Strong monsoon event CIS 4.1
As illustrated in Fig. 3, the Wu3 δ18O record exhibits several strong monsoon events CIS 8–4, similar in timing and pattern to millennial–centennial climatic changes in the North Atlantic region regarding timing and pacing (Anderson et al., 2006). By contrast, the four weak monsoon anomalies probably correlate with ice-rafting events (HE 4, HE 3, and inter-Heinrich ice-rafting events f and h; nomenclature from Bond and Lotti, 1995) in the North Atlantic. In addition, these millennial-scale AM cycles are broadly of opposite sign to those in Antarctica, with weak (strong) monsoon intensity corresponding to high (low) Antarctica temperature. The four weak AM events can be correlated to the warming events in Antarctica [Antarctic Isotope Maximum (AIM) 8, 7, 5 and 4; EPICA Community Members, 2006). The new Wu3 record combined with the previously published data (Liu et al., 2010; Zhao et al., 2010) from the same cave are presented in Fig. 4 showing that these millennial-scale relationships with Greenland and Antarctic ice core records are persistent at this site.
Figure 4. Comparison of AM with Greenland and Antarctica climates. (a) NGRIP δ18O record (Anderson et al., 2006); (b) spliced stalagmite δ18O record from Wulu Cave plotted on their respective time scales (published data are from Liu et al., 2010 and Zhao et al., 2010, respectively); (c) EDML δ18O record given on the GICC05 time scale (EPICA Community Members, 2006). This figure is available in colour online at wileyonlinelibrary.com.
Download figure to PowerPoint
However, a closer inspection of Fig. 5 highlights a minor but significant difference between the cave-based AM intensity and Greenland temperature records. In the Greenland record, the long stadial between GIS 5 and GIS 4 is divided into two phases by the GIS 4.1 event, with the first phase slightly warmer than the second (Fig. 5a). However, at Wulu Cave, the magnitude of the CIS 4.1 event implies a relatively strong monsoon (Fig. 5d), comparable to CIS 5. Moreover, the substantial inferred monsoon weakening evident between other interstadials is not observed between CIS 4.1 and CIS 5. A similar pattern is evident in the stalagmite records from south-west China (Fig. 5e; Zhao et al., 2010), mid-China (Fig. 5c; Zhao et al., 2010), eastern China (Fig. 5b; Wang et al., 2001) and from as far as southern Brazil (Fig. 5f) (Wang et al., 2006) in an anti-phased fashion. A combination of the strong CIS 4.1 and CIS 5 events results in an extended relative strong monsoon period with a duration of ca. 2000 years. Interestingly, the Wu3 (Fig. 5d) and DSB3 (Fig. 5e) δ18O records during the period of CIS 5–4.1 appear different from the Hulu (Fig. 5b) and Sanbao (Fig. 5c) records, where the CIS 5 and CIS 4.1 events appear isolated. Whether the discrepancy arises from the cave locations dominated by the Indian and East Asian monsoons, respectively, is an open question. This is because the previously reported records from Wulu Cave have shown a striking resemblance to the Greenland records (Liu et al., 2010; Zhao et al., 2010), like the Hulu record. Contrasting flow-through times and mixing conditions in the epikarst can also produce the minor difference. Future cave monitoring data and additional records might explain the disparity. Overall, the AM structure is broadly similar, but in an anti-phased fashion, to the EPICA Dronning Maud Land (EDML) ice core record from Antarctica during the episode, despite the short-lived pulse AIM 4.1 (Fig. 5g). This anti-phased correspondence may support the idea that the AM is also influenced by latent heat of cross-equatorial moisture (Clemens et al., 2008) and latitudinal shifts in the intertropical convergence zone (ITCZ) (Fleitmann et al., 2007) via the Southern Hemisphere ‘push’.
Figure 5. Detailed comparison of AM, Greenland and Antarctica temperatures over Heinrich Stadial 3. (a) NGRIP δ18O record (Anderson et al., 2006); (b), (c), (d) and (e) Chinese speleothem records from Hulu (Wang et al., 2001), Sanbao (Zhao et al., 2010), Wulu (this study) and Dashibao (Zhao et al., 2010) caves, respectively; (f) stalagmite δ18O record from southern Brazil (Wang et al., 2006); (g) Antarctica temperature record plotted on the new GICC05 time scale (EPICA Community Members, 2006). Note that the Hulu time scale, not well constrained (Zhao et al., 2010), is shifted younger by 640 years for an easy comparison. The vertical shaded region represents unusual climatic responses in different climate systems. Grey data points are U/Th ages with 2σ uncertainties. This figure is available in colour online at wileyonlinelibrary.com.
Download figure to PowerPoint
The CIS 4.1 event is immediately followed by a weak monsoon event centred at ∼29.8 ka, with a 3‰ increase in δ18O values, coincident with HE 3 in the North Atlantic (Fig. 3). Constrained by layer-counting results, the transitions into and out of the event are 392 ± 25 and 298 ± 18 years, respectively, and its duration at minimum monsoon state lasts 716 ± 32 years, while their counterparts in the NGRIP records are 200 ± 13 years, 1540 ± 101 years and 120 ± 4 years (Anderson et al., 2006). HE 3 is preceded by a rebound-type event GIS 4.1, as expected to be induced by the enhanced Atlantic Meridional Overturning Circulation (AMOC) (Capron et al., 2010). The increased AMOC could be implicated by the strong monsoon event CIS 4.1, because of the strong coupling between them (Sun et al., 2012).
Detailed CIS 8 onset
Previous studies (e.g. EPICA Community Members, 2006; Zhao et al., 2010) have indicated a coherent climatic correspondence between China and North Atlantic millennial-scale climate changes in anti-phase with comparable-scale climate events in Antarctica. Figure 6 presents a snapshot of our record with a 5-year resolution and highlights a monsoonal transition at the onset of CIS 8 with comparable ice core records from Greenland and Antarctica. Three phases are discerned during the transition based on the rate of AM increase. Phase 1 presents a gradual increase in the monsoon intensity, as indicated by a decreasing trend in δ18O from –8.2 to –9.8‰. Phase 2 displays an abrupt decrease in calcite δ18O from –9.3 to –10.0‰ within several decades, suggesting a dramatic jump in monsoon intensity. Phase 3 is another stage of gradual increase in monsoon intensity. A generally gradual transition, relative to abrupt Greenland warmings, was also observed during the transition into the Bølling–Allerød period in the Hulu record (Wang et al., 2001).
Figure 6. Detailed comparison of AM, Greenland and Antarctica temperatures (on the GICC05 age scale) across CIS 8. Three phases can be detected at the CIS 8 onset. If the sharp change of CIS 8 onset is concurrent to abrupt GIS 8 warming and a reversal point of Antarctica temperature variation, the strong/weak AM episodes of CIS 8 would correlate to warm/cold phases of GIS 8, and to cold/warm intervals at Antarctica, strongly supporting the bipolar seesaw mechanism. The vertical bars indicate the boundaries between stadials and interstadials. Grey data points are U/Th ages with 2σ uncertainties. This figure is available in colour online at wileyonlinelibrary.com.
Download figure to PowerPoint
Because of the close relationship between the Greenland temperature and AM intensity (Wang et al., 2001), it is likely that the sharp AM jump, which characterizes the major onset of CIS 8, is synchronous with the rapid temperature rise in Greenland at ∼38.2 ka. This rapid AM change is preceded by the apparently gradual increase (aforementioned Phase 1), which lasts approximately 941 ± 36 years based on our band counting result, equivalent to a half duration of the cold Heinrich Stadial 4. If the sharp changes in the AM and North Atlantic regions are concurrent with a reversal point of Antarctica temperature variations, the initial AM increase would correlate with the late part of AIM 8, which shows a slightly decreasing trend (Fig. 6). This correlation seems to be consistent with an overall anti-phased inter-hemispheric relationship and hence with the atmospheric bipolar seesaw mechanism (Cvijanovic et al., 2013).
This phased transition into CIS 8 supports the idea that abrupt changes happen when the climate systems reach a critical tipping point, which commonly refers to a critical threshold at which a small perturbation can qualitatively alter the state or development of the system (Lenton et al., 2008). Levermann et al. (2009) and Schewe et al. (2012) suggest positive moisture–advection feedback (Webster et al., 1998) as a candidate for the main cause of abrupt changes in monsoon dynamics. The feedback dominates the continent–ocean heat balance (governed by latent heat) of the AM system and could lead to abrupt changes as long as atmospheric humidity over an ocean passes a threshold, which could critically alter the moisture transport to the AM region due to increased latent heat release in the continent. The anti-phased correlation between our stalagmite and the EDML ice core record suggests that the moisture transport associated with Antarctica/Southern Ocean temperature conditions might be an important trigger for abrupt switches of AM (Fig. 3). As illustrated in Fig. 6, the decrease in Antarctica/Southern Ocean temperature, at the turning point to a rapid Greenland warming, might approach a critical threshold and contribute to the subsequent abrupt increases in AM and Greenland temperature. Several modelling (Ganopolski and Rahmstorf, 2001; Schmittner et al., 2002; Alley et al., 2003 for a review) and statistical (Cimatoribus et al., 2012) studies have indicated that abrupt Greenland warmings were also connected to a crossing of a tipping point in heat balance between the North and South Atlantic associated with the AMOC. The simultaneous abrupt climate changes in the AM and North Atlantic regions indicate that the rapid reorganization of oceanic/atmospheric circulations has the potential to push the Earth system across a critical state into another qualitatively different mode of operation. In addition, the phased pattern is also seen at the start of CIS 12 preceded by HE 5 despite low resolution (Wang et al., 2001) and at the end of the Younger Dryas (Liu et al., 2013).
Based on the notion that climate changes might recur in a similar fashion at various temporal scales (Wolff et al., 2009; Liu et al., 2010), the three-phase features discerned at the onset of CIS 8 may be comparable to glacial termination (Terminations I, II and IV) processes that have been linked to AM records (Wang et al., 2001; Cheng et al., 2006, 2006, 2009). These observations confirm that the AM is indeed a potential tipping element of the Earth system, where a critical tipping point can induce abrupt changes whether on the millennial or orbital scales.