Milankovitch theory and monsoon

Abstract

The widely accepted “Milankovitch theory” explains insolation-induced waxing and waning of the ice sheets and their effect on the global climate on orbital timescales. In the past half century, however, the theory has often come under scrutiny, especially regarding its “100-ka problem.” Another drawback, but the one that has received less attention, is the “monsoon problem,” which pertains to the exclusion of monsoon dynamics in classic Milankovitch theory even though the monsoon prevails over the vast low-latitude (∼30° N to ∼30° S) region that covers half of the Earth’s surface and receives the bulk of solar radiation. In this review, we discuss the major issues with the current form of Milankovitch theory and the progress made at the research forefront. We suggest shifting the emphasis from the ultimate outcomes of the ice volume to the causal relationship between changes in northern high-latitude insolation and ice age termination events (or ice sheet melting rate) to help reconcile the classic “100-ka problem.” We discuss the discrepancies associated with the characterization of monsoon dynamics, particularly the so-called “sea-land precession-phase paradox” and the “Chinese 100-ka problem.” We suggest that many of these discrepancies are superficial and can be resolved by applying a holistic “monsoon system science” approach. Finally, we propose blending the conventional Kutzbach orbital monsoon hypothesis, which calls for summer insolation forcing of monsoons, with Milankovitch theory to formulate a combined “Milankovitch-Kutzbach hypothesis” that can potentially explain the dual nature of orbital hydrodynamics of the ice sheet and monsoon systems, as well as their interplays and respective relationships with the northern high-latitude insolation and inter-tropical insolation differential.

Graphical abstract

Public summary

• •Orbital-scale climate variations of Earth are dictated by ice sheet and monsoon
• •Views of “monsoon system science” reinforce the Kutzbach monsoon hypothesis
• •A unified Milankovitch-Kutzbach hypothesis better explains the orbital dual nature

Introduction

One of the foremost advances in Earth science occurred in the 20th century with the revival of the theory of orbital forcing of the late quaternary glacial cycles, known as the Milankovitch theory.¹ The theory postulates that changes in caloric summer half-year insolation (or Northern hemisphere summer insolation [NHSI]) at ∼65° N latitude, a region critical for initiation of glaciers, in response to cyclical changes in the Earth’s orbital geometry (eccentricity, obliquity, and precession) drive the rhythmic growth and retreat of the Northern hemisphere (NH) ice sheet (the sensu stricto theory). The waxing and waning of the ice sheet, in turn, are hypothesized to incite ancillary changes in other parts of the global climate system via various forcing and response/feedback mechanisms (the sensu lato hypothesis).

In its early years, Milankovitch theory was not widely accepted because of a lack of corroborating geological evidence. It was not until publication of seminal work² that the Milankovitch theory acquired widespread acceptance and became the canonical theory of Earth’s orbital-scale climate change. The high-resolution oxygen isotope (δ¹⁸O) time series of planktonic foraminifera was the first record of its kind that revealed variations in the global ice volume over the past ∼450 kiloannums (ka).² The record indicated that variations in the global ice volume occurred in a sawtooth pattern characterized by an ∼100-ka quasi-cycle superimposed by shorter quasi-cycles of ∼41, ∼23, and ∼19 ka, as predicted by the Milankovitch theory. These periodicities were also found independently in long-term variations of eccentricity, obliquity, and climatic precession using an analytical solution of the planetary system.³ In particular, the existence of double precession peaks (the so-called “Berger cycles” of ∼23 and ∼19 ka) in the δ¹⁸O records and astronomical solutions become strong validation of the astronomical theory. Subsequent developments of the marine records presented evidence of the ∼100-ka cycle of glaciation, supporting in part the dominant role of high-latitude processes in global climate variations.^4,5,6,7 However, to date, several unsolved problems associated with the theory remain, such as the “100-ka problem”;² namely, that the orbital eccentricity (∼100-ka cycle) contributes much less to isolation change than precession (∼20-ka cycle) and obliquity (∼40-ka cycle), and yet, the most significant variation in the ice volume over the past ∼800 ka is the ∼100-ka glacial-interglacial cycle, similar to the orbital eccentricity cycle.

Despite various problems, the Milankovitch theory has now become accepted wisdom, and it has led to development of an interpretative framework in which the high-latitude glaciation processes takes central stage, and the low-latitude large-scale tropical monsoon systems^8,9,10 are essentially excluded. The exclusion of the latter in the “conventional” Milankovitch theory presents another problem, the “monsoon problem,” which is described by Imbrie et al.⁶ as follows: “Because our focus is on the process of glaciation, we exclude low-latitude sites where previous work has shown that climate history is dominated by changes in the monsoon circulation. In fact, this strategy may well exclude consideration of low-latitude processes that could be important in changing the concentrations of methane and water vapor in the atmosphere”.

About half of the Earth’s surface lies between ∼30° N and ∼30° S, where a monsoon climate prevails. The vast low-latitude monsoon and ice sheet-dominated high-latitude climate regimes constitute the two major planetary-scale climatic systems with complex interactions.^11,12 Theoretical and proxy considerations indicate that two distinct orbital insolation forcings are prominent, with low-latitude insolation forcing dominating the monsoon variations with strong ∼20-ka precession cycles^{8,13,14,15,16,17,18} and NH high-latitude summer insolation forcing driving ice sheet variations with dominant ∼100-ka cycles.^2,19 Therefore, a comprehensive orbital theory would ideally explain not only high-latitude ice sheets but also the embedded low-latitude monsoons as well as the interplay between high- and low-latitude hydroclimates in the context of the distinctive dual insolation forcings. The Kutzbach orbital monsoon hypothesis (namely, that tropical monsoons vary in response to summer insolation)^8,20 remains in dispute.^21,22 The central issues here involve the spatiotemporal nature of the low-latitude monsoon hydroclimate on an orbital scale, the underlying dynamics, the relation to insolation forcing, as well as key interplay between the monsoon and the ice sheet. A thorough understanding of these issues is prerequisite to formulate a more comprehensive form of the orbital theory of climate.

Breakthroughs in the U/Th dating technique²³ in the past decade have propelled speleothems to the forefront of paleoclimatology. The speleothem records of past climate are now considered the strong fourth leg of the paleoclimate “tetrarchate,”^20,24,25 after the three main orbital-scale paleoclimate archives: the marine, ice core, and loess records. One of the key advantages of speleothem records is their absolute and precise chronology in contrast to the essentially orbitally tuned chronology of marine, ice core, and loess records. In this review, we do not attempt to resolve all outstanding issues surrounding the Milankovitch theory;²⁶ instead, we highlight the millennial-scale variations over the last several glacial-interglacial cycles, inferred from absolutely dated speleothem records, including ice age terminations, to analyze the relationship between insolation and Earth’s hydroclimate responses, which may help to reconcile the “100-ka problem.” We then assess the dominant orbital rhythm of precession variability across the low-latitude monsoon regime, especially the phases and spatial variances over multiple ice age cycles as informed by the speleothem and marine records. We also examine the critical hypotheses of climate responses, such as the “early response” and “late response,” to summer insolation at the precession band^6,13 and high- and low-latitude climate interplay via absolutely dated speleothem records from ice sheet-proximal regions and remote low-latitude tropical monsoon regimes. We show that the crucial discrepancies surrounding the Kutzbach orbital monsoon hypothesis might be superficial and may be reconciled within the framework of “monsoon system science.” Last, we build on the key aspects of the Milankovitch theory^1,2 and Kutzbach orbital monsoon hypothesis^8,20 for high- and low-latitude hydroclimate variations on orbital scales; together with results from previous studies of speleothems, ice cores, and loess and marine sediments, we sketch a primary orbital climate hypothesis that also comprises key elements of monsoonal climate variability at orbital scale and the major climate dynamics.

The “100-ka problem” and millennial-scale events

The so-called “100-ka problem” is a classic issue that has been associated with the Milankovitch theory from the very beginning.² Several competing hypotheses have been put forth to address this problem. The significant ∼100-ka periodicity of glaciation has been linked to the Earth’s eccentricity cycle,^27,28 2–3 obliquity cycles,^29,30,31 a combination of obliquity and precession,^16,32,33,34 and an average of 4–5 discrete precession cycles.^16,35,36 Other hypotheses call for interactions involving internal oscillations in the Earth system,^28,37,38,39 particularly dynamics of global ice volume and atmosphere CO₂ changes, invoking non-linearities to amplify the rather modest changes in insolation forcing.^{2,19,40,41,42,43} The observation that the 100-ka cycles in the geological records become strong at a time when the eccentricity is weak suggests that the 100-ka cycles cannot come from the eccentricity but rather from the non-linear response (especially from the ice sheet)⁴⁴ to the variations of climatic precession of 23- and 19-ka periods.⁴⁵

Theoretically, the apparent mismatch between the benthic δ¹⁸O records (or global ice volume) and insolation (i.e., the 100-ka problem) suggests the presence of some type of internal feedback mechanism of the Earth. Based on simple physics, it has been proposed that the rate of change, rather than absolute global ice volume per se, should be dynamically more important to retain a direct anti-correlation with summertime insolation over northern high latitudes.^46,47 The correlation of 65° N summer insolation with the rate of ice volume (or sea level) change^7,48 is more significant than with ice volume itself (Figure 1A). However, a potential weakness of this correlation is that the chronologies of ice volume and sea level are ultimately “tuned” to insolation, leading to a “circular logic.” To avoid this drawback, an absolutely dated monsoon record over the past ∼640 ka, together with its correlations with other hydroclimate records,^50,51 was utilized to analyze the Milankovitch theory in terms of the relationship between insolation variation and ice age termination events.¹⁶ It was found that all seven conventional terminations over the past ∼640 ka occurred during the rising limbs of NHSI separated by 4, 5, 5, 4, 5, and 5 precession or ∼2, 3, 2, 2, 3, and 3 obliquity cycles with durations of 93, 105, 92, 92, 113, and 115 ka, respectively.¹⁶ Two “extra terminations”, termination IIIa (T-IIIa),²⁹ which occurred about one precession cycle after T-III (between MIS 7.4 and 7.3), and T-VIIa, which occurred about two precession cycles after T-VII (between MIS 15.2 and 15.1) (Figure 1), exhibit a pattern similar to the seven conventional terminations with a comparable amplitude of marine benthic δ¹⁸O or sea level changes.^7,48 Under this circumstance, each of the nine terminations is separated from adjacent terminations by an integral number of precession cycles, suggesting precession pacing terminations. All nine terminations took place when the obliquity was substantially higher than the average, suggesting its causal role as well.^29,30,32,34

Figure 1.

Relationships between the benthic δ¹⁸O change rate, suborbital-scale AM and Antarctic temperature variations, and insolation

(A) Benthic δ¹⁸O (LR04)⁷ change rate (gray) and its 10-ka low pass (blue). June insolation at 65°N is shown by the brown curve.⁴⁹ The correlation between the 10-ka low pass of the benthic δ¹⁸O record and June insolation at 65° N is significant, with a correlation coefficient of 0.59.

(B) Suborbital AM variation (Δδ¹⁸O) (green curve in Z score, detrended from the composite δ¹⁸O record by subtracting 21 July insolation at 65° N from the composite AM δ¹⁸O record)¹⁶ and June insolation at 65° N (orange curve in Z score).⁴⁹

(C) AM Δδ¹⁸O (green curve in Z score)¹⁶ and Antarctic suborbital temperature variation (ΔδD) (olive curve in Z score; detrended from the δD record).¹⁶

(D) Spectral analysis results of June insolation at 65° N (brown) and the benthic δ¹⁸O change rate (blue)⁷ and their coherence spectra (black).

(E) Spectral analysis results of 21 June insolation at 65° N (orange)⁴⁹ and AM Δδ¹⁸O records (green)¹⁶ and their coherence spectra (black).

(F) Spectral analysis results of AM Δδ¹⁸O records (green)¹⁶ and detrended Antarctic ΔδD records (olive) and their coherence spectra (black).

Coherence spectra are at an 80% confidence level. Numbers in parentheses show phase differences in degrees between two spectra at precession (∼23- and 19-ka) and obliquity (∼41-ka) bands. Precession and obliquity periodicities of ∼23 and ∼41 ka are significant in all four datasets (the benthic δ¹⁸O change rate, June insolation at 65° N, AM Δδ¹⁸O, and Antarctic ΔδD). Cross-spectrum analyses elucidate a phase coherency (broadly in phase) at precession and obliquity bands between the benthic δ¹⁸O and the June insolation, the AM Δδ¹⁸O and the June insolation, and the Antarctic ΔδD and June insolation. These observations impl that the high-latitude insolation-induced ice sheet melting triggered the millennial-scale events in the North Atlantic mostly at the precession/obliquity pacing, which, in turn resulted in weak AM events and Antarctic warming with approximate in-phase precession/obliquity periodicities. Adapted from Cheng et al.^16,25

The notion that the ∼100-ka cycle encompasses an average of 4–5 discrete precession cycles or 2–3 discrete obliquity cycles invokes missed “beats” between major terminations. The underlying mechanism may involve Earth’s internal nonlinear responses/feedback (e.g., ice sheet dynamics, CO₂ changes, and ocean circulation reorganizations). One piece of evidence is that the middle Pleistocene transition (MPT) from the “40-ka world” to the “100-ka world” occurred at ∼800–1000 ka BP (before present, where present = 1950 CE) without associated considerable shifts in astronomical parameters,^49,52 implying that internal forcing, rather than external insolation, changed the tempo of the Earth system response to the external orbital forcing. On the other hand, several studies have focused on the cause of skipped precession and/or obliquity beats between major terminations.^29,33,35 Subsequent studies found that small-scale millennial events with a nature similar to the major termination events occurred proximate to times when large termination beats were missed.^16,51 Therefore, these millennial events have been named “low-amplitude version of terminations,”⁵³ “failed terminations,”⁵⁴ or “unfinished terminations.”⁵⁵ Explanations for full- versus small-scale terminations have also focused on interplay and feedback among factors internal to the Earth’s climate system, such as ice sheet dynamics and atmospheric CO₂ in the context of certain orbital forcing states.^{33,56,57,58,59}

Conventional ice age termination events and other millennial-scale events have been shown to be essentially similar except for their amplitudes.^16,25,50,51 To analyze the relationship between insolation and millennial events, time series of millennial-scale variations were obtained by removing the orbital insolation signal from the Asian Monsoon (AM) speleothem δ¹⁸O record (Δδ¹⁸O) and the long-term trend from the Antarctic ice core δD (temperature proxy) record (ΔδD) (Figure 1B and 1C).¹⁶ These detrended records largely characterize millennial-scale climate variations involving both conventional ice age terminations and low-amplitude versions of termination events. The millennial-scale climate variations deciphered from detrended AM (Δδ¹⁸O) and Antarctic temperature (ΔδD) records show prominent precession and obliquity cycles (Figures 1E and 1F),¹⁶ similar to the change rate of ice volume (Figure 1D). The phase analyses indicate that large change rates of ice volume (or ice sheet melting) and significant millennial events, including conventional terminations, tend to occur at the precession minima (P_min; around NH June insolation maxima)^16,60 under the high-obliquity state, resulting in weak AM and Antarctic warming, a manifestation of the “bipolar see-saw” pattern.⁶¹ A recent model simulation also shows that a precession-controlled increase in low-latitude boreal summer insolation around P_min raises the temperature over the tropical Atlantic, which weakens trade winds and reduces Atlantic water vapor, resulting in a decrease in Atlantic sea surface salinity (SSS) and Atlantic meridional overturning circulation (AMOC) slowdown.⁶² These observations and simulations indicate that insolation, as a possible ice age termination trigger, might never really miss a beat at precession/obliquity pacing. What was missed was the same amplitude of Earth’s internal response/feedback in terms of the ice sheet/CO₂ changes. In this interpretative framework, a northern high-latitude summer insolation increase would accelerate ice sheet melting (Figure 1A)^46,47 and trigger millennial events (Figure 1B) and associated atmospheric CO₂ variations,^50,51 leading to the observed different amplitude of terminations in the proxy records.¹⁶ Therefore, the longstanding and most critical problem associated with the Milankovitch theory, the “100-ka problem,” seems to be not vital any longer, provided that the central concern lies in high-latitude NHSI driving the ice sheet change mostly at precession/obliquity pacing rather than the ultimate outcomes that are largely dependent on Earth’s internal nonlinear responses/feedback, which are not particularly explained by the theory.^46,47

Low-latitude monsoon

The monsoon climate prevails over about half of the Earth’s surface between ∼30° N and ∼30° S (Figure 2).^17,18 The amount of solar energy received in the region is an order of magnitude more than that received above 65° N. Until now, the Milankovitch theory essentially does not involve monsoon dynamics,⁶ called here the “monsoon problem.” This is a critical limitation of the current orbital theory of climate, which focuses mainly on the Earth’s climate system at high latitudes. A more encompassing version of the orbital theory should include, rather than exclude, the global monsoon system. To involve the monsoon in an orbital theory, a systematic understanding of monsoon dynamics on orbital scale is essential.

Figure 2.

Modern monsoon systems

Upper Panel: Shown is the precipitation difference (1979–2016) between JJA and DJF from Global Precipitation Climatology Project (GPCP) version 2.3.⁶³ The green lines show ∼80 (or -80) mm/month contours, which are used to approximately indicate spatial extents of six sub-monsoon systems: the Asian monsoon (AM), North African monsoon (NAFM) and North American monsoon (NAM), Indonesian-Australian monsoon (IAM), South African monsoon (SAFM), and South American monsoon (SAM) systems.⁶⁴ Lower Panel: The precipitation δ¹⁸O difference between the JJA and DJF was simulated by IsoGSM2⁶⁵ (1979–2016). Boreal and austral summer monsoon rainfalls show more negative δ¹⁸O values than their winter precipitations.

The monsoon is nowadays recognized as a global-scale, three-dimensional atmospheric circulation characterized by seasonal reversal in surface wind/rainfall and migration of the inter-tropical convergence zone (ITCZ). The latter is linked closely to the reversal in temperature gradients between continents and ocean and/or the NH and Southern hemisphere (SH) in response to the annual solar radiation cycle.^64,66 Similarly, in paleoclimate world, the global monsoon also defines the major hydroclimate mode across a vast portion of the Earth’s tropics-subtropics on a wide range of timescales (Figure 2).^{15,17,18,67,68} The major monsoon subsystems in the world include the AM, North African monsoon (NAFM), and North American monsoon (NAM) in NH low to mid-latitudes and Indonesian-Australian monsoon (IAM), South African monsoon (SAFM), and South American monsoon (SAM) in SH low latitudes (Figure 2).⁶⁴ Although each of these regional monsoon systems has specific characteristics, all are coordinated largely by the annual cycle of solar radiation, thus displaying a similar fashion (Figure 2). Comparably, on the orbital scale, the insolation variations dominated by precession also appear to coordinate variations of monsoon changes as the major forcing, although the global ice volume and greenhouse gases also influence this as internal forcing/feedback to some extent, especially in the mean annual precipitation and temperature in the monsoon regime.^69,70 In terms of hydrological cycles, vapor-water transition constitutes the essence of the low-latitude monsoon regime, with a dominant precession cycle on orbital scale, in contrast to the liquid-solid transition prevailing in the high-latitude regime, with dominant ∼100-ka glacial-interglacial cycles in the Late Quaternary.^{16,67,71,72,73} In fact, the monsoon has long been theoretically demonstrated to be primarily driven by tropical solar insolation⁸ with dominated precession cycles. If so, then the orbital-scale expression of monsoons ought to be dominated by precession cycles accordingly. A significant body of observational data^16,74 and model simulations^{14,71,75,76,77} supports this hypothesis. However, the precession phases of monsoon orbital variations are apparently more complex, constituting one of the fundamental paradoxes, as discussed below.

Monsoon variations at the precession band

From the astronomical point of view, precession variations overwhelmingly dominate insolation changes over the tropical-subtropical monsoon regime.^49,52 The monsoon variability has been shown to be a magnitude more sensitive to summer insolation forcing than to glacial-interglacial boundary conditions from observational^15,16,17,74 and theoretical considerations,^{13,18,75,78,79} manifesting a primary rhythm of precession cycles. This is largely because the low-latitude monsoon circulation and associated precipitation respond sensitively to the thermal gradient resulting from the difference between thermal properties of continents and oceans, which is amplified (weakened) by larger (smaller) seasonal solar radiation extremes induced by precessional changes.⁷⁸ The Earth’s orbital precession affects the seasonal distribution of solar radiation, resulting in an increase in northern tropical-subtropical summer insolation that is balanced by a decrease in southern tropical-subtropical summer insolation and vice versa. Therefore, it is also expected that monsoons should be registered as an inter-hemispherically anti-phased relationship at the precession band, which has been substantiated by observations of the precessional see-sawing between speleothem records from the AM system in the NH¹⁶ and the SAM system in the SH,^74,80 similar to the seasonal variability between the two monsoon systems (Figures 2 and 3). Although obliquity and eccentricity also contribute to insolation variations, they are generally less important, at least for daily and seasonal mean insolation, especially in the low-latitude monsoon regime.^49,52 Monsoons are typically characterized by the moist summer circulations that provide most of the annual rainfall to terrestrial lands in the tropics and subtropics. Dynamically, the sensitivity of the monsoonal hydrological cycle to orbital parameter changes is also significantly larger in summer than in winter.⁷⁸ Therefore, unless otherwise specified, we highlight the summer monsoon variations in the following discussions.

Figure 3.

Comparison of AM and SAM speleothem δ¹⁸O records with insolation and ice volume changes at the precession band

(A) AM speleothem δ¹⁸O record (green, ice volume corrected),¹⁶ 30º N–30º S June 21 (orange) and July 21 (pink) insolation⁴⁹ and the sea level variation at the precession band (dark gray) (filtered using Gaussian filtering at the precession band; frequency, 0.0435; bandwidth, 0.01).⁴⁸

(B) SAM speleothem δ¹⁸O record (blue, ice volume corrected),⁷⁴ 30º S–30º N December 21 (orange) and January 21 (pink) insolation,⁴⁹ and filtered sea level variation at the precession band (dark gray).⁴⁸ The major millennial-scale event/stadials (Younger Dryas [YD]; HS1, HS4, and HS6; and Greenland stadials 22 and 25) occurred at the rising (falling) limbs of insolation in the NH (SH), which have a similar duration compared with the 2- to 3-ka lag of ASM (SAM) to the June 21 (December 21) insolation at the precession band.

Ruddiman^13,81 proposed two types of hydroclimate responses to summer insolation changes at the precession band. (1) The “late response”: NH summer (or June 21) insolation forces a significantly lagged (∼5-ka) ice sheet response and transfers to proximal regions, which is one of the key aspects of the Milankovitch theory. (2) The “early response”: hydroclimate changes in regions remote to the ice sheet, such as monsoons, respond directly to summer insolation without significant lags. Theoretically, the monsoon precessional variations are presumably sensitive to changes in low-latitude insolation and, thus, most likely to lead the ice sheet change. However, the relationship of precession phases between the ice volume and monsoon variations lacks vigorous testing. This is because the ice volume change at the precession band was orbitally tuned with an assumed lag of ∼5 ka to NH June 21 insolation,^2,4,6 which lacks independent verification from precisely dated geological archives. Of particular interest are the recent speleothem records reconstructed from Buckeye Creek Cave (BCC) in east-central North America at the southern fringe of the maximum North American ice sheet,⁸² which may potentially provide a test for the longstanding issue of precession phase lag, because the climate variability at the ice sheet-proximal regions is presumably sensitive to the ice sheet extent signals, which influence the proximal regions via atmospheric/oceanic processes.^6,13 The phase analysis shows that the BCC δ¹⁸O record indeed lags the NH June 21 insolation (NHSI) at the precession band by ∼5 ka (in phase with September insolation) (Figure 4), consistent with the modeling results⁸³ and supporting the conventional assumption of an ∼5 ka lag of the ice sheet change relative to NH June insolation at the precession band. This lagged response of the ice sheet agrees well with the “late response” scenario. On the other hand, several absolutely dated speleothem δ¹⁸O records from the AM, Asian Westerly (AW), and SAM systems suggest prominent precession pacing of mid- to low-latitude hydroclimate.^{16,74,80,84,85,86,87,88,89} At the precession band, these records are essentially in phase with ∼July 21 insolation (i.e., AM-AW records)^16,89,90 and ∼January 21 insolation (i.e., SAM records)^74,80 (Figures 3 and 4), consistent with model simulations^14,91 and the “early response” hypothesis.¹³

Figure 4.

Phase comparison between North American (NA) and AM climate variations at the precession band

(A) The normalized NA speleothem δ¹⁸O record (purple)⁸² and September 21 insolation at 65° N (orange).⁴⁹

(B) The composite AM speleothem δ¹⁸O record (green)¹⁶ and July 21 30° N–30° S insolation (brown).⁴⁹

(C) Filtered NA (purple) and AM (green) δ¹⁸O records, using Gaussian filtering at the precession band (frequency, 0.0435; bandwidth, 0.01).

(D) Spectral analysis results of NA (purple) and AM (green) δ¹⁸O records and their coherence spectra (red). The AM and NA variations follow ∼July 21 and ∼September 21 insolation, respectively, with the AM leading NA hydroclimate variations by ∼2.7 ± 0.8 ka at the precession band.

The apparent ∼2- to 3-ka lag of the AM (SAM) to June 21 (December 21) insolation in the speleothem records may be explained by millennial-scale weak (strong) monsoon events with durations comparable with the lag that often occurred at the rising (falling) limbs of insolation in the NH (SH) (Figure 3).^16,51,92,93 Another interpretation underlines the ice volume forcing, which lags the June 21 insolation by ∼5 ka at the precession band and, thus, could theoretically produce the observed lag of ∼2–3 ka in combination with the in-phase insolation forcing.^94,95,96 However, when we take monsoon variations in SH into consideration, the ice volume hypothesis becomes rather complicated. Although ice volume forcing seems to be able to account for the AM lag, it would be difficult to explain the same temporal lag of the SAM, unless there exists a completely different mechanism in the SH regarding the ice sheet’s influence on monsoons (Figure 3). Provided that the above interpretation is sound, the ice volume effect on low-latitude monsoons would be somewhat small at the precession band.

Based on transient simulations, Kutzbach et al.¹⁴ also found an ∼2- to 3-ka lag of the monsoon precipitation response to the orbital summer insolation forcing at the precession band so that, on average, the monsoon July (January) precipitation maxima and JJA (DJF) precipitation maxima have high coherence. They suggested that the lag was well within the phase limit constrained by the control factors of monsoon precipitation— both thermal dynamics (land/ocean temperature contrast and land sea-level pressure minima) and hydrological dynamics relating to the sea surface temperature (SST), evaporation from the ocean, and landward transport of water vapor.¹⁴ Because changes in atmospheric greenhouse gases, atmospheric dust loading, ice volume, and sea level were not considered in this model simulation, the 2- to 3-ka lag of the monsoon response apparently could be explained by orbital insolation forcing alone without the necessity to involve the ice volume.

Another hypothesis emphasizes that monsoon is largely driven by intertropical insolation differences (or sensible heat difference) or the resultant pressure gradient between approximately the Tropics of Cancer (∼23° N) and Capricorn (∼23° S), the summer inter-tropical insolation gradient (SITIG).⁹⁷ This idea can account, to some extent, for the SH effect on the AM but differs mechanistically from another longstanding hypothesis, that the AM is driven largely by the latent heat originating from the Southern Indian Ocean, which reaches the maxima on the orbital scale near the time of the minima of SH winter insolation (∼11-ka lag to NH June 21 insolation).^94,95 Similarly, several studies also considered the interhemispheric insolation differential (e.g., the insolation differential between 30° N and 30° S) as the forcing of the AM.^93,98,99 In fact, the June 21 insolation differential between 30° N and 30° S (or between 23° N and 23° S) and the insolation at the high latitude of 65° N are virtually identical in terms of their phase, net range, and pattern of variations. However, their influence on Earth’s hydroclimate systems is quite different. Mechanistically, June 21 insolation at NH high latitude is commonly considered the critical forcing driving changes in ice sheet extent (Milankovitch theory), whereas the June 21 (December 21) insolation differential between the interhemispheric tropics propels NH (SH) monsoon variations on the orbital scale. Although it is conventional wisdom that NH (SH) tropical summer insolation drives NH (SH) monsoon on the orbital scale,⁸ physically, the interhemispheric differential in tropical insolation (or SITIG) appears to be a more integrated forcing of monsoons. This is because the insolation difference involves “pull” and “push” forcings from two hemispheres that propel monsoon changes.^{93,97,98,99,100,101,102} One line of evidence lies in the variation of the AM across the ∼60- to 25-ka BP interval in the last glacial, which manifests as a slightly declining trend akin to the insolation differentials (Figure 4B), rather than local NH tropical summer insolation, which has a distinct precessional insolation minimum around ∼45 ka BP in the interval.⁹⁸ Notably, however, the role of the SH in driving the AM was apparently strong during glacials and weak during interglacials.^101,102,103 In the aforementioned interpretative framework, the monsoon early response and the ice sheet late response suggest that the sensu lato Milankovitch theory (or the ice volume forcing drives Earth’s hydrological changes) may not be perfectly suitable in the monsoon world at the precession band,¹³ though the sensu stricto Milankovitch theory retains validity.

Phases of monsoon variations at the precession band

The orbital forcing-induced hydroclimate changes in the monsoon regime show extensive precession variability. However, a closer look suggests that the precession phases of monsoonal hydroclimate variations are quite complex, depending on the regional dynamic in different parts of the monsoon system. In the current interpretative framework,^25,93 when the modern AM becomes stronger with the annual increase of land-sea thermal contrast from spring to summer, the spatial scale of the summer monsoon circulation expands progressively, and more remote moisture is transported farther northwest into the Asian continent. The summer monsoon fringe extends northward, and overall summer monsoon precipitation increases over the continent, specifically over the northwestern fringe regions of the AM system, whereas the overall precipitation δ¹⁸O becomes more negative in most continental regions (Figure 2). These modern observations^25,104,105 provide an approximate analog (similar seasonality dynamics) to the AM intensification process from low to high NHSI conditions or from the precession maximum (P_max) to the P_min state and vice versa. These processes are now demonstrated consistently in many modeling results (Figure 5).^{14,71,75,76,77,113,114,115,116,117,118,119,120} Even over the arid regions (e.g., the Tarim Basin), the precipitation seems to follow these insolation-induced AM changes as well.⁸⁷ A large body of paleo-monsoon records has also shown that the AM changes from P_max to P_min (low to high NHSI) manifest an overall increased summer rainfall and decreased rainfall δ¹⁸O in most parts of the Asian continent.^16,25,99 In contrast, the oceans adjacent to the Southeast Asian continent show a broadly opposite trend at the precession band regarding rainfall (Figure 5) and rainfall δ¹⁸O.¹²¹ This ocean-continent contrast is dubbed the “sea-land precession-phase paradox.”⁹³

Figure 5.

Simulated changes in precipitation rate (mm/day) and 850 hPa wind vectors (m/s) over the AM and adjacent regions during boreal summer (JJA) between the precession minimum (P_min) and maximum (P_max) from CESM experiments (modified from Shi et al.¹²)

The gray line region indicates the Chinese Loess Plateau. Red asterisks indicate locations of major loess-paleosol sections in the Chinese Loess Plateau (including the Jingyuan, Yimaguan, Xifeng, Lingtai, Baoji, Luochuan, and Jingbian sections). The Zhaitang section is at the NE plateau. Yellow dots show marine core locations: MD97-2141,¹⁰⁶ ODP1145,¹⁰⁷ MD06-3067,¹⁰⁸ MD01-2386,¹⁰⁹ U1429,²¹ and MD97-2140¹¹⁰ in the western Pacific and SK237-GC09,¹¹¹ U1446,²² and NGHP17¹¹² in the Indian Ocean.

The nonconformity in sea-land precession phases is evident to first order empirically and observationally, but the exact precession phases remain less known. The modeled P_min-P_max differences theoretically do not provide their precise phases at the precession band. In fact, marine proxy records suggest a large array of potentially different precession phases (∼1- to ∼10-ka lags to June 21 insolation), which has sparked a longstanding and vigorous dispute, leading to different monsoon hypotheses based on diverse monsoon-insolation relationships.^{21,22,40,94,96,112,122,123,124,125} These contrasting ideas have hampered full acceptance of the Kutzbach orbital-monsoon hypothesis as an orbital monsoon theory.

The planktonic foraminifera δ¹⁸O_sw-iv-sst (calibrated δ¹⁸O of sea-surface water based on the reconstructed SST and ice volume isotopic effects) is determined largely by the regional evaporation minus precipitation (E − P).¹²⁶ Huang et al.¹²¹ noted that the δ¹⁸O_sw-iv-sst from the tropical western Pacific shows a broad negative correlation with speleothem δ¹⁸O records from the Asian continent, and they suggested that the strengthened AM at the P_min enhanced moisture flux from the western Pacific into the Asian continent and vice versa, contributing to the broad see-saw pattern on the precessional scale. It appears that orbital-scale variations in the δ¹⁸O_sw-iv-sst over the tropical Pacific and Indian Oceans adjacent to the Asian continent show, to first order, a nearly positive correlation with NHSI (Figure 6), consistent well with model simulations (Figure 5).

Figure 6.

Surface seawater δ¹⁸O_sw-iv-sst from the Indian Ocean and the western Pacific, and comparison with the June 21 insolation differential between 30º N and 30º S

δ¹⁸O_sw-iv-sst is calculated by removal of SST and global mean ice volume effects from δ¹⁸O of planktonic foraminifera using the mean sea level reconstruction.⁴⁸ From top to bottom are ODP1145,¹⁰⁷ MD06-3067,¹⁰⁸ MD01-2386,¹⁰⁹ U1429,²¹ and MD97-2140¹¹⁰ records from the western Pacific region, and U1446²² and NGHP17¹¹² records from the Indian Ocean. Dashed lines show data filtered using Gaussian filtering on the precession band (frequency, 0.0435; bandwidth, 0.01). Blue curves show the differential between 21 June insolation at 30° N and 30° S. δ¹⁸O_sw-iv-sst variations respond primarily to local E-P as well as river runoff in some cases. It appears that the δ¹⁸O_sw-iv-sst records show an overall positive correlation with the June insolation, suggesting a general E-P increase with the increase in summer insolation over the broad oceanic regions in the AM domain. The U1429 record shows almost no precessional variance,²¹ which is consistent with model results that suggest a weak precessional variance in rainfall and runoff originating from the Yangtze River Valley (Figure 5).⁹³

As mentioned above, the “sea-land precession-phase paradox” concerns the AM variations on orbital scales with apparent phase differences (∼1–10 ka) at the precession band between a wide range of proxy records, including speleothem δ¹⁸O and various marine sediment records in the AM regime (Figures 5 and 6).^{15,18,20,25,94,112,122,127,128,129,130,131} A data-model comparison⁹³ was then used to help the interpretation of the observed phase differences at the precession band among some important continental and marine records, including marine records from the East China Sea (U1429)^21,130 and the Andaman Sea (NGHP17),^112,131 and continental speleothem records¹⁶ that show a significant phase difference at the precession band. More broadly, marine δ¹⁸O_sw-iv-sst records show, to first order, a nearly positive correlation with NHSI (Figure 5). Therefore, the summer precipitation (evaporation) inferred from these records decreases (increases) over the marine regions when NHSI is high (low) or the overall monsoon is presumably strong (weak). In other words, the rainfall decreases at the precession band over the broad marine regions were compensated by coupled overall rainfall increases over the Asian continent and, thus, actually signify overall strengthening rather than weakening of the AM (Figure 5).^{93,121,131,132} This scenario is in line with the strong monsoon circulation under high NHSI inferred by the Asian speleothem δ¹⁸O records.⁹³ In essence, these seemingly fundamental differences rest in that the sea-land monsoon records characterized different aspects of the same AM system or dynamics.⁹³

Recently, another set of marine proxy (e.g., δD, δ¹⁸O_sw, and Rb/Ca) records from the northwestern Bay of Bengal (U1446) was developed for the past 900 ka.^22,40 These high-quality records suggest a several-thousand-year-long lag of AM (or Indian monsoon [IM]) precipitation or the resultant river runoff and a few-thousand-year-long lag of changes in IM moisture source and transport paths to the summer insolation forcing at the precession band. This core site, as well as the adjacent catchment basins of the Mahanadi and Brahmani rivers that provide the majority of runoffs to the core site, are located at a unique place, according to the model result,¹³³ where the annual precipitation amount may have different precession phases from the NHSI or the overall precipitation and precipitation δ¹⁸O (commonly correlated with precipitation δD) over the Indian subcontinent (Figure 7).¹³³ That is, the observed precession phases of the U1446 records^22,40 are consistent with model simulations; although, overall, the Indian subcontinental precipitation amount was higher and precipitation δ¹⁸O (or δD) was lower at 9 ka BP (the high NHSI and strong IM scenario) relative to the preindustrial (PI) state (the low NHSI and weak IM scenario), the behavior across the major Mahanadi-Brahmani catchments and the core site appears to be out of phase. In other words, such a differential among procession phases of the proxy records is plausible, and the model results provided an alternative interpretation in the context of new monsoon dynamics.⁹³

Figure 7.

Simulated differences of annual precipitation, annual precipitation δ¹⁸O, and 850-hPa water vapor transport between 9 ka and PI from the COSMOS-wiso

The red star depicts the IODP U1446 location. The major parts of the Mahanadi and Brahmani rivers (blue lines) and their coastal outlets (red squares) are shown. Although the overall continental precipitation amount was higher and precipitation δ¹⁸O was lower at 9 ka (high NHSI and strong AM) relative to the PI (low NHSI and weak AM), the behavior across the major Mahanadi and Brahmani catchments and the IODP U1446 site are considerably out of phase. These modeling results agree broadly with the observations from the U1446 records.^22,40 Modified from Zhang et al.¹³³

Another important progress in monsoon research is a recent model simulation of precession-scale changes in the upwelling along the western Arabian Sea.¹³⁴ This involves a classic issue of whether the upwelling intensity (or the related ocean productivity) records from the Arabian Sea^94,96 are a direct expression of IM variability, particularly regarding their precession phases, which considerably lag NHSI.²⁰ Idealized experiments by fully coupled climate models under different precession configurations show that the area of upwelling is narrow (wide) during high (low) NHSI time or strong (weak) IM times. The underlying mechanism is the effect of convective heating over northeastern Africa and the western equatorial Indian Ocean on the width and meridional location of the low-level jet.¹³⁴ Although these sensitivity simulations have difficulty to explore the precession phases precisely, new transient climate simulations on the Arabian Sea upwelling intensity appear to suggest a precession phase approximately opposite to NHSI and, thus, akin to the Arabian Sea records (X. Zhang, personal communication). Therefore, a series of upwelling proxy data obtained over the past ∼30 years from the Arabian Sea ultimately has another alternative interpretation in the new interpretative framework in terms of precession phases.^93,134 This new development to some extent explains an important part of the sea-land precession-phase paradox that obstructs acceptance of the Kutzbach orbital monsoon hypothesis.

The “Chinese 100-ka problem”

Another longstanding issue regarding the monsoon hypothesis stems from the Chinese loess magnetic susceptibility (MS) records (as an AM precipitation proxy) from the central Chinese Loess Plateau, which show dominant ∼100-ka cycles (or glacial-interglacial cycles),⁷⁰ implying a major ice volume forcing of monsoon. This is in contradiction with the conventional insolation hypothesis^8,81 and Chinese speleothem records,¹⁶ namely the “Chinese 100-ka problem.”⁹³ Cheng et al.⁹³ brought out the non-mainstream notion that MS “flux,” rather than the MS, would be a more appropriate proxy of precipitation in principle¹³⁵ because it accounts for durations of in situ pedogenesis and, thus, matches to the essence of the physical relationship between the magnetic proxy and precipitation. As expected, the MS flux data exhibit more precession power and less 100-ka power compared with the original MS records,⁹³ which is consistent with a recent spliced loess sand content-MS record.¹³⁶ However, the precession power of MS flux data remains less significant than that of the Chinese speleothem records. A closer look at most modeling results shows that the summer rainfall over the Chinese Loess Plateau may be insensitive to precession change,^{76,77,115,116,119,120,137} although this could be model dependent because different results have also been found in other models.⁶⁹ On the other hand, some model simulations also show that the summer rainfall over the Chinese Loess Plateau is dominated by precession, whereas the annual temperature is more influenced by glacial conditions (CO₂ and ice sheets).^69,70 Therefore, it would be also possible that the loess MS was affected not only by summer precipitation but also by annual temperature and/or other factors/processes.^93,138 A recent study shows that the Tengger desert had also experienced large dry-wet fluctuations with strong 20-ka and 400-ka rhythms as well, probably suggesting East AM (EAM) variations forced dominantly by low-latitude insolation.¹³⁹

The orbital-scale hydroclimate variations in parts of the Indo-Pacific Warm Pool (IPWP), the South China Sea, and the Bay of Bengal appear to show significant ∼100-ka glacial-interglacial cycles as inferred by precipitation and precipitation isotope records.¹⁴⁰ Similar to the alternative interpretations for the Bay of Bengal and South China Sea records (Figures 5 and 6),⁹³ the records from the IPWP (Borneo, Malaysia, and Sumatra, Indonesia) may be tentatively explained by their specific locations in the transition regions between the two “precession regimes” with opposite precipitation amounts and δ¹⁸O variations.^76,77,119 In these regions, like the central Chinese Loess Plateau, muted precession signals may facilitate the powers of other cyclicities. Additionally, the shelf exposure influence because of sea level change on the IPWP hydroclimate is presumably large, affecting regional ocean and atmosphere circulation and amplifying glacial-interglacial cyclicity.¹⁴⁰ The broad anti-phase relationship between NH-SH monsoons on the precession scale implies that the ice volume forcing of monsoons would not be primary because the ice volume forcing is nearly synchronous globally to first order. However, various ice volume forcing effects seem to manifest in the monsoon regime in different ways, particularly in the EAM domain. For example, the ice sheet extent, amplified by the feedback of dust aerosol, can significantly weaken EAM circulation and precipitation.¹⁴¹ It has also been shown that the effect of ice sheets on the EAM is nonlinear depending on the size and location of the ice sheets and the background insolation.^11,79 Recently, a model simulation⁶⁹ has shown that the variation of the summer precipitation over the north of 25° N in the EAM domain is dominated by precession, leading to strong ∼23-ka cycles, whereas over the southern part, the ice volume may play an important role in generating ∼100-ka cycles by influencing the latitude of the ITCZ and the Hadley cell. These findings from model simulations call for further investigation, especially an in-depth data-model comparison. An thorough understanding of the effect of ice volume/CO₂ forcing on monsoons remains challenging.⁶⁹ Sun et al.⁷⁰ proposed that “insolation plays a leading role in affecting changes in temperature, precipitation, and southerly wind during the summer season. Changes in annual mean temperature and precipitation are affected to varying degrees by CO₂ and ice volume.” Overall, even with addition of ice volume/CO₂ forcing with ∼100-ka cycles to the transient simulations, the resultant summer monsoon variations remain predominant with precession cycles in most cases, in line with low-latitude insolation. The main effect of global ice volume/CO₂ forcing seems to lie mainly in the amplitude changes of monsoon variations^12,71,75,93 and/or annual mean temperature and precipitation (or mean state).⁷⁰ In principle, the influence of some of the 100-ka variances on the ASM appears to lie mainly in special changes in boundary conditions, such as certain temperature gradients, ITCZ mean position, and land-sea distribution, instead of direct sensitivity to ice volume changes.¹⁴² These explanations eliminate, in part, another critical obstacle, the generalized “Chinese 100-ka problem,” to acceptance of the Kutzbach orbital monsoon hypothesis.

Monsoon system science

In the past four decades, researchers have commonly focused on just one specific type of proxy from monsoon regimes, such as the loess MS, marine productivity (upwelling), lake pollen, speleothem δ¹⁸O records, etc. To date, a conventional approach has been to use a specific proxy record or multi-proxy records from an individual location to characterize the entire monsoon system/subsystem exclusively, including monsoon periodicities and dynamics; therefore, such an approach has yielded a wide range of superficial discrepancies in our understanding of monsoon variations on orbital scales, particularly regarding the role of insolation forcing.^{20,21,22,93,127,129,133,142} A large body of valuable data has been amassed, which, together with modeling results, has ultimately led to the end of this era itself. Cheng et al.⁹³ analogized the monsoon system to a giant “elephant” and suggested that various proxy records from monsoon regimes characterize different but indispensable parts of the same “elephant”— the whole monsoon system. In other words, most of the apparent differences stemmed from interpretations of diverse proxies that are, in essence, not mutually exclusive. In fact, there is an invaluable complementarity between these different proxy records in the interpretative framework of the whole monsoon system in that they describe distinct dynamic aspects of the same monsoon system and, therefore, are linked closely to each other despite the superficial differences; for instance, in periodicities and precession phases as discussed above. Collectively, the discrete dynamic nature of monsoon calls for a new systematic approach in monsoon research or “monsoon system science” to better understand how the different pieces of the monsoon system work and interact in the context of data syntheses on the monsoon regime scale. Analogous to the legend of “blind men and giant elephant,” a whole, not a partial, view of the monsoon system becomes critical. Therefore, although each climate record is always important, a simple generalization/exemplification from a single record to the whole monsoon system is evidently an outdated approach. This development propels a change in monsoon research to a systematic paradigm or monsoon system science, which underlines a systematic/integrative view of different dynamic natures/responses/feedback across an exceptionally broad range of temporospatial scales in a monsoon system and put them into the context of external insolation and internal (ice volume/CO₂) forcings. This development, including explanations of the two paradoxes mentioned above, may be a decisive step toward acceptance of the Kutzbach orbital monsoon hypothesis as a theory addressing the key aspect of monsoon climates.

Ice sheet and monsoon interplay

The Milankovitch theory postulates that changes in the high-latitude NHSI drive variances in Late Quaternary ice sheet extent, which, in turn, incite ancillary changes in other parts of the global climate system. In the Earth’s tropics-subtropics, however, the monsoon variations are largely driven by tropical-subtropical insolation and manifest as inter-hemispherical anti-phased precession cycles. This dual nature in orbital forcings and their hydroclimate responses/feedback are distinctive, but their interplay is inevitable and complex as well. In the following discussions, we enumerate several relevant issues.

The pattern of glacial-interglacial temperature variations

The most prominent feature of global climate change is the broad similarity between global temperature and atmospheric CO₂ variations on the glacial-interglacial scale globally. For example, the long-term variations of Antarctic temperature are virtually as same as the global ice volume/CO₂ or northern high-latitude temperature changes.^7,143 The SST records across the NH-SH also show patterns broadly similar¹⁴⁴ to the ice volume and Antarctic temperature, consistent with the sensu lato Milankovitch theory: the ice sheet effect on global climate (particularly the changes in the temperature pattern). The observation that changes in the northern high-latitude ice sheet drive global temperature variations predominantly in a similar pattern on the glacial-interglacial scale is important for a better understanding of monsoon dynamics. This is because changes in the temperature gradient (and, in turn, atmosphere-ocean circulation) are likely more important to monsoon dynamics to first order than similar temperature changes associated with glacial-interglacial cycles.¹²

The AM and Antarctic temperature variations are broadly similar on the millennial scale, with a broad anti-phased variability (Figure 1C). As suggested previously, the AM is influenced by a combination of NH “pull” and SH “push” mechanisms.^{16,100,101,102,145} The essence of these mechanisms rests on the influence of the interhemispheric temperature gradient on the monsoon variability on the millennial scale, which is comparable mechanistically with the similar driving forcing of monsoon on the orbital scale: the interhemispheric insolation differential (e.g., the SITIG⁹⁷ or 30° N–30° S^98,99).

CH₄ and CO₂ feedback

It has long been recognized that atmospheric CO₂ change, although mechanistically complex,¹⁴⁶ has a very close relationship with global sea level/ice volume changes, and at the precession band, it is approximately in phase with the NH September insolation (∼5-ka lag to NH June 21 insolation) (Figure 4).^143,146,147 On the other hand, the AM speleothem δ¹⁸O¹⁶ and ice core CH₄ records¹⁴⁸ show prominent precession pacing of low-latitude hydroclimate changes with a nearly in-phase relation with the NH July insolation (∼2- to 3-ka lag to NH June 21 insolation) at the precession band. This phase relation between AM and CH₄ supports the notion that greenhouse gas CH₄ responses to the NH monsoon change without observable lags,^{16,79,149,150} suggesting early CH₄ feedback at the precession band compared with the ice volume and CO₂ in the Earth’s climate system or a potential low- to high-latitude directionality of climate interaction.

On the other hand, the rapid CO₂ increase¹⁵¹ as a critical feedback during the last deglaciation could be also critical for the full glacial termination; for example, it might have propelled the deglaciation striding across the cooling periods of the Younger Dryas (YD) (with an ∼25 ppm CO₂ increase) and Heinrich stadial 1 (HS1) (with an ∼50 ppm CO₂ increase) (Figure 8E).^151,154 This is because, mechanistically, an ∼15 ppm CO₂ rise during a millennial-scale event (e.g., YD and HS1) is sufficient to alter atmospheric moisture transport across Central America and, subsequently, modulate the North Atlantic freshwater budget, ultimately resulting in transition from a weak to a strong AMOC mode and, in turn, intensified AM (Figure 8).¹⁵⁶ Essentially, these processes indicate a high- to low-latitude directionality of climate interaction.

Figure 8.

The AM forcings across the last deglaciation process

(A) The insolation forcing is depicted by 30° N–30° S July 21 insolation.⁴⁹

(B) Ice volume forcing is inferred from the stacked benthic δ¹⁸O record.⁷

(C) Freshwater (or AMOC) forcing is inferred by ²³¹Pa/²³⁰Th ratios of marine sediments.^152,153

(D) Comparison of insolation (orange)⁴⁹ and AMOC (dark blue)^152,153 forcings and their effects on the AM variations (green).¹⁶

(E) Atmosphere CO₂ concentration.^151,154 Orange arrows show CO₂ increases across the YD and HS1.

(F) SAM cave δ¹⁸O records (blue,⁸⁰ green and purple¹⁵⁵) show a persistent dry trend across HS1 (red arrow).

The phase relation between the AM and the ice sheet at the precession band

As mentioned previously, a close comparison of precession phases between absolutely dated AM¹⁶ and North American hydroclimate⁸² records revealed that the AM led the North American hydroclimate (or ice sheet) by ∼2.7 ± 0.8 ka at the precession band (Figure 4). This observation suggests a more active role of low-latitude monsoons in driving global hydroclimate change than previously thought.^{13,14,15,67,73,97,99} That is, at the precession band, low-latitude monsoons appear to be an important driver of global hydroclimate changes, including their potential influence on the high-latitude hydroclimate via input of water vapor and CH₄ to the atmosphere and transport of heat to high latitude through ocean circulation (such as the AMOC).^{25,62,72,73,99}

The role of the AMOC

During the past two decades, the high- and low-latitude climate teleconnections at multiple timescales have been well established empirically and theoretically. One convincing piece of evidence comes from the precise correlation between the millennial-scale events recorded in Greenland ice core and AM speleothem records.^{16,50,149,150} The key process underlying this teleconnection lies in the AMOC dynamics (less in direct ice volume forcing), which links changes in high-latitude temperature to low-latitude monsoon intensity. For example, as inferred by ²³¹Pa/²³⁰Th records^152,157,158 over the last deglaciation process, the high-latitude weak AMOC events associated with HS1 and YD affected the AM profoundly, resulting in pronounced weak monsoon intervals, whereas the influence of a persistent decrease of the ice volume was tentative (Figure 8). In particular, while the ice volume around 18–19 ka BP was very close to its maximum, the AM reached an apparent peak, coherent with the strong AMOC peak, which is clearly not in tune with the ice volume variance (Figure 8). In other words, the ice volume effect appears to be less effective in terms of driving monsoon changes. Nevertheless, provided that the ice volume changes affect the AMOC (very likely) via changing the North Atlantic meltwater forcing for instance, therewith the monsoon will respond much more sensitively. In this respect, the ice volume change rate and the trajectory/distribution of the resultant meltwater in the North Atlantic, not the ice volume per se, would be more critical for low-latitude monsoons.¹⁵⁹ In a pioneer modeling study, Kutzbach and Guetter⁷⁸ showed that “the response of monsoon circulations and tropical precipitation to the orbitally produced solar radiation changes was much larger than the response to changes of glacial-age boundary conditions.” In their experiment, the glacial-age boundary conditions include the ice sheet and related land albedo, sea ice, and SST conditions but not AMOC, consistent with the observations. Another example is a recent model simulation that also suggests that the abrupt low-latitude monsoon weakening events that occurred during the ending processes of interglacials were associated with the astronomically triggered abrupt weakening of the AMOC¹⁶⁰ rather than with a direct ice volume change.

SAM (Amazon River discharge)

Recent studies show that, during the termination processes of millennial-scale events such as HSs, a multi-centennial drying trend occurred as a precursor event in the vast SAM domain, which would have caused a decrease in Amazon River discharge into the tropical Atlantic Ocean (Figures 8 and 9).^150,161 Reduced freshwater input may have induced a positive sea-surface salinity anomaly in the huge Amazon plume region,^166,167 ultimately advected to the deep-water formation areas in the North Atlantic via the great ocean conveyor¹⁶² and contributing to the strengthening of the AMOC^164,165 and vice versa. A strengthened AMOC would induce positive feedback by transporting more saline water to the north, facilitating northward heat transport and a shift in the tropic rain belts, intensification of the AM,^{62,149,150,161} and Greenland warming.¹⁶⁸ This dynamic process provides another example showing that low-latitude monsoon rainfall may affect high-latitude climate.

Figure 9.

Conceptual diagram depicting the climatic dynamics during termination of the millennial-scale climate events.

Based on cave records,^150,161 the weakened SAM resulted in reduced Amazon River runoff into the tropical Atlantic Ocean and, thus, a small Amazon plume region (APR; depicted by the red dashed line) about multi-centuries prior to the millennial event termination at Greenland. The long-term drying trend in the SAM domain may have induced a relatively positive SSS anomaly in the APR, which was subsequently advected to the North Atlantic,^162,163 ultimately contributing to AMOC strengthening.^164,165 White dots show locations of cave records. Purple and white arrows indicate the surface and deep ocean currents, respectively. The light blue arrow depicts the wind direction of the SAM. This figure is modified from Cheng et al.¹⁵⁰ and Dong et al.¹⁶¹ The map shows surface salinity in 2011 as measured by the National Aeronautics and Space Administration (NASA) Aquarius satellite. Orange colors indicate higher salinity and blue colors lower salinity (https://news.uga.edu/wp-content/uploads/2017/12/Amazon-River-Plume.jpg).

Prospect

The Milankovitch theory became an accepted version of the astronomical theory of climate change mainly because the major periodicities of the astronomical parameters were found to be consistent with the geological records of the northern high-latitude ice sheets, and, subsequently, the ice sheets forced nearly coincident global climate changes. However, orbital-scale climate changes of Earth are not dictated by high-latitude ice sheets alone. In fact, more than half a century before Milankovitch’s seminal study,¹ Adhémar¹⁶⁹ and Croll¹⁷⁰ had hypothesized hemispherically asymmetric occurrences of ice ages at precession periods. Although their hypotheses were eventually discarded because of observations that are consistent with the hemispherical coincidence of ice ages and summer, rather than winter, insolation is the most favorable. The core concept that precession-driven changes in insolation drive asynchronous (or anti-phased) hydroclimate fluctuations in the NH and SH found support in the 1980s, when Kutzbach⁸ put forth a simple hypothesis that summer insolation drives monsoon variance: the Kutzbach orbital monsoon hypothesis.¹³ This hypothesis is now tested by empirical and observational data, especially the decisive and elegant evidence from low-latitude monsoon variability inferred, for example, from Asian^16,51 and South American^74,80 speleothem records. We therefore call for a combined Milankovitch-Kutzbach hypothesis— a unified framework for understanding orbital-scale global climate variability. This form of orbital hypothesis not only accentuates that high-latitude NH summer insolation drives changes in ice sheet extent at the Earth’s orbital periods and its nearly synchronous effect globally on glacial-interglacial variability but also emphasizes that inter-tropical insolation differential drives inter-hemispherically out-of-phase low-latitude hydroclimate variability at the precession band (Figure 10). Basically, we suggest blending the global monsoon with the existing framework of the orbital theory to formulate a more comprehensive interpretative framework of Earth’s orbital climate changes, including the dual nature of orbital hydrodynamics of the high-latitude ice sheet and low-latitude monsoon in the context of the dual insolation forcings and their broad interplays (Figure 10). To advance the Milankovitch-Kutzbach hypothesis into a new accepted version of the astronomical theory of climate, there is much more to be done, such as gaining a better understanding of the global-scale low- and high-latitude interactions and feedback and, in particular, research on the forefront of “monsoon system science”— a new monsoon research paradigm.

Figure 10.

Simulated 23- and 100-ka cycle power ratios of annual precipitation illustrate the dual nature of global hydrodynamics in the high- and low-latitude regimes on orbital scales

Shown is simulated cyclicality of annual precipitation variations for the past 300 ka. The transient simulations were initialized from the PI conditions and forced by the orbital (ORB) and greenhouse gas (GHG) forcings as well as the prescribed fluctuations of continental ice sheets (ICE) for the past 300 ka.¹⁷¹ The simulated precipitation time series over each model grid cell was calculated to attain the power spectrum of annual precipitation for each grid cell, and the results were used to estimate the distribution of ∼23-ka and ∼100-ka band-average power. The natural log of this ratio is presented here so that the positive/negative values indicate that the 23-ka cycle of annual precipitation variabilities was stronger/weaker than the 100-ka cycle. The global annual precipitations show stronger ∼23-ka cycles in low-latitude regions (∼30° N–30° S, depicted by dashed lines), in contrast to high-latitude regions, where ∼100-ka cycles are stronger.

Published Online: October 19, 2022

Lead contact website

https://gr.xjtu.edu.cn/web/cheng021/home.