1. Introduction
Previous studies have found that the East Asian summer monsoon (EASM) and East Asian winter monsoon (EAWM) are correlated at different time scales as a result of their different responses to different forcings and mechanisms. For instance, from decadal to millennial scales, the intensities of the EASM and EAWM tend to be negatively correlated, with a strengthened EASM corresponding to a weakened EAWM (Wen et al. 2016; Yan et al. 2020; Zhu et al. 2021) because of modulation by decadal to millennial variabilities in the ocean, including the Pacific decadal oscillation (PDO), Atlantic multidecadal oscillation (AMO), and, particularly, the Atlantic meridional overturning circulation (AMOC). This negative EASM–EAWM relationship from centennial to suborbital time scales is consistent with proxy reconstructions, such as those derived from loess deposits (Kang et al. 2018, 2020). In contrast, at the precessional scale, in response to the precessional forcing of varying seasonality, the EASM and EAWM show an in-phase intensity variation, as analyzed in transient simulations over the past 21 000 years (21 kyr) (Wen et al. 2016). This occurs because during perihelion in, for example, northern summer, the seasonal cycle is increased such that an increase in summer insolation in the Northern Hemisphere (NH) is always accompanied by a decrease in winter insolation in the same region. The opposite occurs during aphelion in boreal summer. That is, when boreal summer occurs at the aphelion, summer insolation in the NH decreases, while winter insolation in the NH increases.
Obliquity forcing has some similarities to precessional forcing in the sense that increased obliquity increases seasonality and, therefore, should lead to increased insolation in boreal summer and decreased insolation in boreal winter. This leads us to hypothesize that the EASM and EAWM variations are also in phase at the obliquity time scales. Nevertheless, loess records from the Chinese Loess Plateau (CLP) have suggested that EASM and EAWM variations at the obliquity time scales are in phase before ∼3 Ma but out of phase afterward (Clemens et al. 2008; Sun et al. 2010). This phase shift was proposed to have been caused mainly by the increased ice volume dominating in the NH after ∼3 Ma (An 2000; Clemens et al. 2008; Sun et al. 2010). This leads to our research question: What is the EASM–EAWM relationship at the obliquity scale?
This work is a follow-up of Wen et al. (2016) and Yan et al. (2020), seeking to further explore the EASM–EAWM relationship at the obliquity time scales. Here, we analyze the monsoon relationship in a set of transient simulations covering the past 300 kyr, in which the forcing is accelerated by 100 times (Lu and Liu 2018). The results show that obliquity indeed forces an in-phase relationship between the EASM and EAWM. In addition, the presence of ice sheets can further modulate the EASM–EAWM relationship with phase shifting. This paper is arranged as follows: The data and methods are introduced in section 2. The spatial distributions of the EASM and EAWM in response to obliquity forcing are reported in section 3. The time-dependent EASM–EAWM relationship is illustrated in section 4. The evolutionary phases between the EASM and EAWM at the obliquity time scales are discussed in section 5. The EASM–EAWM relationship forced by obliquity is compared to that forced by precession in section 6. Conclusions are drawn in section 7.
2. Data and methods
a. Simulations and reconstructions
To understand the effect of obliquity forcing, we first analyze a 300-kyr transient simulation forced only by orbital parameters, including eccentricity, obliquity, and precession (ORB run). This simulation was conducted with the National Center for Atmospheric Research (NCAR) Community Climate System Model, version 3 (CCSM3), with an atmospheric horizontal resolution of ∼3.75° (T31). The orbital forcing spanned from 300 ka BP to the present day (0 ka BP) and was accelerated by a factor of 100 (Lu and Liu 2018). To assess the role of orbital forcing relative to other forcing factors, we also analyze two accompanied, 300-kyr, accelerated transient simulations. One was forced by transient greenhouse gases in addition to the orbital parameters (ORC run), and the other was forced further by the transient global ice sheet in addition to orbital parameters and greenhouse gases (OGG run). Changing the ice sheet involves large topographic changes that influence air pressure; these changes are equivalent to forcing a piece of terrain into the atmosphere, thereby squeezing the air and increasing the air pressure. However, in the OGG run used in this study, the model did not reflect this process. Instead, the global mean sea level pressure (SLP) in the OGG run was adjusted to reflect ice sheet volume changes. In other words, the SLP variable in the OGG run can be used to reflect global ice sheet volume changes but cannot be used to reflect other phenomena.
The external forcings used in the experiments are shown in Fig. 1. Although they are dominated by the 100-kyr cycle, the greenhouse gases and ice sheet forcings also contain modest components at the obliquity time scale (Figs. 1d,f), which could potentially force monsoon responses. The obliquity components of CO2 and the ice sheet (represented by the global SLP in the OGG run) explain approximately 14% and 15% of their total variances, respectively. The obliquity component of the CO2 amplitude is 32.8 ppm, approximately 32% of its total amplitude, which is 103.1 ppm.
External forcings used in the 300-kyr accelerated transient experiments, including (a) eccentricity, (b) obliquity (°), (c) precession index, (d) CO2 concentration (ppm), and (e) global ice volume (106 km3) during the past 300 kyr. (f) The globally averaged SLP (hPa) in the OGG run is adjusted to reflect global ice volume changes and is also shown here as a model proxy of the ice volume. The obliquity components of the CO2 and global SLP (reflecting the ice volume) forcings are explicitly shown in(d) and (f) in dark orange and dark green, respectively.
Citation: Journal of Climate 36, 12; 10.1175/JCLI-D-22-0587.1
External forcings used in the 300-kyr accelerated transient experiments, including (a) eccentricity, (b) obliquity (°), (c) precession index, (d) CO2 concentration (ppm), and (e) global ice volume (106 km3) during the past 300 kyr. (f) The globally averaged SLP (hPa) in the OGG run is adjusted to reflect global ice volume changes and is also shown here as a model proxy of the ice volume. The obliquity components of the CO2 and global SLP (reflecting the ice volume) forcings are explicitly shown in(d) and (f) in dark orange and dark green, respectively.
Citation: Journal of Climate 36, 12; 10.1175/JCLI-D-22-0587.1
External forcings used in the 300-kyr accelerated transient experiments, including (a) eccentricity, (b) obliquity (°), (c) precession index, (d) CO2 concentration (ppm), and (e) global ice volume (106 km3) during the past 300 kyr. (f) The globally averaged SLP (hPa) in the OGG run is adjusted to reflect global ice volume changes and is also shown here as a model proxy of the ice volume. The obliquity components of the CO2 and global SLP (reflecting the ice volume) forcings are explicitly shown in(d) and (f) in dark orange and dark green, respectively.
Citation: Journal of Climate 36, 12; 10.1175/JCLI-D-22-0587.1
Three long-term reconstructions representing East Asian monsoon variations are used for comparison with the simulations. The reconstructions are the composite δ18O from Chinese Caves over the past 640 kyr representing EASM variations (Cheng et al. 2016), and loess data over the past 800 kyr from Xifeng, CLP, in which the loess magnetic susceptibility is used to represent EASM variations and the loess grain size (the contents of the >32-μm fraction) is used to represent EAWM variations (Guo et al. 2009). We also collected the averaged benthic δ18O over North Atlantic sites covering the past 800 kyr (Lisiecki and Raymo 2009, hereafter LR09) to represent the timing of ice volume changes. The reconstruction data are obtained from the National Centers for Environmental Information (NCEI), National Oceanic and Atmospheric Administration (NOAA; https://www.ncei.noaa.gov/access/paleo-search/; last accessed 26 March 2022).
The temporal resolution of the model output is 100 years. We first interpolate the time series to the 1-kyr resolution using the piecewise linear interpolation method. To obtain the variability at the obliquity band (approximately 40 kyr) and precession band (approximately 20 kyr) (Berger and Loutre 1991), we apply a fast Fourier transform (FFT) bandpass filter to the interpolated time series of 1-kyr resolution. The lower and upper cutoff frequencies are 0.02 and 0.03 for the obliquity band and 0.03 and 0.1 for the precession band.
In this study, the correlation coefficient between two time series is calculated based on the filtered time series, of which the sample sizes are relatively small. To check whether the correlation coefficients are statistically significant, we use a Monte Carlo significance test method. The Monte Carlo significance test consists of a comparison between the observed data and random samples generated in accordance with the hypothesis being tested (Hope 1968). The significance values are calculated in accordance with the confidence intervals based on 1000 correlation coefficients of the randomly generated red noise time series. The red noise series are produced based on the lag-1 autocorrelation of the filtered series and white noise series. The white noise series are randomly generated with the same means and standard deviations as the filtered time series.
b. Calendar correction
According to Kepler’s second law of planetary motion, the lengths of seasons vary with orbital parameters. Comparing the monthly insolation using the present calendar for different paleo periods can be misleading (Joussaume and Braconnot 1997) because it causes an artificial phase shift in the monthly insolation forcing and, in turn, the monthly climate responses (Chen et al. 2011). Therefore, a paleo calendar is required for the analysis of climate changes at orbital time scales. There are two ways to define a calendar: One is the fixed-day calendar (Chen et al. 2011; Timm et al. 2008; Joussaume and Braconnot 1997; Bartlein and Shafer 2019), in which the days of each month are fixed for any historical period as in the present. The other is the fixed-angular calendar (Chen et al. 2011), in which the degrees of a month during which Earth sweeps along its orbit are fixed at 30°. In paleoclimate simulation studies, the use of a fixed-day calendar induces biased temporal characteristics of seasonal climate changes at orbital scales, especially for long time scales, such as the obliquity time scale. Therefore, it is more appropriate to use the fixed-angular calendar to study transient processes at long orbital time scales (Timm et al. 2008; Chen et al. 2011; Bartlein and Shafer 2019).
In this study, we first interpolate the standard monthly mean model outputs, which are saved on the fixed-day calendar, to pseudodaily data with a simple linear interpolation method. Then, we aggregate the pseudodaily data to monthly data on a fixed-angular calendar to obtain calendar-corrected monthly mean data. The months (including the month length and the beginning, middle, and ending days of each month) on a fixed-angular calendar over the past 300 kyr with 1-kyr resolution are calculated using the scripts provided by Bartlein and Shafer (2019) (https://github.com/pjbartlein/PaleoCalAdjust; last accessed 16 January 2022).
3. Spatial distributions of the EASM and EAWM in response to obliquity forcing
We first examine the response patterns of the EASM and EAWM under pure orbital forcing. When the obliquity increases, the incoming solar radiation at the top of the atmosphere (SOLIN) increases in the summer hemisphere and decreases in the winter hemisphere, leading to enhanced seasonality, as shown in the seasonal insolation difference between high and low obliquity (Fig. 2d) (Wu and Tsai 2020). As expected, the surface temperatures (TSs) increase in boreal summer and decrease in boreal winter, more so over the Eurasian continent lands (Figs. 2e and 3a,c) than over the nearby oceans due to the relatively small heat capacity of land. This leads to an enhanced land–sea thermal contrast and, in turn, a strengthened land–sea surface pressure gradient (Figs. 3a,c). This strengthened zonal pressure gradient then enhances the northerly EAWM wind and southerly EASM wind (Figs. 2f and 3b,d), according to geostrophic balance. It is interesting to note that the enhanced EASM circulation is accompanied by a dipole precipitation response of “dry south/wet north” over eastern China (Figs. 2f and 3d) because of the enhanced moisture divergence and convergence in the southern–northern region (Liu et al. 2014). This composite analysis shows that both the EAWM and EASM are strengthened during high-obliquity states (Figs. 2e,f and 3), implying a positive EASM–EAWM relationship.
Composite Hovmöller diagrams of (top) global zonal-mean SOLIN, (middle) TS, and (bottom) total precipitation rate (PRECT; shading) and 850-hPa wind field (vectors) over East Asia (110°–120°E) in (a)–(c) low-obliquity stages and (d)–(f) the differences in these factors between high- and low-obliquity states from the ORB run.
Citation: Journal of Climate 36, 12; 10.1175/JCLI-D-22-0587.1
Composite Hovmöller diagrams of (top) global zonal-mean SOLIN, (middle) TS, and (bottom) total precipitation rate (PRECT; shading) and 850-hPa wind field (vectors) over East Asia (110°–120°E) in (a)–(c) low-obliquity stages and (d)–(f) the differences in these factors between high- and low-obliquity states from the ORB run.
Citation: Journal of Climate 36, 12; 10.1175/JCLI-D-22-0587.1
Composite Hovmöller diagrams of (top) global zonal-mean SOLIN, (middle) TS, and (bottom) total precipitation rate (PRECT; shading) and 850-hPa wind field (vectors) over East Asia (110°–120°E) in (a)–(c) low-obliquity stages and (d)–(f) the differences in these factors between high- and low-obliquity states from the ORB run.
Citation: Journal of Climate 36, 12; 10.1175/JCLI-D-22-0587.1
Composite maps of the TS difference (shading) along with the (a),(c) SLP difference (contours) and (b),(d) the total precipitation rate difference (PRECT, shading) along with the horizontal 850-hPa wind field difference (vectors) between high- and low-obliquity states in (a),(b) DJF and (c),(d) JJA from the ORB run. Only values exceeding the 90% confidence level are shown.
Citation: Journal of Climate 36, 12; 10.1175/JCLI-D-22-0587.1
Composite maps of the TS difference (shading) along with the (a),(c) SLP difference (contours) and (b),(d) the total precipitation rate difference (PRECT, shading) along with the horizontal 850-hPa wind field difference (vectors) between high- and low-obliquity states in (a),(b) DJF and (c),(d) JJA from the ORB run. Only values exceeding the 90% confidence level are shown.
Citation: Journal of Climate 36, 12; 10.1175/JCLI-D-22-0587.1
Composite maps of the TS difference (shading) along with the (a),(c) SLP difference (contours) and (b),(d) the total precipitation rate difference (PRECT, shading) along with the horizontal 850-hPa wind field difference (vectors) between high- and low-obliquity states in (a),(b) DJF and (c),(d) JJA from the ORB run. Only values exceeding the 90% confidence level are shown.
Citation: Journal of Climate 36, 12; 10.1175/JCLI-D-22-0587.1
4. Temporal relationship of EASM–EAWM indices at the obliquity time scales
Due to the complexity of the EASM and EAWM monsoon systems, we study the temporal relationship between the EASM and EAWM using multiple monsoon indices. The EASM and EAWM indices are selected according to the regression maps of each variable against the obliquity (Fig. 4) compared to the composite analysis results. For the EASM, we select three indices: we select the JJA mean precipitation over northern China (35°–50°N, 105°–120°E) as the precipitation-based EASM index (EASM_PRECT), the SLP over the 25°–40°N, 70°–110°E region as the SLP-based EASM index (EASM_SLP), and the meridional wind at 850 hPa over eastern China (20°–40°N, 105°–120°E) as the wind-based EASM index (EASM_v850). For the EAWM, we also define three indices: we define the DJF mean TS over eastern China (30°–50°N, 105°–120°E) as the TS-based EAWM index (EAWM_TS), SLP over the 30°–60°N, 130°–150°E region as the SLP-based EAWM index (EAWM_SLP), and the meridional wind at 850 hPa over eastern China (30°–40°N, 105°–120°E) as the wind-based EAWM index (EAWM_v850). We also use the DJF mean SLP difference between land (30°–50°N, 60°–110°E) and ocean (30°–60°N, 130°–150°E) areas as another SLP-based EAWM index (EAWM_dSLP). (Notably, the DJF mean v850 evolves similarly in response to the obliquity as v1000, as can be seen in the comparison of Figs. 4d,g).
Regression maps of (a) the boreal summer total precipitation rate (PRECT), (b) boreal winter TS, (c) summer 850-hPa wind, (d) winter 850-hPa wind, (e) summer SLP, (f) winter SLP, and (g) winter 1000-hPa wind against obliquity from the ORB run. The red boxes indicate the areas used to define monsoon indices. Only values exceeding the 90% confidence level are shown.
Citation: Journal of Climate 36, 12; 10.1175/JCLI-D-22-0587.1
Regression maps of (a) the boreal summer total precipitation rate (PRECT), (b) boreal winter TS, (c) summer 850-hPa wind, (d) winter 850-hPa wind, (e) summer SLP, (f) winter SLP, and (g) winter 1000-hPa wind against obliquity from the ORB run. The red boxes indicate the areas used to define monsoon indices. Only values exceeding the 90% confidence level are shown.
Citation: Journal of Climate 36, 12; 10.1175/JCLI-D-22-0587.1
Regression maps of (a) the boreal summer total precipitation rate (PRECT), (b) boreal winter TS, (c) summer 850-hPa wind, (d) winter 850-hPa wind, (e) summer SLP, (f) winter SLP, and (g) winter 1000-hPa wind against obliquity from the ORB run. The red boxes indicate the areas used to define monsoon indices. Only values exceeding the 90% confidence level are shown.
Citation: Journal of Climate 36, 12; 10.1175/JCLI-D-22-0587.1
Under orbital forcing (ORB run), most of the filtered time series of monsoon indices in the obliquity band covary with the obliquity, including all the EASM indices and the EAWM indices of EAWM_TS and EAWM_SLP (Fig. 5a). Therefore, the intensities of the EASM and EAWM are almost in phase at the obliquity time scales, with most of the correlations being statistically significant at the 90% level, except for EAWM_v850 (Table 1). The correlation coefficients between each monsoon index and the obliquity are also found to be significantly positive, except for EAWM_v850. The less significant phase relationship of EAWM_v850 might be related to the large noise in the wind field related to synoptic eddies.
Time series of each monsoon index, along with the obliquity (yellow) and precession index (green) from the (a) ORB run, (b) ORC run, and (c) OGG runs. The red lines indicate the EASM indices, and the blue lines indicate the EAWM indices. Time series with a 1-kyr resolution are shown in light colors, with the precession band results shown in relatively darker colors; the obliquity band results are shown with the darkest colors.
Citation: Journal of Climate 36, 12; 10.1175/JCLI-D-22-0587.1
Time series of each monsoon index, along with the obliquity (yellow) and precession index (green) from the (a) ORB run, (b) ORC run, and (c) OGG runs. The red lines indicate the EASM indices, and the blue lines indicate the EAWM indices. Time series with a 1-kyr resolution are shown in light colors, with the precession band results shown in relatively darker colors; the obliquity band results are shown with the darkest colors.
Citation: Journal of Climate 36, 12; 10.1175/JCLI-D-22-0587.1
Time series of each monsoon index, along with the obliquity (yellow) and precession index (green) from the (a) ORB run, (b) ORC run, and (c) OGG runs. The red lines indicate the EASM indices, and the blue lines indicate the EAWM indices. Time series with a 1-kyr resolution are shown in light colors, with the precession band results shown in relatively darker colors; the obliquity band results are shown with the darkest colors.
Citation: Journal of Climate 36, 12; 10.1175/JCLI-D-22-0587.1
Correlation coefficients between each monsoon index at the obliquity time scales, along with the correlation coefficients with obliquity and precession derived from the ORB run. An asterisk (*) indicates the correlation coefficient is above the 90% confidence level, as determined by applying the Monte Carlo test.
Despite the overall positive correlations, some small phase shifts among monsoon indices of up to 5 kyr are observed. One notable example is the relationship between EASM_PRECT and EAWM_TS at the obliquity time scales, which has a phase difference of −40°, or approximately 5 kyr in the coherence analysis results (Fig. 6a), with the EASM_PRECT lagging the EAWM_TS. This phase shift is confirmed in the lead–lag correlation between EASM_PRECT and EAWM_TS (Fig. 6b). These phase shifts indicate potential differences in the evolutionary responses of various seasonal monsoon indices to obliquity-induced insolation.
(a) Coherence (blue lines) and phase (green dots) between the EASM_PRECT and EAWM_TS and (b) lead–lag correlations between the EASM_PRECT and EAWM_TS in the ORB run. In (a), the black line indicates the 90% confidence level for the coherence square. The green dashed line denotes 0° of phase difference. The phases in which the coherence square passed the 90% confidence level are marked with red circles.
Citation: Journal of Climate 36, 12; 10.1175/JCLI-D-22-0587.1
(a) Coherence (blue lines) and phase (green dots) between the EASM_PRECT and EAWM_TS and (b) lead–lag correlations between the EASM_PRECT and EAWM_TS in the ORB run. In (a), the black line indicates the 90% confidence level for the coherence square. The green dashed line denotes 0° of phase difference. The phases in which the coherence square passed the 90% confidence level are marked with red circles.
Citation: Journal of Climate 36, 12; 10.1175/JCLI-D-22-0587.1
(a) Coherence (blue lines) and phase (green dots) between the EASM_PRECT and EAWM_TS and (b) lead–lag correlations between the EASM_PRECT and EAWM_TS in the ORB run. In (a), the black line indicates the 90% confidence level for the coherence square. The green dashed line denotes 0° of phase difference. The phases in which the coherence square passed the 90% confidence level are marked with red circles.
Citation: Journal of Climate 36, 12; 10.1175/JCLI-D-22-0587.1
Kutzbach et al. (2008) suggested that the phase shifts between the evolutionary responses of monsoons to orbital insolation might depend upon the relative importance of thermal–dynamic and hydrologic–dynamic processes during the monsoon process. Taking the EASM_PRECT and EAWM_TS as examples, the TS-based EAWM is found to be related to the thermodynamic process, whereas the precipitation-based EASM is related to the hydrologic process. The lead–lag correlation illustrates that the EAWM_TS leads obliquity by approximately 2 kyr, while the EASM_PR lags obliquity by approximately 2 kyr. This difference might induce a phase shift between EASM_PR and EAWM_TS of approximately 5 kyr.
Next, we investigate the modulation effects of greenhouse gases on the EASM–EAWM relationship at the obliquity time scales by additionally considering greenhouse gas forcings in the ORC run. At the obliquity time scale, the monsoon responses in the ORC run (Fig. 5b) are similar overall to those in the ORB run (Fig. 5a), indicating the small influence of greenhouse gases at the obliquity time scale. This negligible impact of greenhouse gases may be caused by the greenhouse gas forcing component being too small at the obliquity scale, at approximately 32% of the total amplitude and only approximately 14% of the total forcing variability, as was discussed in section 1 and Fig. 1.
When the ice sheet forcing is imposed in the OGG run, the significant positive correlation between the EASM and EAWM still holds (Fig. 5c, Table 2). Moreover, the monsoon responses also show some visible differences from the ORB and ORC runs, indicating a notable role of the ice sheet. In the OGG run, due to the global mass adjustment associated with the continental ice sheet volume change on the SLP, as was introduced in section 2a, the SLP value itself is no longer used as a monsoon index. Therefore, in the OGG run, we utilize only four monsoon indices: the EASM_PRECT, EASM_v850, EAWM_TS, and EAWM_v850.
As in Table 1, but for the OGG run. An asterisk (*) indicates the correlation coefficient is above the 90% confidence level, as determined by applying the Monte Carlo test.
In the OGG run, the correlation coefficients between the EASM indices and EAWM_v850 become significant. However, the correlation coefficient between the EAWM_v850 and obliquity is still nonsignificant (Table 2). We hypothesize that the significant relationships between the EASM indices and EAWM_v850 are related to the ice sheet topography, which has a notable impact on East Asian circulation (Gao et al. 2020). On the other hand, the phase shifts between the EASM and EAWM indices, with the EASM lagging behind the EAWM, at the obliquity time scales become larger in the OGG run (Fig. 5c). Again, we hypothesize that these phase shifts might be caused by the existence of ice sheets. We will focus on these phase shifts in the following section.
5. Phase responses of EASM and EAWM at obliquity time scales
As mentioned above, the evolutions of the East Asian monsoon exhibit phase shifts from the forcings, and these phase shifts are larger in the OGG run than in the ORB run, especially in the circulation-based monsoon indices. In this section, we focus on these phase shifts and the implied roles of external forcings on East Asian monsoon evolutions at the obliquity time scales using four monsoon indices: the EASM_PRECT, EASM_v850, EAWM_TS, and EAWM_v850.
The phase shifts of monsoon indices at the obliquity time scales are presented in the phase-wheel diagram (Fig. 7), with the baseline phase 0 as the maximum summer monsoon forcing (maximum June insolation) or, equivalently, the maximum winter monsoon forcing (minimum December insolation) on the top atmosphere at 45°N (Clemens and Prell 2003; Clemens et al. 2021). In the ORB run (Fig. 7a), all the monsoon indices evolve closely in phase with the monsoon forcing at the obliquity scale. The EASM indices are in phase with the June insolation variation while the EAWM indices are in phase with the December insolation, with the phase differences ranging between approximately ±20° [or approximately 2 kyr ≈ (20°/360°) × 41 kyr]. Therefore, under orbital forcing alone, the EASM and EAWM show relatively coherent variations at the obliquity time scales in phase with the orbital forcing.
Phase wheels representing phases between the monsoon indices and 45°N June insolation maximum at obliquity time scales from the (a) ORB, (b) ORC, and (c) OGG runs. The phases between reconstructions and insolation are marked with stars. The phases between simulated indices are marked with circles. The EASM indices are shown in red, the EAWM indices are shown in blue, and the ice volume change indices are shown in green. The ice volume changes are represented by the global SLP in the simulation and benthic δ18O (LR09) in the reconstruction. CO2 is shown in orange. The numbers in parentheses indicate the phases. GS in each phase wheel refers to the loess grain size, and MSUS refers to the loess magnetic susceptibility.
Citation: Journal of Climate 36, 12; 10.1175/JCLI-D-22-0587.1
Phase wheels representing phases between the monsoon indices and 45°N June insolation maximum at obliquity time scales from the (a) ORB, (b) ORC, and (c) OGG runs. The phases between reconstructions and insolation are marked with stars. The phases between simulated indices are marked with circles. The EASM indices are shown in red, the EAWM indices are shown in blue, and the ice volume change indices are shown in green. The ice volume changes are represented by the global SLP in the simulation and benthic δ18O (LR09) in the reconstruction. CO2 is shown in orange. The numbers in parentheses indicate the phases. GS in each phase wheel refers to the loess grain size, and MSUS refers to the loess magnetic susceptibility.
Citation: Journal of Climate 36, 12; 10.1175/JCLI-D-22-0587.1
Phase wheels representing phases between the monsoon indices and 45°N June insolation maximum at obliquity time scales from the (a) ORB, (b) ORC, and (c) OGG runs. The phases between reconstructions and insolation are marked with stars. The phases between simulated indices are marked with circles. The EASM indices are shown in red, the EAWM indices are shown in blue, and the ice volume change indices are shown in green. The ice volume changes are represented by the global SLP in the simulation and benthic δ18O (LR09) in the reconstruction. CO2 is shown in orange. The numbers in parentheses indicate the phases. GS in each phase wheel refers to the loess grain size, and MSUS refers to the loess magnetic susceptibility.
Citation: Journal of Climate 36, 12; 10.1175/JCLI-D-22-0587.1
After the imposition of greenhouse gases, among the four simulated monsoon indices, only the EAWM_TS exhibits a slight shift of approximately 15° [or approximately 1.7 kyr ≈ (15°/360°) × 41 kyr] toward the CO2 minimum in the ORC run compared to the ORB run (Figs. 7a,b). This means that greenhouse gases may have a slight modulating effect on winter temperature variabilities in response to orbital-induced insolation but have little impact on summer monsoon variabilities, especially monsoon-circulation variabilities. The simulated EASM–EAWM relationship in the ORC run still shows relatively coherent variations at the obliquity time scales.
In the OGG run, after the imposition of the ice sheet forcing, the phases between the monsoon indices and monsoon forcing exhibit notable changes (Fig. 7c), especially the winter monsoon index. The winter monsoon variability shifts notably toward the ice sheet maximum (Fig. 7c), with the phase lead more than doubling from approximately 20° in the ORB run to approximately 40°–50° in the OGG run. This EAWM shift toward the ice sheet maximum was expected in previous studies (e.g., Clemens et al. 2008), although the magnitude of the phase shift in the current model is still smaller than that in observations, as inferred from loess grain-size data. It is interesting that the EASM_PRECT index also exhibits a shift. In the ORB and ORC runs, the EASM_PRECT and EASM_v850 indices show the same responses to insolation, slightly lagging the June insolation. In the OGG run, however, the EASM_v850 index still largely follows the June insolation variations, but the EASM_PRECT index leads the June insolation by 25°. This phase change may indicate the sensitivity of regional monsoon precipitation to different external forcings at the obliquity time scale.
Finally, we compare our simulations with the reconstructions and discuss the implication of our simulation work to the reconstruction work. The phases between the EASM_v850 and insolation in the simulations (the ORB and ORC runs) are the same as those between the composite speleothem δ18O and insolation, lagging the June insolation by approximately 20°. This finding is consistent with the notion that the speleothem δ18O over East China is a good proxy for the EASM circulation (Liu et al. 2014; He et al. 2021; Cheng et al. 2021) and is forced mainly by orbital-induced insolation.
In the reconstruction, the phase difference between the composite δ18O (representing the EASM; Cheng et al. 2016) and the loess grain size (representing the EAWM; Guo et al. 2009) with regard to insolation is approximately 123°. The direct phase between the composite δ18O and loess grain size is approximately 107° [or approximately 12 kyr ≈ (107°/360°) × 41 kyr], close to the phase difference. Whereas in the loess records, the phase difference and the direct phase between the loess magnetic susceptibility and loess grain size are almost the same, at approximately 175°. The slight difference (e.g., 123° − 107° = 16°) might have been related to the calculation method. The phase difference is calculated by obtaining the difference between the separate phases relative to insolation. The direct phase is obtained from the coherence between the two variables, that is, the composite δ18O and the loess grain size. To determine their direct relationship, we discuss the direct phase between the EASM and EAWM indices in the following section.
In the ORB run, the direct phase between the EASM_v850 and EAWM_v850 indices at the obliquity scale, illustrated by coherence analysis, is approximately 60° [or approximately 7 kyr ≈ (60°/360°) × 41 kyr]. In the OGG run, the phase difference between the EASM_v850 and EAWM_v850 indices is approximately 102° [or approximately 11 kyr ≈ (102°/360°) × 41 kyr], much closer to the reconstruction. This indicates an obvious modulation effect of the ice sheet. The phase shift is still observed between the EASM_v850 and EAWM_TS indices (not shown). However, the summer precipitation–based EASM is positively correlated with the winter TS at lag 0 and shows a much weaker correlation with the winter circulation index (not shown). This provides another piece of evidence that δ18O represents monsoon circulation instead of precipitation itself.
Our findings are qualitatively consistent with the reconstructions of Clemens et al. (2008) and Sun et al. (2010). Their proxy studies suggested that at the obliquity time scales, EASM and EAWM variations were in phase prior to the great NH glacial cycles that occurred in the early period of 2.75–3.15 Ma but out of phase in the later period dominated by NH glacial cycles. Here, our ORB run can be considered an approximation of the early period prior to the NH glacial cycle, while the OGG run can be considered for the later period of NH glacial cycles. Therefore, the positive EASM–EAWM relationship in the ORB run is consistent with the reconstructed in-phase variations in the earlier period before the NH glacial cycles, and the expanded phase difference between the EASM and EAWM in the OGG run shifts toward the reconstructed out-of-phase relationship in the later-period NH glacial cycle. The ice sheet forcing likely increases the phase difference mainly by shifting the EAWM earlier toward the maximum ice volume (Clemens et al. 2008).
Notably, the abovementioned, reconstructed, out-of-phase (∼180°) EASM–EAWM variation was based on the loess data from the CLP (Clemens et al. 2008; Sun et al. 2010). However, the phase between the composite δ18O-based EASM and loess grain-size–based EAWM variation is approximately 107°, which is also true for that between the simulated EASM_v850 and EAWM_v850 in the OGG run. Moreover, in the composite δ18O-EASM and loess grain-size–EAWM and in the simulated EASM and EAWM, differences are observed between the phase difference and direct phase, whereas in the loess magnetic susceptibility–EASM and loess grain-size–EAWM, almost no difference between the phase difference and direct phase is observed. This raises questions about the regional monsoon climate response, a topic to be further studied in the future through both models and observations.
On the other hand, the phases in the simulation between the EASM indices and insolation are consistent with those between the composite δ18O and insolation but slightly different from those between the loess magnetic susceptibility and insolation. This discrepancy between the simulated EASM indices and loess magnetic susceptibility (MSUS) is yet another interesting topic that needs to be investigated in the future.
6. Comparison of the obliquity and precessional time scales
Consistent with previous studies of the EASM–EAWM relationship performed at the precessional time scales (Wen et al. 2016; Yan et al. 2020), our simulated EASM and EAWM show a robust positive correlation at the precessional time scale, even with the ice sheet forcing considered in the OGG run (Fig. 8, Table 3). The robust, positive correlations found between the EASM and EAWM at both the precessional and obliquity time scales, especially in response to orbital forcing only, imply a consistent response mechanism of the EASM and EAWM to orbital forcing at both time scales. This is theoretically expected because monsoon indices mainly follow orbital-induced insolation, and both obliquity and precession may lead to enhanced or weakened seasonality of the NH insolation (Figs. 2d and 8a). This type of seasonal forcing, increasing (decreasing) insolation in summer and decreasing (increasing) insolation in winter, leads to a positive EASM–EAWM relationship.
Composite Hovmöller diagrams of (a) the global zonal-mean SOLIN difference, (b) global zonal-mean TS difference, (c) East Asian total precipitation rate (PRECT; shading) difference and 850-hPa wind field (vectors) difference between the low- and high-precession stages from the ORB run.
Citation: Journal of Climate 36, 12; 10.1175/JCLI-D-22-0587.1
Composite Hovmöller diagrams of (a) the global zonal-mean SOLIN difference, (b) global zonal-mean TS difference, (c) East Asian total precipitation rate (PRECT; shading) difference and 850-hPa wind field (vectors) difference between the low- and high-precession stages from the ORB run.
Citation: Journal of Climate 36, 12; 10.1175/JCLI-D-22-0587.1
Composite Hovmöller diagrams of (a) the global zonal-mean SOLIN difference, (b) global zonal-mean TS difference, (c) East Asian total precipitation rate (PRECT; shading) difference and 850-hPa wind field (vectors) difference between the low- and high-precession stages from the ORB run.
Citation: Journal of Climate 36, 12; 10.1175/JCLI-D-22-0587.1
As in Table 1, but for precessional time scales. An asterisk (*) indicates the correlation coefficient is above the 90% confidence level, as determined by applying the Monte Carlo test.
Quantitatively, the responses of the monsoon climate to obliquity forcing are weaker than those to precessional forcing (Fig. 2 versus Fig. 8). This is expected because the changes in insolation seasonality forced by obliquity and precession are qualitatively similar, but those forced by obliquity have smaller magnitudes (Figs. 2d and 8a).
It is also interesting to note that in our simulations, the obliquity and precession signals are more mutually comparable in winter than in summer, except for the SLP-based indices, reflecting the greater response of the EAWM than the EASM to obliquity (Fig. 5). Specifically, the precession signal is much stronger than the obliquity signal in the EASM indices but becomes almost comparable with the obliquity signal in the EAWM indices. For example, the ratios of the standard deviation between the obliquity band and precession band are 0.39 and 0.28 for EASM_PRECT and EASM_v850, respectively, while these ratios increased by ∼10% to 0.47 and 0.45 for EAWM_TS and EAWM_v850, respectively (Table 4). These quantitative changes seem to be consistent with the recent reconstruction work by Sun et al. (2021). In their study, the authors decomposed the speleothem δ18O record (representing the EASM) and loess grain-size stack (representing the EAWM) into three orbital components. Although the spectral density at the precession cycle was much higher than that at the obliquity cycle in δ18O, their spectral densities became comparable with regard to the loess grain size (Fig. 6 in Sun et al. 2021). These seasonal changes in the ratio are related mainly to insolation. The ratio of the simulated SOLIN at 45°N between the obliquity band and precession band changes from 0.2 in JJA to 0.5 in DJF (Table 4).
Ratios of the standard deviation between the obliquity band and precession band of each variable.
7. Conclusions
Wen et al. (2016) and Yan et al. (2020) discussed the EASM–EAWM relationship at multidecadal to precessional time scales and found that their relationship changed from a negative correlation at multidecadal to millennial time scales to a positive correlation at the precessional time scales. In this study, we extend their study of the EASM–EAWM relationship to the obliquity time scales using a set of experiments spanning the past 300 kyr. Our results show that, forced by the enhanced seasonality of insolation associated with obliquity changes, the EASM and EAWM are positively and robustly correlated across different indices, mainly in response to insolation forcing. Moreover, ice sheets play a notable modulating role in the responses of the EAWM circulation to obliquity forcing. As a result, the circulation-based EASM–EAWM relationship in the OGG run exhibits an increased phase difference that is qualitatively consistent with the reconstructed EASM–EAWM relationship. The detailed mechanism behind the modulating effect of ice sheets on the East Asian monsoon evolution is beyond the scope of this study and will be investigated in future studies.
In line with our previous works, here we propose that the phase relationship between the EASM and EAWM depends critically on the forcing, with a negative correlation found in response to a dominant annual forcing and a positive correlation found in response to a dominant seasonal forcing. Annual forcings can influence the surface temperature throughout the year in the same direction and tend to have opposite effects on the summer and winter monsoon intensities. For example, a weakened AMOC cools East Asia in both summer and winter, and this annual forcing leads to a negative EASM–EAWM relationship (Wen et al. 2016; Yan et al. 2020). In contrast, seasonal forcings change temperatures in the opposite directions in summer and winter, thereby exerting the same effect on the summer and winter monsoon intensities. For example, a precession minimum (or high obliquity) leads to warmer summers and cooler winters in East Asia and thus a positive EASM–EAWM relationship, as we show in the present study (see also Wen et al. 2016). These findings have implications for better understanding the factors controlling Asian climate evolution revealed by paleoproxies.
This work raises questions regarding other important aspects of orbital-scale paleomonsoon variations that should be considered in the future, that is, the phases and/or phase differences between monsoon indices (and proxies) and orbital-induced insolation and those between summer and winter monsoon indices (and proxies). For proxy variations in particular, the phase differences measured when using different proxies might be dependent on the chronology methods employed. Therefore, the phases and/or phase differences measured from different indices and proxies and the mechanisms behind these phase differences require further studies of both modeling and reconstruction methods.
Acknowledgments.
We thank Drs. Kang Shugang and Shi Zhengguo for helpful discussions. We acknowledge three anonymous reviewers for their constructive suggestions that helped improve the manuscript. This work is jointly supported by the Science and Technology Innovation Project of Laoshan Laboratory (Grant LSKJ202203303), the National Natural Science Foundation of China (Grants 42130604, 42075049, 41971108, 41971021, and 42111530182), and the Priority Academic Program Development of Jiangsu Higher Education Institutions (Grant 164320H116). The authors declare that there are no conflicts of interest regarding the publication of this article.
Data availability statement.
The data analyzed in this study were a reanalysis of existing data. The time series of model-based monsoon indices used in this paper, the code used to perform the Monte Carlo significance test, and the tool applied for FFT bandpass filtering can be found at https://doi.org/10.5281/zenodo.7349651.
REFERENCES
An, Z., 2000: The history and variability of the East Asian paleomonsoon climate. Quat. Sci. Rev., 19, 171–187, https://doi.org/10.1016/S0277-3791(99)00060-8.
Bartlein, P. J., and S. L. Shafer, 2019: Paleo calendar-effect adjustments in time-slice and transient climate-model simulations (PaleoCalAdjust v1.0): Impact and strategies for data analysis. Geosci. Model Dev., 12, 3889–3913, https://doi.org/10.5194/gmd-12-3889-2019.
Berger, A., and M. F. Loutre, 1991: Insolation values for the climate of the last 10 million years. Quat. Sci. Rev., 10, 297–317, https://doi.org/10.1016/0277-3791(91)90033-Q.
Chen, G.-S., J. E. Kutzbach, R. Gallimore, and Z. Liu, 2011: Calendar effect on phase study in paleoclimate transient simulation with orbital forcing. Climate Dyn., 37, 1949–1960, https://doi.org/10.1007/s00382-010-0944-6.
Cheng, H., and Coauthors, 2016: The Asian monsoon over the past 640,000 years and ice age terminations. Nature, 534, 640–646, https://doi.org/10.1038/nature18591.
Cheng, H., and Coauthors, 2021: Orbital-scale Asian summer monsoon variations: Paradox and exploration. Sci. China Earth Sci., 64, 529–544, https://doi.org/10.1007/s11430-020-9720-y.
Clemens, S. C., and W. L. Prell, 2003: A 350,000 year summer-monsoon multi-proxy stack from the Owen Ridge, northern Arabian Sea. Mar. Geol., 201, 35–51, https://doi.org/10.1016/S0025-3227(03)00207-X.
Clemens, S. C., W. L. Prell, Y. Sun, Z. Liu, and G. Chen, 2008: Southern Hemisphere forcing of Pliocene δ18O and the evolution of Indo-Asian monsoons. Paleoceanogr. Paleoclimatol., 23, PA4210, https://doi.org/10.1029/2008PA001638.
Clemens, S. C., and Coauthors, 2021: Remote and local drivers of Pleistocene South Asian summer monsoon precipitation: A test for future predictions. Sci. Adv., 7, eabg3848, https://doi.org/10.1126/sciadv.abg3848.
Gao, Y., Z. Liu, and Z. Lu, 2020: Dynamic effect of last glacial maximum ice sheet topography on the East Asian summer monsoon. J. Climate, 33, 6929–6944, https://doi.org/10.1175/JCLI-D-19-0562.1.
Guo, Z., A. Berger, Q. Z. Yin, and L. Qin, 2009: Strong asymmetry of hemispheric climates during MIS-13 inferred from correlating China loess and Antarctica ice records. Climate Past, 5, 21–31, https://doi.org/10.5194/cp-5-21-2009.
He, C., and Coauthors, 2021: Hydroclimate footprint of pan-Asian monsoon water isotope during the last deglaciation. Sci. Adv., 7, eabe2611, https://doi.org/10.1126/sciadv.abe2611.
Hope, A. C. A., 1968: A simplified Monte Carlo significance test procedure. J. Roy. Stat. Soc., 30B, 582–598, https://doi.org/10.1111/j.2517-6161.1968.tb00759.x.
Joussaume, S., and P. Braconnot, 1997: Sensitivity of paleoclimate simulation results to season definitions. J. Geophys. Res. Atmos., 102, 1943–1956, https://doi.org/10.1029/96JD01989.
Kang, S., X. Wang, H. M. Roberts, G. A. T. Duller, P. Cheng, Y. Lu, and Z. An, 2018: Late Holocene anti-phase change in the East Asian summer and winter monsoons. Quat. Sci. Rev., 188, 28–36, https://doi.org/10.1016/j.quascirev.2018.03.028.
Kang, S., J. Du, N. Wang, J. Dong, D. Wang, X. Wang, X. Qiang, and Y. Song, 2020: Early Holocene weakening and mid- to late Holocene strengthening of the East Asian winter monsoon. Geology, 48, 1043–1047, https://doi.org/10.1130/G47621.1.
Kutzbach, J. E., X. Liu, Z. Liu, and G. Chen, 2008: Simulation of the evolutionary response of global summer monsoons to orbital forcing over the past 280,000 years. Climate Dyn., 30, 567–579, https://doi.org/10.1007/s00382-007-0308-z.
Lisiecki, L. E., and M. E. Raymo, 2009: Diachronous benthic δ18O responses during late Pleistocene terminations. Paleoceanography, 24, PA3210, https://doi.org/10.1029/2009PA001732.
Liu, Z. Y., and Coauthors, 2014: Chinese cave records and the East Asia Summer Monsoon. Quat. Sci. Rev., 83, 115–128, https://doi.org/10.1016/j.quascirev.2013.10.021.
Lu, Z., and Z. Liu, 2018: Orbital modulation of ENSO seasonal phase locking. Climate Dyn., 52, 4329–4350, https://doi.org/10.1007/s00382-018-4382-1.
Sun, Y., Z. An, S. C. Clemens, J. Bloemendal, and J. Vandenberghe, 2010: Seven million years of wind and precipitation variability on the Chinese Loess Plateau. Earth Planet. Sci. Lett., 297, 525–535, https://doi.org/10.1016/j.epsl.2010.07.004.
Sun, Y., S. C. Clemens, F. Guo, X. Liu, Y. Wang, Y. Yan, and L. Liang, 2021: High-sedimentation-rate loess records: A new window into understanding orbital- and millennial-scale monsoon variability. Earth-Sci. Rev., 220, 103731, https://doi.org/10.1016/j.earscirev.2021.103731.
Timm, O., A. Timmermann, A. Abe-Ouchi, F. Saito, T. Segawa, 2008: On the definition of seasons in paleoclimate simulations with orbital forcing. Paleoceanography, 23, PA2221, https://doi.org/10.1029/2007PA001461.
Wen, X., and Coauthors, 2016: Correlation and anti-correlation of the East Asian summer and winter monsoons during the last 21,000 years. Nat. Commun., 7, 11999, https://doi.org/10.1038/ncomms11999.
Wu, C.-H., and P.-C. Tsai, 2020: Obliquity-driven changes in East Asian seasonality. Global Planet. Change, 189, 103161, https://doi.org/10.1016/j.gloplacha.2020.103161.
Yan, M., Z. Y. Liu, L. Ning, and J. Liu, 2020: Holocene EASM-EAWM relationship across different timescales in CCSM3. Geophys. Res. Lett., 47, e2020GL088451, https://doi.org/10.1029/2020GL088451.
Zhu, X., and Coauthors, 2021: Simulation of the relationship between East Asian winter and summer monsoon in three typical periods over the past millennium (in Chinese). Quat. Sci., 41, 537–551.
