Introduction

Tropical precipitation is concentrated in a narrow meridional band, called the intertropical convergence cone (ITCZ), as the Hadley circulation causes moisture convergence in the equatorial region and moisture divergence in the subtropics. Figure 1 highlights the inhomogeneity of the climatological tropical precipitation distribution using the average over 1979–2018 based on GPCP (Global Precipitation Climatology Project) data [1]: the upper half of the area-integrated precipitation within 20°S-20°N is concentrated over 26% of the total area (solid white), while half of the total area encompasses nearly 80% of the tropical precipitation (dashed white). Because of the sharp meridional gradients associated with the ITCZ precipitation, a small displacement in its position can cause dramatic variations in local precipitation. The monthly precipitation changes congruent with a change in the monthly ITCZ position, defined as the zonal-mean precipitation centroid from 20°S to 20°N, suggest that a 1° northward ITCZ shift would accompany local precipitation variations up to 30% relative to the tropical mean (Fig. 2). I caution that the precipitation distribution change associated with the ITCZ shift is not unique; that is, the same ITCZ shift can be associated with a different spatial pattern of precipitation change [2]. Figure 2 is a simple illustration of how an ITCZ shift could induce local precipitation changes. Notable rainfall changes associated with a meridional ITCZ shift occur regionally over the Sahel and monsoon regions. Thus, a shift in the ITCZ position will affect billions of people residing in the tropics who rely heavily on agriculture for their livelihood. For instance, it is hypothesized that a southward ITCZ shift, causing an extended dry period, might have contributed to the collapse of the Classic Maya civilization [3]. Furthermore, meridional shifts in the ITCZ modulate the El Niño-Southern Oscillation activity [4]; the genesis [5], intensity, and frequency [6, 7] of tropical cyclones; the Hadley circulation width [8, 9]; and the midlatitude jet position [10, 11]. Despite the far-reaching climatic impacts of ITCZ shifts, most global climate models have difficulty in correctly simulating the present-day tropical precipitation distribution, with deficient precipitation close to and north of the equator and excessive precipitation south of the equator compared to observations [12]. This problem, commonly referred to as the double ITCZ bias, has persisted for several generations of coupled general circulation models [13]. The notorious climatological bias in tropical precipitation calls into question the reliability of future tropical precipitation pattern projections [14]. Thus, it is imperative that we enhance our understanding of the factors controlling the ITCZ position.

Fig. 1
figure 1

The tropical precipitation climatology between 1979 and 2018 from GPCP data (colors). The half of the area-integrated precipitation within 20°S-20°N is demarcated by the solid white line. The half of area within 20°S-20°N is demarcated by the dashed white line

Full size image
Fig. 2
figure 2

The tropical precipitation climatology between 1979 and 2018 from GPCP data (solid contours; contour interval = 2 mm day−1). A regression of the monthly precipitation on the monthly ITCZ location, defined as the zonal-mean precipitation centroid from 20°S-20°N, expressed as a percentage of the tropical mean precipitation between 20°S and 20°N (colors)

Full size image

Historically, the mechanisms that govern the tropical precipitation distribution have been thought to be intrinsic to the tropics. For this reason, previous studies tend to search in the tropics for the causes of the double ITCZ bias [15], but more recent studies point to the extratropics as the potential origin of the double ITCZ bias [16,17,18]. In fact, pronounced extratropically driven tropical precipitation shifts can be found in early atmosphere-slab ocean model simulations for an ice age, in which a southward displaced Atlantic ITCZ is evident due to the massive ice sheets and associated cooling in the Northern Hemisphere [19]. The ability of extratropical processes to displace the ITCZ was also noticed during the development of an early atmosphere-slab ocean climate model [20]. The model simulated a climatological ITCZ position in the Southern Hemisphere (see Fig. 10 in [20]), possibly due to an underestimation of cloud amount in the southern extratropics as a result of the mistake that the prescribed zonally averaged cloud amount was accidentally switched between the two hemispheres (personal communication with Dr. Stouffer). Years later, this unintentional mistake led to the study that investigated the mechanism by which ITCZ displacements can be forced from the extratropics [21]. Only over the past two decades, it has been widely recognized that extratropical radiative perturbations are effective drivers of tropical precipitation changes and that significant progress has been made in understanding the relevant physical mechanisms.

The goal of this paper is to review our current understanding of how extratropical radiative perturbations can result in meridional ITCZ shifts. The so-called energetics framework allows us to identify both radiative feedbacks and ocean dynamics as key factors controlling the ITCZ position. Here, my focus is to synthesize the research on both factors and illustrate their interplay. The manuscript is organized as follows. “Evidence for Extratropical Influence on the ITCZ Position” reviews the literature demonstrating extratropical impact on the ITCZ position from paleoproxy data as well as climate model simulations. “Energetics Framework for the ITCZ Position” outlines the physical mechanism that provides theoretical constraints on the ITCZ position. “Emerging Challenges” discusses limitations of the energetics framework. Lastly, “Conclusions” offers a summary and discussion of future directions.

Evidence for Extratropical Influence on the ITCZ Position

While the tropics have long been recognized as a driving source of extratropical climate variability on interannual timescales [22, 23], the extratropics-to-tropics teleconnection was not well recognized until the early 2000s due to a lack of observational evidence. The possibility of tropical precipitation being shifted due to an extratropical cause has been inferred from paleoclimate records that indicate strong coupling between tropical and high-latitude climate variations on a variety of timescales [24]. For instance, sediment records from the Cariaco Basin over the Holocene and last glacial cycle indicate that the hydrological cycle variability in the western tropical Atlantic is synchronous with climate oscillations at high latitudes from decadal through glacial-interglacial timescales [25,26,27]. Much of the Northern Hemisphere tropics and subtropics underwent rapid and dramatic climate changes during the interstadial (relatively warm episodes during a glacial) and stadial (relatively cold episodes during a glacial) phases of Dansgaard-Oeschger events [28]. In addition, the timing of the Younger Dryas (a relatively cold event that occurred after the last glacial maximum and before the warming of the Holocene) in the western tropical Atlantic sea surface temperature (SST) record from the Cariaco Basin is synchronous with the temperature evolutions obtained from Greenland ice cores [29]. The Younger Dryas event is also evident in sediment records from the Sulu Sea in the western tropical Pacific [30] and in ice core records from Peru [31]. Moreover, runoff proxies into the Cariaco Basin during the Holocene are strongly correlated with reconstructions of the temperature contrast between the extratropics of the Northern and Southern Hemispheres [32]. Put together, paleoclimate records suggest significant tropical climate changes that occur concomitantly with high-latitude climatic events.

The paleoclimate literature has motivated numerous climate modeling studies to confirm the existence of the global teleconnection originating from the high latitudes. One of the first modeling studies to highlight the extratropical impact on the ITCZ shows that last glacial maximum conditions resulted in a meridional shift of the ITCZ to the south of the present-day position, due primarily to the land ice sheet albedo effect rather than different CO2 concentration and orbital parameters [33]. The follow-up study focuses on the ITCZ response to high-latitude ice changes, demonstrating that the ITCZ shifts away from the hemisphere with additional high-latitude ice cover [34]. A number of modeling studies that have examined the global effect of a weakened thermohaline circulation, which is induced by freshwater input to the North Atlantic (so-called “water-hosing” experiments), also indicate that cooling of the North Atlantic is accompanied by a southward ITCZ shift [35,36,37]. In addition, sulfate aerosols that preferentially cool the northern extratropics are simulated to induce a southward ITCZ shift [38], while a future reduction in the aerosol emissions following the passage of clean air acts is projected to yield a northward ITCZ shift [39,40,41]. Consistently, observational data analysis suggests that anthropogenic aerosol emissions have caused a southward ITCZ shift during the twentieth century [42, 43]. Moreover, there is modeling evidence that high-latitude volcanic eruptions in the Northern Hemisphere can cause strong hemispheric cooling and a southward ITCZ shift [44]. Anthropogenic land cover changes during the Holocene that predominantly cooled the northern extratropics have been simulated to shift the ITCZ southward [45]. In contrast, the northern extratropical afforestation in climate model experiments displaces the ITCZ northward due to the albedo effect [46,47,48]. Southern Ocean heat uptake under global warming, which delays warming in the southern extratropics, is argued to shift the ITCZ northward [49]. Southern Ocean open-ocean convection on multidecadal timescales in a model is associated with an overall increase in the Southern Hemisphere SSTs and a southward shift of the ITCZ [50]. Finally, lowering the surface orography of Antarctica is simulated to cause a local warming, which in turn results in an enhanced outgoing longwave radiation and a northward ITCZ shift [51].

In summary, paleoproxy data and model simulations both show that the extratropical climate change can considerably influence the ITCZ position. It has been even argued that the extratropical impact on tropical climate could be as strong as the tropical impact on extratropical climate [52]. Moreover, results from an idealized model experiment suggest that high-latitude forcing can cause a larger ITCZ shift than tropical forcing of same magnitude [53]. Collectively, we see that the ITCZ position responds to the interhemispheric contrast of temperature (or more precisely, atmospheric net energy input) even when the interhemispheric contrast is introduced well outside the tropics.

Energetics Framework for the ITCZ Position

An energetics theory has been developed over the past two decades to understand the zonal-mean ITCZ shift to interhemispheric contrasts in atmospheric energy budget [21, 24, 32, 54,55,56,57,58]. The central idea is that an anomalous energy gain or loss in one hemisphere necessitates an anomalous cross-equatorial energy transport into the hemisphere with net energy loss. This is accomplished by baroclinic eddies spreading extratropical energy perturbations into the tropics [59], followed by a shift of the ascending branch of the Hadley circulation into the hemisphere with net energy gain, given that on average the Hadley circulation transports energy in the direction of the flow in its upper branch. As warm and moist air masses converge toward the ascending branch of the Hadley circulation, the ITCZ consequently shifts toward the hemisphere with net energy gain. The argument implicates that the ITCZ position (ϕITCZ) covaries with the latitude at which the meridional atmospheric energy transport vanishes (the so-called energy flux equator ϕEFE [55]), for which a robust relationship on seasonal and longer time scales exists [60, 61]. The atmospheric energy budget in a statistically steady state is

$$ \nabla \cdotp F=\mathrm{NEI}=R-O, $$
(1)

which states that the divergence of vertically integrated meridional moist static energy transport by the atmosphere F is balanced by the net energy input into the atmosphere column (NEI), consisting of the difference between net downward top-of-atmosphere (TOA) radiation R and ocean heat uptake O. A linearization of the atmospheric energy budget offers an expression for the energy flux equator based on the energetics [32, 56]:

$$ {\phi}_{\mathrm{EFE}}\approx {\phi}_{\mathrm{ITCZ}}\approx -\frac{1}{a}\frac{F_0}{R_0-{O}_0}, $$
(2)

with a the radius of the Earth and the subscript 0 indicating quantities evaluated at the geographical equator. The energy flux equator and hence the ITCZ position are anti-correlated with the cross-equatorial atmospheric energy transport F0 [54, 55]. In particular, the tropical precipitation centroid exhibits roughly a 3° latitude northward shift per 1 Petawatt (PW) of southward F0 on a wide range of timescales from seasonal to geological [62]. Even with no changes in the cross-equatorial atmospheric energy transport F0, the ITCZ position can be shifted through changes in the net energy input into the equatorial atmosphere (R0 − O0) [32, 56, 63]. For example, the increased net energy input into the equatorial atmosphere could explain the equatorward ITCZ shift during El Niño [32, 56]. However, the interhemispheric forcing in the extratropics impacts the ITCZ position primarily through changes in F0 [64].

The steady-state atmospheric energy budget, Eq. 1 can be rearranged to give an expression for the cross-equatorial atmospheric energy transport F0:

$$ {F}_0=\left\langle \mathrm{NEI}\right\rangle =\left\langle {R}_{CLR}+{R}_{CRE}-O\right\rangle $$
(3)

where brackets denote the spatial integral of the global-mean removed quantities over the Southern Hemisphere (or, equivalently, the negative of the spatial integral over the Northern Hemisphere). The net TOA radiation R is divided into the clear-sky component RCLR and the cloud radiative effect RCRE. According to Eq. 3, the cross-equatorial atmospheric energy transport F0 and hence the ITCZ position are determined by physical processes that affect the atmospheric energy budget, such as cloud and water vapor radiative feedbacks and ocean dynamics; no direct information about tropical SSTs are needed to predict the precipitation response. This seems to be in contrast to the classical studies that argue the tropical precipitation is driven largely by local SSTs [65, 66]. The energetics framework views the SST response as a result of the surface energy budget rather than a driver of the tropical precipitation [67]: however, this is not to say that the SST response is unnecessary for the ITCZ response. The intimate coupling between the surface energy budget, SST, and precipitation implicates that the ITCZ cannot be displaced if the tropical SSTs are not allowed to respond.

Now, we proceed to discuss the factors that determine the ITCZ response to hemispherically asymmetric radiative perturbations, which are summarized in Fig. 3. Here we are considering a scenario in which the atmospheric energy budget is perturbed via an external radiative forcing, for instance, by a change in orbital parameters. In this case, the different components of the climate system (i.e., the atmospheric and oceanic circulation) tend to work together to compensate for the radiative perturbation (Fig. 3), although their relative effectiveness may vary. Importantly, the shallow wind-driven and thermohaline ocean circulations tend to adjust in a similar way, that is, they act to balance out the hemispheric energy perturbation. In contrast, in water-hosing experiments, in which the Atlantic Meridional Overturning Circulation (AMOC) is forced to weaken via a freshwater forcing in the North Atlantic, the wind-driven and thermohaline circulations adjust in opposite ways with regard to their cross-equatorial energy transport.

Fig. 3
figure 3

Schematic illustration of the physical mechanisms that determine the response of the zonal-mean ITCZ position to interhemispheric extratropical energy perturbations, which corresponds to the case with the Northern Hemisphere cooling and the Southern Hemisphere warming. Extratropical energy perturbations are spread into lower latitudes by baroclinic eddies. The effective forcing amplitude is modulated by radiative feedbacks: robust positive feedbacks include the water vapor feedback associated with the ITCZ shift and low cloud feedback in the cooled extratropics, while the cloud radiative effect associated with the ITCZ shift is uncertain due to the subtle compensation between the shortwave and longwave components. The interhemispheric energy perturbation is partially compensated by the energy transport toward the cooled hemisphere, mostly by the Hadley circulation in the atmosphere and by the ocean circulation such as shallow cross-equatorial cell and the deep Atlantic Meridional Overturning Circulation (AMOC). The cross-equatorial cell arises as a result of the Ekman pumping associated with anomalous easterlies in the cooled Northern Hemisphere extratropics and the Ekman suction associated with anomalous westerlies in the warmed Southern Hemisphere extratropics. The AMOC strengthens as the subduction in the subpolar North Atlantic strengthens in response to the Northern Hemisphere cooling. As the Hadley circulation transports moist static energy in the direction of the flow in its upper branch, the moisture in the lower troposphere is transported toward the warmed Southern Hemisphere, accompanying the ITCZ shift. [Courtesy of Y. Shin]

Full size image

Sensitivity to Radiative Feedback

Radiative feedback processes are crucial for determining the sensitivity of ITCZ shifts to interhemispheric forcing as they determine the net radiation response, yet large model uncertainties exist in their representation, ranging from the details of the moist convection parameterization to cloud radiative transfer schemes. The relative roles of different physical processes are often explored by adopting hierarchical general circulation models with radiation schemes of varying complexity, which includes a gray radiative transfer with neither water vapor nor cloud feedbacks [68], a full radiative transfer code but without clouds [69], and a comprehensive scheme with complete water vapor and cloud feedbacks [e.g., 55]. More rigorously, the sensitivity to the treatment of physical processes is often assessed by contrasting model experiments with an interactive and locked climate variables associated with the feedback processes of interest (water vapor or clouds) seen by the radiation code [53, 54, 69, 70]. Moreover, radiative feedbacks can be attributed to individual physical processes such as changes in temperature, lapse rate, water vapor, and clouds by using such methods as a partial radiative perturbation analysis [71, 72] or a radiative kernel technique [73].

Regardless of the analysis approach, the water vapor feedback is robustly suggested to act as a dominant positive feedback, amplifying the ITCZ response to a given interhemispheric forcing [11, 69]. Changes in water vapor can impact atmospheric column energy in multiple ways. For instance, an increase in column water vapor generally enhances atmospheric cooling through increased downward longwave flux to the surface [74]. Increases in upper tropospheric water vapor associated with convection, however, can increase atmospheric heating by reducing outgoing longwave radiation [75]. The robustness of the amplifying effect from the water vapor feedback indicates that the vertical redistribution of water vapor is the dominant factor. Since the ITCZ shifts toward the hemisphere with the greater net column energy content, increased longwave water vapor absorption in the new ITCZ location enhances the total hemispheric asymmetry in net column energy, leading to a larger ITCZ shift. In contrast, the local amplification of extratropical forcing by the water vapor feedback is a secondary effect [69].

The cloud radiative feedback is complex because of its dependence on the forcing distribution. For example, the sign of cloud radiative feedback is sensitive to the vertical structure of prescribed atmospheric heating, whether it peaks above or below clouds. As a result, there is a possibility that two prescribed atmospheric heating perturbations differing in their vertical profile can cause an opposite responses in the ITCZ shift [76]. In case of surface heat flux or insolation perturbations, the cloud radiative feedbacks vary regionally but tend to be positive in the extratropics. The prescribed heating in the extratropics decreases the lower tropospheric static stability from the subtropics to midlatitudes of the forced hemisphere, leading to a reduction of low stratiform clouds; in contrast, the prescribed cooling in the extratropics increases the lower tropospheric static stability from the subtropics to midlatitudes of the forced hemisphere, leading to an increase of low stratiform clouds [e.g., 55]. Hence, the radiative effect from extratropical cloud changes typically acts as a positive feedback, regardless of the degree of atmosphere-ocean coupling [77], thereby amplifying the changes driven from the extratropics [11, 53, 55, 78, 79]. This positive cloud feedback in the extratropics is stronger in response to surface cooling than to surface warming [53, 80, 81], with the contrast being largest for subtropical forcing and gradually decreasing for higher-latitude forcing [81]. This nonlinearity causes zonally asymmetric forcing (alternating heating and cooling with zero zonal mean) to shift the ITCZ away from the forced hemisphere [80, 81]. Moreover, the magnitude of cloud radiative feedbacks varies between ocean basins. For instance, experiments with energy perturbations prescribed over the Southern Hemisphere extratropics indicate that the southeastern Pacific exhibits a positive cloud feedback, the strength of which modulates the ability of Southern Hemisphere extratropical forcing to cause an ITCZ shift [82]. In addition, experiments with anomalous heating applied to different surface locations identify the North Pacific as a region with strong positive cloud feedbacks [83].

In the tropics, the cloud radiative effect directly associated with the ITCZ shift depends on the relative dominance (and the degree of their compensation) between the shortwave and longwave components. The tropical cloud radiative effect acts as a negative feedback if the shortwave component associated with low cloud cover changes dominates over the longwave component associated with high cloud cover changes [e.g., 55]. Conversely, the tropical cloud radiative effect acts as a positive feedback when it is dominated by the longwave component [11, 84]. Indeed, tropical clouds are considered an important source of model spread in the cloud radiative impact [70]. Regardless of the sign of tropical cloud radiative effect, models coupled to a slab ocean tend to show that high-latitude forcing produces a larger ITCZ shift than low-latitude forcing due to the amplifying effects from both water vapor and extratropical cloud feedbacks [53, 85, 86]. As such, uncertainties in model physics can create uncertainties in the nonlocal impacts of extratropical energy perturbations on tropical rainfall.

Compensation by Ocean Dynamics

Theoretical progress on the energetics framework and ITCZ shifts has been made from idealized simulations with a slab ocean [e.g., 21, 55]. Recent model experiments with a dynamic ocean show that high-latitude energy perturbations are less effective in shifting the ITCZ than suggested by slab ocean simulations [87,88,89]. This is expected from the energetics framework because the interhemispheric energy perturbations in the fully coupled system need not be balanced by the atmosphere alone, but oceanic circulation reduces the burden of the atmospheric energy transport. Inclusion of ocean dynamics not only dampens the atmospheric energy transport response but also alters the sensitivity to the latitudinal position of energy perturbations. Under the assumption of a passive ocean, the energy budget of the atmosphere is solely controlled by radiative feedbacks, which tend to be more positive for extratropical energy perturbations than for tropical energy perturbations, so that the efficiency in shifting the ITCZ increases as energy perturbations are introduced at higher latitudes [53, 85, 86]. In contrast, fully coupled model experiments indicate that the atmospheric energy transport response and hence the ITCZ shift are consistently larger when the tropics are perturbed as opposed to when the extratropics are perturbed [90,91,92,93].

Motivated by [16] that partly attributed the double ITCZ bias to the Southern Ocean warm bias, fully coupled models were employed to investigate if the double ITCZ bias is alleviated when the southern extratropics are artificially cooled [82, 88, 89]. Despite the southern extratropical bias reduction, the tropical precipitation bias remained because the cross-equatorial energy transport response occurred not primarily in the atmosphere but in the ocean. The cross-equatorial oceanic energy transport response occurred mostly in the Pacific basin [89] whose majority of meridional energy transport is accomplished by wind-driven subtropical cells [94]. This led studies to propose the mechanical coupling between the Hadley cell and the oceanic subtropical cell via zonal surface wind stress as the oceanic damping pathway [57, 58, 95, 96]. However, the Ekman mass transport cannot produce a large cross-equatorial wind component with the oceanic upwelling zone kept on the equator, even when the ascending branch of the Hadley cell above shifts toward a warmer hemisphere. The ocean, instead, accomplishes the anomalous cross-equatorial energy transport through frictional western boundary currents that cross the equator beneath the lower limbs of oceanic subtropical cells (refer to Fig. 9 in [57]). It forms the so-called cross-equatorial cell together with anomalous Ekman pumping and suction in the extratropics of the cooled and warmed hemisphere, respectively (Fig. 3). As a consequence, the Ekman coupling ensures that the Hadley cell and oceanic cross-equatorial cell drive the cross-equatorial energy transport in the same direction, thereby reducing the burden on atmospheric energy transport and damping the ITCZ shift response [57]. The efficiency of Ekman damping effect is proportional to the gross stability of the oceanic cross-equatorial cell, which is a measure of the temperature contrast between its upper and lower branches [97, 98]. Because the surface water subducted in the cooled extratropics is transported equatorward by the lower branch of the cross-equatorial cell while conserving its temperature, the gross stability of the cross-equatorial cell is set by the SST contrast between the equator and the region of anomalous Ekman pumping in the cooled hemisphere. Hence, the Ekman damping efficiency increases with the cross-equatorial cell width, which gets wider for higher-latitude energy perturbations associated with a larger shift of surface westerly jet. The strengthening of Ekman damping effect with the latitudinal position of energy perturbations is consistent with a more muted ITCZ shift response for higher-latitude energy perturbations [91].

Subsequent fully coupled model experiments with energy perturbations at different geographical locations indicate that the cross-equatorial oceanic energy transport response is not always dominated by the Pacific basin but can be controlled by the Atlantic basin. In fact, the important role of the Atlantic in setting the cross-equatorial oceanic energy transport can be inferred from earlier studies. For example, a slowdown of the AMOC reduces the predicted northward ITCZ shift under the RCP8.5 climate change scenario [99]; the AMOC weakening associated with future aerosol reduction partially explains the weaker ITCZ shift in the Atlantic than the Pacific basin [100]. Moreover, the Atlantic Ocean contributes the most to the cross-equatorial oceanic energy transport in climatology [e.g., 101]. Indeed, a Pacific-dominated response is only the case for the Southern Hemisphere forcing; for the Northern Hemisphere forcing, the Atlantic dominates the cross-equatorial oceanic energy transport response [83]. Also, the Atlantic contribution to the cross-equatorial oceanic energy transport response systematically increases as the interhemispheric energy perturbation is moved from lower to higher latitudes, while the Indo-Pacific contribution stays nearly constant [92]. As a result, a higher-latitude forcing induces a more ocean-centric cross-equatorial energy transport response [90]. The AMOC-involved damping effect will depend on the forcing region associated with a sensitivity of the AMOC response to forcing specifics [102].

A relative importance between the Ekman and AMOC-linked oceanic damping pathways depends on the geographical distribution of energy perturbation. Both oceanic damping pathways indicate that the ITCZ response will be more damped when the same amount of energy perturbation is introduced in the extratropics than in the tropics. However, a simplified model with an idealized representation of Ekman mass transport and its associated heat transport indicates that an unrealistically large gross stability of the oceanic cross-equatorial cell would be needed to reproduce the partitioning of cross-equatorial energy transport between the atmosphere and ocean in fully coupled models [103]. That is, the wind-driven cross-equatorial cell cannot be explained solely by the Ekman transport but needs to take into account a deeper western boundary current [57]. The factors determining the relative damping efficiency of different ocean circulation components need to be better explored.

Interplay Between Radiative Feedback and Ocean Dynamics

Interdependence of net TOA radiation response and the ocean heat uptake response in Eq. 3 complicates the attribution of cross-equatorial atmospheric energy transport response. For example, radiative feedbacks largely determine the SST response that modulates the ocean heat uptake response; in turn, ocean states mediate radiative feedbacks by setting the bottom boundary conditions for atmospheric dynamics and thermodynamics that are critical to radiative feedback processes. Thus, the ocean circulation changes neglected in slab ocean configuration will alter the radiative feedbacks, making radiative feedbacks in fully coupled models to be distinct from those in slab ocean models [91, 104]. Dependence of cloud radiative feedback on ocean circulation has been discussed in the context of global warming [105, 106].

In the presence of a dynamical ocean, the SST response is damped, and thus the SST-mediated radiative feedbacks are weakened. For example, fully coupled models show a substantially weakened positive SST-low cloud feedback in the extratropics, which in slab ocean simulations is a crucial factor for amplifying the tropical response to extratropical energy perturbations. Although the dampened SST response also weakens the negative Planck feedback, fully coupled model experiments indicate that the negative Planck feedback overwhelms the positive shortwave cloud feedback for high-latitude energy perturbations [92, 93]. However, this net negative feedback is offset by the positive ice-albedo feedback and noncloud shortwave effect, resulting in a relatively insensitive net TOA radiation response to the forcing location [92, 93]. It is yet to be examined if the resiliency of net TOA radiation response is an intrinsic feature of Earth’s climate system.

In addition to its role as a buffer of the SST response, the ocean dynamics can modulate the net TOA radiation response by altering the meridional distribution of net energy input (NEI) into the atmospheric column. In a model with intermediate complexity, the oceanic heat transport response to extratropical energy perturbations makes the NEI distribution to be more amplified in the extratropics because the oceanic compensation by the cross-equatorial cell occurs in the tropics, away from the perturbed extratropical region [91]. Because the efficiency of the atmosphere at radiating heating anomalies away to space varies with latitude, a change in the NEI distribution can alter the relative importance between the cross-equatorial atmospheric energy transport and the hemispheric mean TOA radiation in the atmospheric energy balance, Eq. 3. The gray radiation atmosphere that includes water vapor feedback radiates heating anomalies out to space more efficiently from the extratropics, because of a large and positive water vapor feedback in the tropics [91]. As the ocean dynamics redistributes the NEI to higher latitudes at which the radiative feedbacks are more efficient in balancing the heating anomalies, the burden on atmospheric energy transport is further reduced as a result of the interaction of anomalous ocean heat uptake with radiative feedbacks.

However, the interplay between the radiative feedbacks and ocean dynamics is not guaranteed to always dampen the cross-equatorial atmospheric energy transport response. First, the effect of oceanic heat transport in modulating the NEI distribution will depend on the location of energy perturbation that affects the efficiency of AMOC response whose damping effect is most prominent in the extratropics unlike the cross-equatorial cell. For example, the energy perturbations over the subpolar North Atlantic are expected to be damped locally by the AMOC, in which case the NEI distribution becomes less amplified in the extratropics. Second, the meridional distribution of net radiative feedbacks is model dependent. The gray radiation atmosphere, with no cloud radiative and water vapor feedbacks, radiates heating anomalies out to space more efficiently at lower latitudes, which can be attributed to the lapse rate feedback being more negative in the tropics [107]. Comprehensive models with clouds tend to exhibit the net radiative feedback relatively more positive in the tropics, but with a large uncertainty resulting from the shortwave cloud feedback [108].

Emerging Challenges

Validity of a Key Assumption in the Energetics Framework

The energetics framework links ITCZ location to cross-equatorial atmospheric energy transport F0, which consists of mean \( \hat{F_0} \) and eddy \( {F}_0^{\prime } \) components. The eddy component \( {F}_0^{\prime } \) is assumed to be small. The mean component can be expressed as \( \hat{F_0}=V\Delta m \) where V is the mean meridional mass flux and Δm, the ratio of energy transport to mass transport, is the gross moist stability that represents the energy transport efficiency of the Hadley circulation [97]. Thus, at the heart of the energetics framework lies the assumption that the cross-equatorial atmospheric energy transport is mainly driven by the mean meridional mass flux changes rather than changes in eddy components or gross moist stability. If the eddy energy transport dominates the Hadley circulation energy transport, the cross-equatorial atmospheric energy transport is no longer related to the mean meridional mass flux changes and hence the ITCZ. While in most model simulations the Hadley circulation plays a primary role in transporting changes in the cross-equatorial atmospheric energy, there are some model simulations where the eddy component dominates the cross-equatorial atmospheric energy transport response [2, 109]. Moreover, the observed interannual variance in Hadley cell strength is shown to be driven largely by extratropical eddy stresses [110]. Even when the mean component dominates the cross-equatorial atmospheric energy transport response, the energetics framework can still be inappropriate if the mean cross-equatorial atmospheric energy transport response is induced not by changes in the mass flux but by changes in the gross moist stability [61, 111, 112]. Further study is needed to understand when the key assumptions in the energetics framework are invalidated, whether it is sensitive to the differences of experiment design or to the model formulation difference.

Spatial Pattern of Tropical Precipitation Response Associated with ITCZ Shift

The energetics framework is a theory for the zonal-mean ITCZ location and thus has limited relevance to regional changes. For example, the same zonal-mean ITCZ location can be associated with a vastly different precipitation distribution [2]. This indicates that caution should be taken when applying the energetics framework to interpret paleoclimate proxy records that represent regional changes. In an aquaplanet slab ocean setup, the zonal-mean ITCZ location properly depicts the spatial pattern of tropical precipitation response to a localized extratropical forcing because the tropical precipitation shifts in a zonally symmetric manner as the forcing effect spreads into the tropics by baroclinic eddies and midlatitude westerlies [59]. This mechanism is also at play in a more realistic configuration with real-world land distribution and topography, so that the extratropical forcing with a varying structure yields a common tropical precipitation response pattern that is antisymmetric about the equator [79, 85]. However, the tropical precipitation response pattern to a localized extratropical forcing becomes distinct in a fully coupled model framework [87, 113]. Ocean dynamics creates a sensitivity of tropical precipitation response pattern to the extratropical forcing profile: for example, the Northern Hemisphere extratropical forcing tends to induce a meridional shift of tropical precipitation, while the same forcing in the Southern Hemisphere tends to induce a nearly equatorially symmetric response [93]. It is, however, of question whether this sensitivity depends on model biases in the mean state [14].

Issues regarding zonal asymmetries may be resolved through recent efforts that sought to extend the energetics framework to two dimensions, in longitude and latitude, by introducing the concept of an “energy flux potential” [114, 115]. While the ascending branch for a meridional overturning circulation occurs along the energy flux equator (i.e., zero line of the meridional divergent energy flux), the ascending branch for a zonal overturning circulation occurs along a so-called energy flux prime meridian (i.e., zero line of the zonal divergent energy flux). In comparison with the zonal-mean framework, the two-dimensional framework has not yet been widely applied for interpreting shifts in regional precipitation features. Thus far, the extended theory has been applied to explain continental rainfall shifts during the mid-Holocene [115], the observed seasonal and interannual migrations of the zonally varying ITCZ location [114], and the South Pacific Convergence Zone (SPCZ) response to the El Niño-Southern Oscillation (ENSO) [116].

Transient Adjustment of Tropical Precipitation Response

Model simulations with persistent extratropical energy perturbations clearly demonstrate that meridional ITCZ shifts can be driven from the extratropics. Observation-based analysis also finds that the reconstructed ITCZ position is correlated with the North Atlantic SSTs on multidecadal timescales. However, on interannual timescales, the North Atlantic SSTs are forced by the atmosphere rather than the ocean, thereby hampering an observational approach to identify the same connections between the North Atlantic SST variability and the ITCZ position at high frequency [117]. That is, we currently lack observed evidence of extratropics-to-tropics teleconnections on shorter-than-decadal timescales. The question naturally arises if there exists a critical frequency of extratropical forcing for its impact to be detected in the tropics, which involves an investigation of the sequential propagation mechanism. The energetics framework is fundamentally diagnostic and hence does not offer physical insight into the temporal evolution of an equatorward propagation of extratropical impacts. An examination of the transient response to reduced Arctic ice cover reveals two distinct stages: on fast (decadal) timescales, the ITCZ shifts toward a warmer Northern Hemisphere in a way resembles its response in slab ocean setting; on slow (multidecadal and longer) timescales, the initial response is reversed in association with a slower oceanic adjustment that involves the AMOC changes [118]. It is yet to be examined whether this two-stage evolution is robust, as similar studies on the global climatic effect of Arctic ice melt, yet bearing differences in the models and experiment design, show a rather fast oceanic adjustment within 25 years with an equatorially symmetric tropical climate response [119, 120]. The propagation mechanism even in the absence of ocean dynamics is not well established as the proposed wind-evaporation-SST feedback is claimed to be not essential [e.g., 59, 121]. Despite important implications for the interpretation of observed features, a mechanistic theory for the transient adjustment of tropical response to extratropical forcing is yet to be formulated.

Conclusions

Research over the past two decades has made significant progress in our fundamental understanding of the mechanisms controlling the ITCZ location. The energetics framework links meridional shifts in the zonal-mean ITCZ position to interhemispheric contrast in atmospheric heating. This implies that any forcing agents can displace the ITCZ regardless of the forcing location as long as they perturb interhemispheric balances in atmospheric heating. Thus, remote extratropical processes are now recognized as an important factor in determining the ITCZ location [55] as opposed to the traditional view that emphasizes local tropical processes [122] such as local continental/topographic features [123, 124]. The mean ITCZ in the Northern Hemisphere has been attributed to the AMOC that preferentially warms the Northern Hemisphere extratropical atmosphere relative to the Southern Hemisphere counterpart [125, 126]. Moreover, the simulated mean ITCZ bias toward the Southern Hemisphere compared with observation has been hypothesized to be caused by excessive shortwave absorption over the Southern Ocean [16], in contrast to prevalent studies that emphasize local influences [15, 127].

Extratropical mechanisms have been tendered as an important control on the ITCZ position based on idealized simulations with a slab ocean. However, fully coupled climate models with the southern extratropical energy bias reduction yield little improvements to tropical precipitation bias because the resulting cross-equatorial heat transport changes primarily occurred in the ocean [88, 89]. Nevertheless, the importance of extratropical control on the ITCZ position cannot be dismissed because the ability of southern extratropical energy perturbation to induce a meridional ITCZ shift depends on radiative feedbacks such as the stratocumulus-SST feedback in the southern subtropics, which is not properly represented in many models [82]. More importantly, the southern extratropics happens to be the region with a strong oceanic damping effect and a weak positive radiative feedback [83], making the extratropical mechanisms appear inefficient at causing the ITCZ shift. The Northern Hemisphere extratropical forcing tends to be more effective at shifting the ITCZ than the equivalent forcing in the Southern Hemisphere, because of the smaller ocean contribution to the cross-equatorial energy transport changes [93]. In particular, the North Pacific exhibits the strongest efficiency in shifting the ITCZ, even more so than the tropical ocean basins, associated with strong positive cloud feedbacks [83]. Although the Southern Ocean is where the energy budgets are most poorly simulated in models [128], the energy biases over other extratropical regions than the Southern Ocean may be more important in terms of their tropical manifestations, due to the regional dependence of ocean dynamical adjustment and radiative feedbacks.

The question then arises as to which region would be most effective at causing a meridional ITCZ shift. The answer is most likely model dependent because the ability of extratropical forcing to meridionally displace the ITCZ is sensitive to cloud radiative feedbacks. The dependence on models is also expected from the fact that the ocean response to radiative forcing varies widely across models [129, 130]. Concerns about multimodel spread have motivated the Extratropical-Tropical Interaction Model Intercomparison Project (ETIN-MIP), a community-wide effort to investigate mechanisms underlying the linkages between radiative biases over various regions and associated diversity of ITCZ shifts [93]. The ETIN-MIP employs nine independent fully coupled models to perform common numerical experiments where insolation is reduced over three latitudinal bands of persistent model biases. It provides a critical platform to examine the robustness of the sensitivity of (1) the partitioning of energy transport response between the atmosphere and ocean, (2) the effectiveness of different ocean circulation components in compensating the prescribed forcing, and (3) the local and remote radiative feedbacks in modulating the regional forcing effect, to the forcing region.

Here I have reviewed the mechanisms by which extratropical forcing affects the annual- and zonal-mean ITCZ position, a topic that has been extensively tested over the past two decades, but open questions still remain. While studying the extratropical impact alone is interesting in and of itself, it is of practical importance as such idealized experiment configurations with a forcing over finite latitudinal bands allow us to distinguish between the local and remote contributions of regional climate change patterns [131, 132]. The processes governing the ITCZ position have been successfully explored by adopting idealized experiments in a hierarchy of models, ranging from aquaplanet models with simplified gray radiation schemes to aquaplanet models with comprehensive radiation schemes and to slab ocean models with a realistic continental configuration. Although model hierarchy studies tend to not include fully coupled climate models due to the high computational cost, recent advances in the field indicate that a fully coupled climate model should be included in a hierarchical modeling approach to further stimulate mechanistic understanding relevant to the Earth’s climate, which is in line with the spirit of ETIN-MIP.