not show clear hydrological changes between the Medieval Climate Anomaly and Little Ice Age, but they do show widespread multidecadal oscillations associated with solar activity and a connection with Indian Ocean sea surface temperature (Verschuren et al. 2000, Russell & Johnson 2005, Stager et al. 2005, Tierney et al. 2013).

Numerous studies have focused on hydrological changes within American monsoon regions over the past millennium, revealing pronounced regional differences (Figure 7). Hydrological records from the northern limit of the ITCZ (e.g., Cariaco Basin and Yucatán Peninsula) document similar variability (Haug et al. 2001, Hodell et al. 2005). This region was drier during the Little Ice Age than during the Medieval Climate Anomaly, supporting a southward migration of the ITCZ during the Little Ice Age (Figure 7). The southward migration of the ITCZ is also supported by results from ice cores, stalagmites, and lake sediments from the South American Andes, which indicate wet conditions during the Little Ice Age (Thompson et al. 1986, Reuter et al. 2009, Bird et al. 2011). However, some hydrological records from both the eastern (Novello et al. 2012) and western (Moy et al. 2002, Conroy et al. 2008) coasts of South America indicate dry conditions during the Little Ice Age, which cannot be explained by an ITCZ migration (Figure 7). Other factors, such as the ENSO and the Atlantic Multidecadal Oscillation, may also contribute to the centennial- and multidecadal-scale hydrological changes in South America (Moy et al. 2002, Novello et al. 2012).

In summary, the centennial-scale southward migration of the ITCZ during the relatively cold Little Ice Age can explain most of the hydrological records in global monsoon areas (Sachs et al. 2009), although some monsoon records and simulated results suggest different patterns and mechanisms (Liu et al. 2009, Yan et al. 2011). Multidecadal monsoon variability is widely recognized in hydrological records of regions affected by the global monsoon, and these records are probably influenced by external changes in solar activity and internal oscillations within Earth's climate system (Kumar et al. 1999, Verschuren et al. 2000, Russell & Johnson 2005, Stager et al. 2005, Tan et al. 2011).

3.4. Interdecadal Variability

Global monsoon precipitation has intensified over the past three decades, mainly due to a significant increasing trend in the Northern Hemisphere summer monsoon. The intensification of global monsoon precipitation results primarily from an enhanced east-west thermal contrast between the Pacific and Indian Oceans that features rising pressure in the subtropical eastern Pacific and decreasing pressure over the Indo-Pacific warm pool. This mechanism tends to amplify both the Northern and Southern Hemisphere summer monsoons. However, the stronger warming trend in the Northern Hemisphere and the weaker warming trend in the Southern Hemisphere create a hemispheric thermal contrast that enhances the Northern Hemisphere summer monsoon but weakens the Southern Hemisphere summer monsoon. The strengthened Pacific zonal thermal contrast results from both natural variability and global warming, and anthropogenic forcing can intensify the hemispherical thermal contrast (Luo et al. 2012, Wang et al. 2012). However, some have suggested that the increased west-east sea surface temperature contrast is the result of observational bias (Tokinaga et al. 2012); thus, the mechanism by which the increased zonal sea surface temperature gradient in the Pacific may increase global monsoon precipitation requires further study.

Indian summer monsoon rainfall shows coherent multidecadal variability with an approximate periodicity of 55 to 60 yr, without a trend or a climate change signal. Generally, Indian summer monsoon rainfall was above normal from 1891 to 1900 and from 1930 to 1960 and was below normal from 1901 to 1930 and from 1971 to 2000 (Goswami 2006). The interdecadal variability of Indian summer monsoon rainfall is attributed to the complex ENSO-like interdecadal variability of the Pacific sea surface temperature (Graham 1994, Kawamura 1994). A significant negative relationship between the interdecadal variability of Indian summer monsoon rainfall and Niño 3 sea surface temperature indicates a close association on interdecadal timescales (Parthasarathy et al. 1994, Torrence & Webster 1999, Krishnamurthy & Goswami 2000). An empirical inverse relationship between the Indian summer monsoon and the ENSO has weakened remarkably since the late 1970s (Figure 8) (e.g., Kumar et al. 1999). In addition to oceanic forcing, aerosols can also alter cloud density and affect the radiative balance in the atmosphere, leading to changes in cloud microphysics and atmospheric stability, which can either suppress or foster clouds and precipitation. Recent studies (e.g., Ganguly et al. 2012) suggested that the widespread summertime drying over South Asia during the second half of the twentieth century can be primarily attributed to human-influenced aerosol emissions—that is, the drying results from a slowdown in the tropical meridional overturning circulation, which compensates for the aerosol-induced energy imbalance between the Northern and Southern Hemispheres.

The traditional strong East Asian summer monsoon signifies the northward extension of the rainfall belt and above-normal rainfall over northern China. Since the end of the 1970s, the East Asian summer monsoon has followed an evident weakening trend with a “southern China flood and northern China drought” rainfall pattern (Zhou et al. 2009a). One possible reason is that warming in the Indian Ocean and far western Pacific Ocean has induced a westward extension of the western Pacific subtropical high by enhancing convection over the equatorial Indian Ocean and Maritime Continent (Zhou et al. 2009b). Meanwhile, an increase in the extent and depth of snow cover and a weakening in the sensible heat source over the Tibetan Plateau (Zhao et al. 2007a, Duan et al. 2013) have apparently weakened monsoon circulation and produced a southward retreat of the monsoon rainfall belt. The subtropical jet stream is also an important factor to consider in relation to the East Asian summer monsoon (Molnar et al. 2010). Relatively large-amplitude warming in the higher latitudes of the Eurasian continent could decelerate the westerly jet and produce a decadal-timescale change in the East Asian summer monsoon. All of these mechanisms are related, to a certain extent, to anthropogenic global warming. Moreover, anthropogenic aerosols might be a significant factor responsible for such decadal changes. Jiang et al. (2013) suggest that anthropogenic aerosols have suppressed precipitation in North China but enhanced precipitation in South China and adjacent oceanic regions. However, some studies (e.g., Ding et al. 2008, Lei et al. 2011) argue that decadal changes in the East Asian summer monsoon likely result from internal oscillations in the climate system. Recent decadal variability in the Asian summer monsoon may thus result from forcings related to both natural variability and human activity in the climate system.

The interdecadal variability of the West African monsoon is well documented and has motivated considerable research efforts (e.g., Fontaine & Janicot 1996, Le Barbé et al. 2002). The dramatic change from wet conditions in the 1950s and 1960s to much drier conditions in the 1970s, 1980s, and 1990s over the whole region represents one of the strongest interdecadal signals on the planet in the twentieth century. The decadal-scale rainfall deficit from 1970 to 1990 is dominated by lower rainfall during the peak of the rainy season in August and early September (Le Barbé & Lebel 1997, Le Barbé et al. 2002). Furthermore, the Sahel rainfall variability from July to September associated with the northern African summer monsoon declines distinctly from the wetter-than-average 1950s to the progressively drier 1960s, 1970s, and 1980s.

Earlier studies have attributed the long drying trend to human activity and the associated positive land-atmosphere feedback (Charney 1975). More recent modeling studies propose that such rainfall variability in the Sahel is a response of the African summer monsoon to oceanic forcing, amplified by land-atmosphere interaction (Giannini et al. 2008). The role of positive land-atmosphere interaction is also controversial. In spite of some earlier modeling evidence of positive land-atmosphere feedback through surface albedo and soil moisture, there has been no observational evidence of positive land feedback on African rainfall. Instead, a recent analysis of remote sensing observations did not find statistically significant evidence of positive vegetation feedback on African rainfall (Liu et al. 2006). Therefore, the role of land-atmosphere feedback and human effects on the African monsoon remains controversial.

The domain of the North American monsoon is large, extending from the western United States to northwestern Mexico (Adams & Comrie 1997). The North American monsoon system's strength, onset, and retreat over northwestern Mexico exhibited multidecadal variations from 1948 to 2009. Two dry regimes, associated with late onsets, early retreat, and weaker rainfall rates, occurred from 1948 to 1970 and from 1991 to 2005, whereas a wet regime, associated with early onset, late retreat, and stronger rainfall rates, occurred from 1971 to 1990. Several studies suggest that North American monsoon variability is modulated by the Pacific Decadal Oscillation (Higgins & Shi 2000, Castro et al. 2007), the Atlantic Multidecadal Oscillation (Hu & Feng 2008), and the Arctic Oscillation/Northern Hemisphere Annular Mode (Hu & Feng 2010). Arias et al. (2012) argue that the multidecadal variation is linked to the variability of sea surface temperature anomalies in conjunction with the Atlantic Multidecadal Oscillation and the warming sea surface temperature trends. However, for the South American summer monsoon, the decadal variability featured above-normal rainfall over northeastern Brazil and from the central Andes to Gran Chaco and below-normal rainfall north of the equator and over the southern Amazon Basin during most of the 1980s. The reverse trend appears to have occurred during the late 1980s and early 1990s (Zhou & Lau 2001). Decadal- and longer-timescale variability is related to ocean surface changes on those timescales in the Pacific and Atlantic Oceans (Zhou & Lau 2001).

Overall, the global monsoon exhibited decadal variation during the past century, but the temporal-spatial characteristics differed substantially among specific monsoon regions. Oceanic forcing seems to play a very important role, although the effects over specific oceans are distinct. Decadal ENSO-like variability determines long-term changes in the Asian-Australian monsoon, whereas decadal variability over the tropical Atlantic and Indian Oceans seems to determine changes in the African summer monsoon. In addition, anthropogenic aerosols may suppress precipitation in the Asian summer monsoon regions. Specifically, in the case of the East Asian summer monsoon, aerosols decrease sensible heating over the Tibetan Plateau, weakening monsoon circulation and resulting in a southward retreat of the summer rainfall belt in East China. For the American summer monsoon, decadal variability is primarily influenced by the Pacific Decadal Oscillation, Atlantic Multidecadal Oscillation (Hu & Feng 2008), and the Northern Hemisphere Annular Mode.

3.5. Interannual Variability and Intraseasonal Oscillations

Among the wide range of timescales for monsoon variability, the interannual and intraseasonal timescales are the most extensively studied due to their socioeconomic impacts on monsoon regions.

The interannual variability of the global monsoon is a dominant year-to-year fluctuation. Rainfall over the Indian and Indonesian-Australian monsoon regions exhibits biennial variability during certain decades (Yasunari 1991, Webster et al. 1998), and similarly, a dominant oscillation at a period of 2–3 yr exists over East Asia (Chang et al. 2001, Li & Zeng 2005). Figure 9 shows the interannual variability of monsoon intensity measured by the DNS index (Li & Zeng 2002) for some major monsoon subsystems. Besides distinct quasi-biennial fluctuations, the anomalous monsoon for each subsystem is apparently related to an ENSO event. This is because ENSO events usually induce anomalous vertical motions in the equatorial Indian Ocean via modification of the Walker circulation, and thus they affect the regional Hadley circulation associated with the Indian summer monsoon (Krishnamurthy & Goswami 2000, Kumar et al. 2006). The Indian summer monsoon and the ENSO were negatively correlated on the interannual timescale prior to the late 1970s (Kumar et al. 1999). An anomalous Indian summer monsoon may influence the interannual variability of the Australian summer monsoon: A weak Indian summer monsoon may be followed by a delayed Australian summer monsoon, and a strong Indian summer monsoon by an early Australian summer monsoon (Joseph et al. 1991). The Indian summer monsoon is also teleconnected with the West African summer monsoon on the interannual timescale; the physical link in this case is the low-level inflow into the ascending branch of the African ITCZ over the eastern Mediterranean region (Rodwell & Hoskins 1996, Raicich et al. 2003). The interannual variability of the North American summer monsoon is controlled by various anomalies in ocean and land surface conditions (e.g., sea surface temperature, soil moisture) associated with tropical climate anomalies such as the ENSO (Higgins et al. 1998). The East Asian summer monsoon, characterized by Meiyu-Baiu-Changma (the major rainy season from central China across South Korea to Japan, brought by a zonally elongated rainband from June to July), is far more complicated than other monsoons; thus, its interannual variability is forced by multiple factors including the ENSO, the Indian summer monsoon, winter snow cover in Eurasia and the Tibetan Plateau, and the Northern and Southern Hemisphere Annular Modes (Wang et al. 2001a, Nan & Li 2003, Ding & Chan 2005).

In summary, ENSO-related sea surface temperature anomalies in the Pacific and Indian Oceans are believed to be a primary factor responsible for the interannual variability of the global monsoon. Snow cover in Eurasia, thermal forcing of the Tibetan Plateau, and the Northern and Southern Hemisphere Annular Modes may also contribute to the interannual activity of the Asian summer monsoon.

Intraseasonal oscillations are an important component of monsoon variability; they are characterized by several active and break sequences occurring within each summer monsoon season. They significantly affect the local weather and climate as well as the global atmospheric circulation.

The predominant intraseasonal oscillation in a broad period ranging from 30 to 60 days was identified in the Asian summer monsoon regime (Mao & Chan 2005), propagating both eastward and northward (Lawrence & Webster 2002, Mao et al. 2010). Another dominant intraseasonal oscillation that modulates Asian summer monsoon activity is a 10–20-day westward-propagating oscillation (Krishnamurti & Ardanuy 1980, Mao & Chan 2005). The active and break sequences of the Indian summer monsoon are associated with shifts of the monsoon trough position (Webster et al. 1998). The active-break cycles in rainfall over the Indian summer monsoon zone are linked with fluctuations in the intensity of the continental tropical convergence zone (TCZ) (Sikka & Gadgil 1980, Gadgil 2003); the continental TCZ results from northward migration of the Indian Ocean TCZ during late boreal spring to boreal summer. This northward propagation is driven by the north-south gradient of total heating that results from the meridional gradient of convective stability and moisture availability; the maximum convective heating is displaced toward the north of the maximum vertical ascent (Gadgil & Srinivasan 1990). The active and break sequences in Meiyu rainfall of the East Asian summer monsoon are related to convective anomalies over the South China–Philippine Sea, which result from a Rossby wave–like coupled circulation-convection system that propagates north-northwestward from the equatorial western Pacific (Mao et al. 2010).

The timing of summer monsoon onset is one of the most significant manifestations of intraseasonal variability (Webster et al. 1998). In the Australian-Asian monsoon regime, the yearly summer monsoon onset is often associated with an eastward-propagating tropical intraseasonal oscillation event (Madden & Julian 1994). In northern Australia, the monsoon onset date is defined to be the beginning of the first westerly wind burst in each wet season (Wheeler & McBride 2012). The Asian summer monsoon onset arises primarily from the arrival of a wet phase in the tropical intraseasonal oscillation or a phase lock between intraseasonal oscillation modes (Ding & Chan 2005). The onset of the Asian summer monsoon occurs in the eastern Bay of Bengal in early May, in the South China Sea in mid-May, and in South Asia in early June (Wu & Zhang 1998, Mao & Wu 2007). It may be concurrent with the reversal of the meridional temperature gradient (MTG) in the upper troposphere south of the Tibetan Plateau (Flohn 1957, Li & Yanai 1996). Thus, the area-averaged MTG in the middle-upper troposphere has been proposed as an index to measure the onset of the Asian summer monsoon for different regions (Mao & Wu 2007). Rajagopalan & Molnar (2012) suggest that the MTG-based Indian summer monsoon onset and withdrawal dates (Goswami & Xavier 2005) depend on the ENSO much more clearly than those derived from other variables, such as rainfall and large-scale circulation (Joseph et al. 2006).

Although the anomalous convection during each active phase of the Asian and Australian summer monsoons is initiated from the equatorial Indian and western Pacific Oceans, where air-sea interactions may help generate the convective disturbances, the intraseasonal oscillations of summer monsoons are considered to represent an atmospheric internal variability with large-scale coupling between atmospheric circulation and deep convection. The poleward migration of such anomalous convection into the continental monsoon zone supports an intrinsic dynamic mechanism—such as cloud-radiation feedbacks on the position of the TCZ (Gadgil 2003) or instability of the coupled ocean-atmosphere system (Webster et al. 1998)—for the intraseasonal variability of summer monsoons. Extratropical disturbances such as the passage of midlatitude cold surges can also help trigger the tropical convection. The intraseasonal oscillations of the Asian-Australian monsoon may be related to tropical-extratropical interactions (Hsu 2012, Wheeler & McBride 2012).

4. THE TIBETAN PLATEAU AND THE CENOZOIC ASIAN MONSOON

4.1. Cenozoic Asian Monsoon Formation

The formation and development of the Asian summer monsoon during the Cenozoic were closely linked to land-sea redistribution (including the retreat of the Paratethys Sea) and Tibetan Plateau growth; they were also affected by factors such as changes in global ice volume, sea level, and atmospheric CO₂ (Prell & Kutzbach 1992, deMenocal & Rind 1993, Kutzbach et al. 1993, Ramstein et al. 1997, Liu & Yin 2002). The earliest Cenozoic Asian summer monsoon was previously thought to have formed during the very late Oligocene to early Miocene (Qiang et al. 2001, Sun & Wang 2005, Guo et al. 2008). During the Eocene, however, coal and oil shale were widely deposited along the southern margins of the subtropical arid zone (Gu & Renaut 1994), evergreen trees began to appear in South China (Guo 1965, 1983), and mammals preferring moist and forested habitats became common in southeastern China (Qiu & Li 2005). These observations imply that the Asian paleomonsoon may have initiated in the southern tropical zone as early as the Eocene, associated with a commensurate northward retreat of the arid zone and westerly planetary wind system (Guo 1965, Qiu & Li 2005, Huber & Goldner 2012).

Numerical experiments demonstrate that when there is only extratropical continent located between 0° and 120°E and between 20°N–30°N and the North Pole, a weak monsoon rainband appears along the southern border of the continent (Liang et al. 2005, Privé & Plumb 2007, Wu et al. 2012a). Paleogeographic reconstructions suggest that the Indo-Eurasia collision produced a large extratropical continent in the Northern Hemisphere during the Eocene (Molnar & Stock 2009). Thus, according to this line of thought, the Cenozoic Asian summer monsoon should have initiated in the Eocene. Furthermore, numerical experiments show that a strong Asian summer monsoon circulation can be induced by solar forcing when the Tibetan Plateau elevation is half that of today (Prell & Kutzbach 1992). The appearance of high-altitude vegetation after 38 Ma in the Xining Basin (Dupont-Nivet et al. 2008), rapidly increased sedimentation rates at ∼40 Ma in the Hoh Xil Basin (Wang et al. 2008b), isotope-based paleoaltimetry in the Lunpola Basin (Rowley & Currie 2006), and increased exhumation and erosion at ∼35 Ma in the Kunlun Mountains (Clark et al. 2010) indicate that the Tibetan Plateau had already reached a high elevation during the late Eocene. In addition, the Paratethys Sea retreated westward from the Tarim Basin since the Eocene (Bosboom et al. 2011), and this may have also contributed to the initiation of the Asian summer monsoon (Ramstein et al. 1997).

4.2. Tibetan Plateau Growth and Asian Monsoon Evolution on Tectonic Timescales

The Tibetan Plateau began to grow after India collided with Eurasia at 55–45 Ma (Molnar et al. 2010). The expansion of the Tibetan Plateau followed a multiphase process of northward and eastward growth and uplift from its central and southern parts. At 35–20 Ma, the central Tibetan Plateau was raised to a height of 3,000–4,500 m (Rowley & Currie 2006, DeCelles et al. 2007). At 15–8 Ma, the Tibetan Plateau extended eastward and northeastward, and its height and range approached present-day conditions (Yuan et al. 2013). Since the Pliocene, limited growth may have occurred in its northern, northeastern, and eastern margins (e.g., Zheng et al. 2000, An et al. 2001). Tibetan Plateau growth and uplift during the Cenozoic had profound net effects on the Asian summer monsoon and global climate change.

The thermal and mechanical effects of the Tibetan Plateau on general atmospheric circulations were first recognized from meteorological observations (Bolin 1950, Yeh 1952, Flohn 1957, Yeh et al. 1957) and from the “with/without mountains” experiments using earlier numerical models (Kasahara et al. 1973, Manabe & Terpstra 1974, Hahn & Manabe 1975). With more complicated models, sensitivity experiments demonstrate that the Tibetan Plateau intensifies both the winter and summer monsoons in Asia (e.g., Kutzbach et al. 1989, 1993; Ruddiman et al. 1989; Prell & Kutzbach 1992; An et al. 2001). Because the Tibetan Plateau blocks northward moisture delivery from the tropical Indian Ocean and forces strong subsidence air flows, inland Asian aridity (Manabe & Broccoli 1990, Broccoli & Manabe 1992) and the associated atmospheric dust cycle (Shi et al. 2011) are significantly intensified. An experiment with phased uplift of the Tibetan Plateau by an ideal 10% increase in all regions indicated that the response of the East Asian summer monsoon is more sensitive than that of the Indian monsoon (Liu & Yin 2002). This response can be intensified by oceanic feedbacks (Kitoh 2004). Furthermore, Tibetan Plateau uplift can amplify the response of the East Asian summer monsoon to orbital forcing (Liu et al. 2003a).

Kutzbach and colleagues were among the first to simulate the impacts of plateau growth on Asian summer monsoon evolution with four different “realistic” scenarios of temporal and spatial expansions based on limited geological evidence (An et al. 2001). As the Tibetan Plateau grew, the East Asian summer and winter monsoons, inland aridity, and the westerly jet were strengthened. This supports a coupling between Asian climate evolution and phased Tibetan Plateau growth. The distinct responses of the East Asian and Indian summer monsoons to the uplift of different parts of the Tibetan Plateau have also been verified by recent numerical experiments (Zhang & Liu 2010, Tang et al. 2013). Furthermore, a recent study (Shi et al. 2014) demonstrated that Mongolian Plateau uplift also contributes substantially to the strengthened westerly jet.

Using geological evidence and simulation results, An et al. (2001) showed a close coupling between the evolution of the Asian summer monsoon and the phased uplift of the Tibetan Plateau since the Late Miocene. Subsequently, An et al. (2006) further suggested a four-stage increase in summer monsoon intensity and Asian interior aridification at 25–22, 16–14, 10–7, and 4–2.6 Ma, based on a broad range of paleoclimatic records. The global ice volume remained relatively stable, or even increased in certain stages, and global CO₂ concentration remained at a relatively consistent low level after ∼25 Ma (Zachos et al. 2001, Tipple & Pagani 2007). The four stages of monsoon intensification can hardly be explained by changes in ice volume and CO₂ concentration. Therefore, they could be attributed to the stepwise Tibetan Plateau growth revealed by sedimentary, geochemical, and tectonic evidence (An et al. 2006). They are also supported by synchronous enhanced chemical weathering of South China Sea (Wei et al. 2006). In particular, eolian red clay consisting of alternating loess and paleosol layers on the west Loess Plateau was deposited after ∼25 Ma (Qiang et al. 2011). Molnar (2005) suggested an important Tibetan Plateau uplift at ∼8 Ma based on ecological, tectonic, and sedimentary records. This is consistent with the eastward expansion of red clay from the west Loess Plateau after 8 Ma (Sun et al. 1998, Ding et al. 1999, Qiang et al. 2001), which indicates further intensification of the Asian summer monsoon and aridification. In addition, the onset of major Northern Hemisphere glaciation after 2.6 Ma appears to have influenced the development of the Asian summer monsoon. The magnetic susceptibility of Chinese loess indicates that after 2.6 Ma, the Asian summer monsoon began to fluctuate in a pattern closely following that of glacial cycles.

4.3. The Tibetan Plateau and the Asian Monsoon

Monsoons are not controlled by land-sea thermal contrast alone; zonal asymmetric diabatic heating and large-scale orography can also significantly affect them (e.g., Hahn & Manabe 1975, Molnar et al. 1993, Chakraborty et al. 2002, Liu et al. 2007). Mechanical forcing by the Tibetan Plateau is dominant in winter, whereas heating is more important than topography in forcing the summer stationary waves in the tropical and subtropical monsoons (Wu et al. 2005, 2007). Recently, Wu et al. (2012a,b) reinvestigated the influences of the land-sea distribution and large-scale orography on the formation of the modern Asian summer monsoon system. They emphasized the significance of tropical land on the genesis of cross-equatorial flow. In aquaplanet experiments, no monsoon is observed (Liang et al. 2005). When only extratropical land is present in the model, a weak summer monsoon is produced in association with an intense ITCZ. When tropical land is also prescribed, the ITCZ over the longitude domain of the extratropical land disappears in response to strong surface cross-equatorial flow from the winter hemisphere to the summer hemisphere, and a prototype of the Asian summer monsoon—especially the tropical Asian summer monsoon—is produced. However, this monsoon remains in southern India and southern China and cannot extend to the north (Wu et al. 2012a).

The surface sensible heating on large mountain slopes in summer can pump up the surrounding atmosphere to produce surface convergent flow, and cooling in winter can pump it down to produce divergent flow; together, these actions combine to form a sensible heat–driven pump and thereby affect monsoon circulation (Wu et al. 1997, 2007). The uplifted thermal forcing in summer strengthens the coupling between the subtropical and tropical circulation, and between the lower and upper troposphere. The integration of the Iran Plateau into the Tibetan Plateau generated an extra cyclonic circulation in the lower troposphere, which contributes to the dryness in North Africa and the heavy precipitation over the Arabian Sea and northern India, enhances the Indian and East Asian summer monsoons, and spurs the development of a mid-Asia desert (Figure 10) (Wu et al. 2012a).

Boos & Kuang (2010, 2013) recently proposed that Himalayan mechanical forcing exerts a block effect on the colder and drier air advection from the north and suggested that the high surface entropy over North India plays a more important role than the Tibetan Plateau does in producing the local monsoon rainfall and upper warm center over North India. Wu et al. (2007, 2012b), in contrast, emphasize that the sensible heat–driven pump effect of large mountains drives low-level moisture advection from the Indian Ocean to the southern Tibetan Plateau to support the high surface entropy over North India and maintain the north branch of the Asian summer monsoon. Because high surface entropy requires high surface potential temperature as well as high specific humidity, both local surface heating and water vapor transport from ocean to land are required. In any case, both results demonstrate that the Asian summer monsoon is thermally controlled.

The intensity of Tibetan Plateau heating contributes to monsoon variability. Duan et al. (2006) and Duan & Wu (2008) found that the amount of temporal change in boreal spring surface air temperature and in boreal spring and summer land temperature over the Tibetan Plateau has increased significantly in recent decades. Surface sensible heating over the Tibetan Plateau decreased steadily from the mid-1970s to the end of the twentieth century, because surface wind speed decreased during that time and the change in wind speed is much greater than the change in temperature difference between the land surface and the air (Liu et al. 2012). The weakening trend in forcing by the Tibetan Plateau during summer weakens the surface cyclonic circulation and contributes significantly—at least in terms of the underlying mechanism—to a decreasing JJA precipitation trend over North China but an increasing trend over northwestern China and South China (Wu et al. 2012a,b). Numerical experiments based on an atmospheric general circulation model and an ocean-atmosphere coupled model have verified that the variation in thermal forcing by the Tibetan Plateau contributes to the rainfall pattern with wetter climate in the south and drier climate in the north on decadal timescales (Liu et al. 2012).

5. SUMMARY AND PERSPECTIVES

5.1. Common Features of Monsoon Climate

The monsoon definition we put forth in Section 1 emphasizes the common characteristics and dynamic coherence within the global monsoon. Changes in the climates of the various monsoon regimes all have significant seasonality and similar changes in periods and trends on orbital timescales (Table 1). For example, the precession cycle is distinctive in all the monsoonal proxies, implying that the global monsoon shares the coherent variability induced by solar radiation. However, owing to regional diversity, such as differences in underlying surface properties, land-sea distribution, and cloud-radiation forcing, regional monsoons can respond differently even under the same forcing. For example, the vast Eurasian continent, the Tibetan Plateau, and the Indo-Pacific warm pool have made the Asian monsoon the strongest and largest monsoon on the planet.

The periodicity of monsoon changes on different timescales should be attributed to responses to external and/or internal forcings; irregular monsoon changes also develop in response to nonlinear processes and stochastic perturbations. Instability or abrupt changes in the global monsoon are identified as a rapid jump from one stable state to another, reflecting a nonlinear response to the continuing variation of external forcing parameters. For example, abrupt monsoon weakening events recorded in the last glacial stalagmite and loess deposits occurred in response to the North Atlantic cooling events. The onset date and rainfall belt migrations of the Asian summer monsoon (Ding & Chan 2005, Li & Zhang 2009) and the southward retreat of maximum precipitation belts over East Asia in the Holocene (An et al. 2000) exemplify asynchronous monsoon properties. The irregularity, abruptness, and asynchroneity of monsoons increase the complexity and limit the predictability of monsoon evolution.

5.2. Multiscale Monsoon Dynamics

Monsoon dynamics can be considered as the combined effects of external forces (orbital parameters and solar activities) and internal forces (surface boundary conditions and land-air-sea interactions) at different timescales. Monsoon variability encompasses a wide range of timescales, expressed as intraseasonal to interannual cycles and interdecadal, multidecadal, centennial, suborbital (millennial), orbital, and tectonic variability (Figure 11). However, the primary driving forces of monsoon variability on different timescales are not unique. Moreover, multiscale monsoon dynamics involves inherent interactions among individual timescales.

Tectonic-scale monsoon evolution and variability are forced predominantly by mountain uplift/growth and changes in sea-land configuration (e.g., Tibetan Plateau uplift and Paratethys Sea retreat), as well as Northern and Southern Hemisphere ice-sheet changes. Orbital-timescale monsoon variations are primarily controlled by seasonal variations in insolation imparted through changing orbital parameters (e.g., obliquity, precession, and eccentricity) and are strongly modulated by changes in ice volume, greenhouse gas concentrations, and interhemispheric contrast. Millennial-scale monsoon variability is related to North Atlantic climate changes induced by external solar activity and internal interactions. Internal land-air-sea interactions (e.g., thermohaline circulation, sea-ice extent, sea surface temperature), which might be related to solar output and anthropogenic forcing (e.g., greenhouse gases, aerosols, and land use), apparently impact centennial to intraseasonal monsoon variability.

Multiscale interactions increase the complexity of monsoon variability. Longer-timescale monsoon variability usually provides a background for change on shorter timescales. For instance, Tibetan Plateau uplift and Paratethys Sea retreat since the Eocene set the sea-land configuration in Asia, which provided a background for the ensuing monsoon variations. Orbital-timescale monsoon variability may be amplified by Tibetan Plateau growth (Prell & Kutzbach 1992, Liu et al. 2003a). After the onset of the Northern Hemisphere glaciations, the amplitude and frequency of monsoon variations were modulated by changes in ice volume and atmospheric greenhouse gases. The amplitude and frequency of millennial climatic events apparently differ between glacials and interglacials, indicating that ice volume might modulate millennial-scale events (Sima et al. 2004, Wang & Mysak 2006, Wang et al. 2008b). Asian summer monsoon weakening events at centennial to multidecadal timescales were smaller in amplitude than millennial monsoon weakening events within the Holocene (Wang et al. 2005b), implying a constraint that millennial forcing places on centennial to multidecadal variability.

5.3. Perspectives on Future Monsoon Research

Monsoonal climate changes are driven by external and internal forces that primarily manifest as changes in circulation and precipitation. These external and internal factors can influence monsoon variability by affecting the thermal difference in the underlying surface and the resultant pressure gradient. Owing to the intrinsic mechanisms, the responses of the global monsoon to these forcings show some coherent features in spatial-temporal variations. However, deciphering the dynamics of the global monsoon remains challenging due to unresolved issues, such as the roles of low-latitude processes and the interactions between high- and low-latitude climates and between Southern and Northern Hemisphere climates across various timescales. Particularly, further investigation is needed into the relationship of monsoon variability on tectonic timescales with the growth of mountains and the opening and closure of ocean gateways; with ice sheet expansions in the Northern and Southern Hemispheres; and with atmospheric CO₂ changes. In the context of orbital timescales, assessing the physical meaning of different monsoonal indices and identifying the common proxies for global monsoon changes will help to resolve discrepancies among multiple monsoon proxies and elucidate the underlying monsoon dynamics. North Atlantic climate change is believed to contribute to global monsoon changes on millennial to centennial timescales. However, the ultimate cause of abrupt climate change in the North Atlantic, and whether the impacts of such change on monsoon dynamics are conveyed through atmospheric and/or oceanic circulations, is still unclear. Monsoonal climate variability on decadal and shorter timescales is mainly associated with land-air-sea interactions and anthropogenic forcings; however, the relative contributions are not well understood.

Challenges also remain in uncovering the connections among the forces that drive monsoons across different spatial-temporal scales, as well as connections among monsoonal and other atmospheric circulations. The inhomogeneous and asynchronous properties of monsoonal circulation and related precipitation changes also remain to be elucidated. It will be essential to carry out integrated studies that combine past and present monsoon changes and compare data from observations, proxies, and models. Research on monsoon changes since the Industrial Revolution, especially in the past several decades, should also be emphasized. Priorities for future research should include numerical simulation, impact assessment, and prediction and projection of monsoon variability on different temporal-spatial scales—particularly on seasonal to centennial timescales. Changes in global and regional monsoon trends and environmental effects must be evaluated from the perspective of both natural and anthropogenic contributions, especially under global warming conditions. The role of anthropogenic forcings such as greenhouse gases, aerosols, vegetation, and land use need to be understood in the context of the larger land-air-sea coupling system.

disclosure statement

The authors are not aware of any affiliations, memberships, funding, or financial holdings that might be perceived as affecting the objectivity of this review.