Intra-interglacial climate variability from Marine Isotope Stage 15 to the Holocene

Introduction Conclusions References Tables Figures


Intra-interglacial climate variability from
The Quaternary period is characterized by the cyclic growth and decay of continental ice sheets associated with global environmental changes (e.g., Lisiecki and Raymo, 2005;Tzedakis et al., 2006;Jouzel et al., 2007;Lang and Wolff, 2011).While it is commonly accepted that the transitions between glacial and interglacial stages are ultimately triggered by varying orbital insolation (Hays et al., 1976), climate research is just beginning to understand the internal climate feedbacks that are required to shift the Earth system from one state to the other (e.g., van Nes et al., 2015).The astronomical forcing, with its characteristic periods of ca.(obliquity), and ca.19 and 23 kyr (precession), also acts as an external driver for longterm climate change within the interglacials (i.e. the long-term intra-interglacial climate variability) and likely contributes to interglacial diversity since the evolution of orbital parameters differs between all Quaternary interglacial stages (cf.Tzedakis et al., 2009).Understanding both interglacial climate diversity and intra-interglacial variability helps to estimate the sensitivity of the Earth system to different forcings and to assess the rate and magnitude of current climate change relative to natural variability.
While the present and the last interglacial have been extensively investigated with fully coupled atmosphere-ocean general circulation models (e.g., Braconnot et al., 2007;Lunt et al., 2013), earlier interglacial periods have received much less attention by climate modellers.Coupled general circulation model (CGCM) studies of earlier interglacial climates have recently been performed for Marine Isotope Stage (MIS) 11 (Milker et al., 2013;Kleinen et al., 2014) and MIS 13 (Muri et al., 2013).Using the CGCM CCSM3 (Community Climate System Model version 3), Herold et al. (2012) presented a set of interglacial climate simulations comprising the interglaciations of MIS 1, 5, 9, 11 and 19.Their study, however, focussed on peak interglacial forcing (i.e.Northern Hemisphere summer occurring at perihelion) and intercomparison of interglacials (i.e.interglacial diversity) only.Here, we present a different and complementary CGCM (CCSM3) study which takes intra-interglacial climate variability into account by simulating two or more time slices for each interglacial stage of MIS 1,5,11,13,and 15.The goal of this study is to disentangle the effects of obliquity, precession and greenhouse gases (GHG) on global surface climate.In contrast to previously performed climate model experiments with idealized orbital forcing, in which obliquity and precession have usually set to extreme values (e.g., Tuenter et al., 2003;Mantsis et al., 2011Mantsis et al., , 2014;;Erb et al., 2013;Bosmans et al., 2015), our analyzes are based on realistic orbital configurations and hence climate states.Special focus is on the sensitivity of the West African and Indian monsoon systems to obliquity and precession forcing.In particular, the applicability of the global monsoon concept (Trenberth et al., 2000;Wang et al., 2014) will be tested for orbital timescales.Introduction

Conclusions References
Tables Figures

Back Close
Full

Model description
We use the fully coupled climate model CCSM3 with the atmosphere, ocean, sea-ice and land-surface components interactively connected by a flux coupler (Collins et al., 2006).We apply the low-resolution version of the model (Yeager et al., 2006) which enables us to simulate a large set of time slices.In this version, the resolution of the atmosphere is given by T31 spectral truncation (3.75 • transform grid) with 26 layers, while the ocean model has a nominal horizontal resolution of 3 • (as has the sea-ice component) with 25 levels in the vertical.The land model shares the same horizontal grid with the atmosphere and includes components for biogeophysics, biogeochemistry, the hydrological cycle as well as a Dynamic Global Vegetation Model (DGVM) based on the Lund-Potsdam-Jena (LPJ)-DGVM (Sitch et al., 2003;Levis et al., 2004;Bonan and Levis, 2006).The DGVM predicts the distribution of 10 plant functional types (PFT) which are differentiated by physiological, morphological, phenological, bioclimatic, and fire-response attributes (Levis et al., 2004).In order to improve the simulation of landsurface hydrology and hence the vegetation cover, new parameterizations for canopy interception and soil evaporation were implemented into the land component (Oleson et al., 2008;Handiani et al., 2013;Rachmayani et al., 2015).PFT population densities are restored annually, while the land and atmosphere models are integrated with a 30 min time step.

Setup of experiments
To serve as a reference climatic state, a standard pre-industrial (PI) control simulation was carried out following PMIP (Paleoclimate Modelling Intercomparison Project) guidelines with respect to the forcing (e.g., Braconnot et al., 2007).The PI boundary conditions include orbital parameters of 1950 AD, atmospheric trace gas concentrations from the 18th century (Table 1) as well as pre-industrial distributions of at-Introduction

Conclusions References
Tables Figures

Back Close
Full mospheric ozone, sulfate aerosols, and carbonaceous aerosols (Otto-Bliesner et al., 2006).The solar constant is set to 1365 W m −2 .The PI control run was integrated for 1000 years starting from modern initial conditions, except for the vegetation which starts from bare soil.In total, 13 interglacial time slice experiments were carried out, all branching off from year 600 of the PI spin-up run and running for 400 years each.Boundary conditions for the selected time slices which are spanning the last 615 kyr comprise orbital parameters (Berger, 1978) and GHG concentrations as given in Table 1, while other forcings (ice sheet configuration, ozone distribution, sulfate aerosols, carbonaceous aerosols, solar constant) were kept as in the PI control run.The mean of the last 100 simulation years of each experiment was used for analysis.

Selection of interglacial time slices
Our selection of interglacial time slices takes into account different aspects of interand intra-interglacial variability and associated astronomical forcing.As such, our approach differs from and complements previous model studies that focussed on peak interglacial forcing and intercomparison of interglacials (Yin and Berger, 2012;Herold et al., 2012).
For MIS 1, the mid-Holocene time slice of 6 kyr BP using standard PMIP forcing (Braconnot et al., 2007) was complemented by an early-Holocene 9 kyr BP simulation when Northern Hemisphere summer insolation was close to maximum (Fig. 1).Two time slices, 125 and 115 kyr BP, were also chosen for the last interglacial (MIS 5e).Similar to 9 kyr BP, the 125 kyr BP time slice is also characterized by nearly peak interglacial forcing, although the MIS 5 insolation forcing is stronger due to a greater eccentricity of the Earth's orbit.Moreover, the global benthic δ 18 O stack is at minimum around 125 kyr BP (Lisiecki and Raymo, 2005).By contrast, boreal summer insolation is close to minimum at 115 kyr BP, which marked the end of MIS 5e (Fig. 1).GHG concentrations for the MIS 5 time slices were taken as specified by PMIP-3 (Lunt et al., 2013).Introduction

Conclusions References
Tables Figures
The time slices of 394 and 416 kyr BP are characterized by almost identical precession and similar GHG concentrations (Table 1), but opposite extremes of obliquity (maximum at 416 kyr BP, minimum at 394 kyr BP; Fig. 1).This allows to study the quasi-isolated effect of obliquity forcing (Berger, 1978) during MIS 11 by directly comparing the results of these two time slices.As opposed to idealized simulations of obliquity forcing (e.g., Tuenter et al., 2003;Mantsis et al., 2011Mantsis et al., , 2014;;Erb et al., 2013) (Yin et al., 2009).In addition, the 504 kyr BP time slice was picked because of peak Northern Hemisphere summer insolation forcing, while obliquity has an intermediate value (Fig. 1).Finally, two time slice experiments were performed for MIS 15 to assess the climatic response to minimum (579 kyr BP) and maximum (609 kyr BP) precession.Accordingly, Northern Hemisphere summer insolation is near maximum and minumum at 579 and 609 kyr BP, respectively.In addition, a third MIS 15 experiment was carried out (615 kyr BP) with insolation forcing in between the two others (Fig. 1).Moreover, the 615 kyr BP time slice has a very special seasonal insolation pattern as we will see in the next section.All three MIS 15 time slices coincide with minimum δ 18 O values (Lisiecki and Raymo, 2005).
Table 1 summarizes  distribution of methane sources and sinks over the latitudes, values of atmospheric CH 4 concentration derived from Antarctic ice cores present a lower estimate of global CH 4 concentration.We further note that some results from the MIS 1 (6 and 9 kyr BP), MIS 5 (125 kyr BP), and MIS 11 (394, 405, and 416 kyr BP) experiments were previously published (Lunt et al., 2013;Milker et al., 2013;Kleinen et al., 2014;Rachmayani et al., 2015).

Insolation anomalies
Annual cycles of the latitudinal distribution of insolation at the top of the atmosphere (as anomalies relative to PI) are shown in Fig. 2 for each experiment.The insolation patterns can be divided into three groups which differ in their seasonal distribution of incoming energy.Group I is characterized by high Northern Hemisphere summer insolation as exhibited for the 6 and 9 kyr BP (MIS 1), 125 kyr BP (MIS 5), 405 and 416 kyr BP (MIS 11), 504 kyr BP (MIS 13), and 579 kyr BP (MIS 15) time slices.In most (but not all, see below) cases this is due to an orbital configuration with northern summer solstice at or close to perihelion.Group II comprises anomalies with low boreal summer insolation as shown for 115 kyr BP (MIS 5), 495 and 516 kyr BP (MIS 13), and 609 kyr BP (MIS 15).In these cases, northern winter solstice is near perihelion.Group III is characterized by changes in the sign of the Northern Hemisphere insolation anomalies from spring to summer and consists of two dates (394 and 615 kyr BP).At 394 (615 kyr BP) the insolation anomaly spring-to-summer change is from positive (negative) to negative Introduction

Conclusions References
Tables Figures

Back Close
Full

JJAS surface temperature anomalies
The response of boreal summer (June-July-August-September, JJAS) surface temperature to the combined effect of insolation and GHG in all individual climates (Fig. 3) shows warm conditions (relative to PI) over most parts of the continents in Group I (6, 9, 125, 405, 416, 504, and 579 kyr BP) with the three warmest anomalies at 9, 125, and 579 kyr BP.The warm surface conditions can largely be explained by the immediate effect of high summer insolation and a reduction of the Northern Hemisphere sea-ice area by about 15-20% (not shown) relative to PI.The large thermal capacity of the ocean explains a larger temperature response over land than over the ocean (Herold et al., 2012;Nikolova et al., 2013).Simulated cooling over North Africa (10-25 • N) and India in the Group I experiments is caused by enhanced monsoonal rainfall in these regions, which is associated with increased cloud cover, i.e. reduced shortwave fluxes, and enhanced land surface evapotranspiration, i.e. greater latent cooling (e.g., Braconnot et al., 2002Braconnot et al., , 2004;;Zheng and Braconnot, 2013).The 416 kyr BP time slice, however, differs from the other Group I members by anomalously cold conditions over the Southern Hemisphere continents.Again, this behaviour can be explained by the immediate effect of the insolation, which shows negative anomalies in the Southern Hemisphere during the JJAS season (Fig. 2).As such, the 416 kyr BP time slice must be considered a special case in Group I.While high Northern Hemisphere summer insolation is related to low precession in most Group I members, positive anomalies of Northern Hemisphere summer insolation at 416 kyr BP are attributable to a maximum in obliquity (Fig. 1), yielding the Northern-versus-Southern Hemisphere insolation contrast.
In contrast to Group I, Group II climates exhibit anomalously cold JJAS surface temperatures globally with the three coldest anomalies at 115, 516, and 609 kyr BP.Again, the temperature response can largely be explained by the direct response to insolation forcing, amplified in high latitudes by an increase of the sea-ice cover (about 5% in the Introduction

Conclusions References
Tables Figures

Back Close
Full Arctic compared to PI).Due to a particular combination of high precession and orbital eccentricity with low obliquity the insolation forcing and surface temperature response is strongest for the 115 kyr BP time slice.Group II warming in the North African and Indian monsoon regions is associated with increased aridity and reduced cloudiness.Group III climates (394 and 615 kyr BP) show rather complex temperature anomaly patterns, especially in the tropics.In the 394 kyr BP time slice, however, northern continental regions show a distinct cooling, whereas continental regions exhibit an overall warming in the Southern Hemisphere.To a large extent, the 394 kyr BP time slice shows a reversed JJAS temperature anomaly pattern compared to the 416 kyr BP simulation over the continental regions, except for Antarctica.

DJF surface temperature anomalies
Boreal winter (December-January-February, DJF) surface temperature anomalies are presented in Fig. 4. Generally low DJF insolation in Group I time slices (Fig. 2) results in anomalously cold surface conditions over most of the globe, particularly strong in the 579 kyr BP (MIS 15) time slice.However, anomalously warm conditions in the Arctic stand in contrast to the global DJF cooling at 6, 9, 125, 405, and 416 kyr BP.The Arctic warming is due to the remnant effect of the polar summer insolation through ocean-sea ice feedbacks (Fischer and Jungclaus, 2010;Herold et al., 2012;Yin and Berger, 2012;Kleinen et al., 2014).Anomalous shortwave radiation during the Arctic summer leads to enhanced melting of sea ice and warming of the upper polar ocean.The additional heat received by the upper ocean delays the formation of winter sea ice, reduces its thickness and finally leads to a warming of the winter surface atmospheric layer by enhanced ocean heat release.For the 504 kyr BP (MIS 13) and 579 kyr BP (MIS 15) time slices, however, the summer remnant effect is masked by a global cooling that is induced by low GHG concentrations typical for early Brunhes (MIS 13 and before) Introduction

Conclusions References
Tables Figures

Back Close
Full reverses to a substantial cooling in the Arctic region.Temperature anomaly patterns in Group III are, again, rather complex.Interestingly, most Northern Hemisphere continental regions remain relatively cold during boreal winter (as in summer) in the 394 kyr BP simulation.

JJAS precipitation anomalies
Boreal summer precipitation shown in Fig. 5 exhibits intensified precipitation in the monsoon belt from North Africa to India, via the Arabian Peninsula, in all Group I simulations in response to high summer insolation (Prell and Kutzbach, 1987;de Noblet et al., 1996;Tuenter et al., 2003;Braconnot et al., 2007).By contrast, the same monsoon regions experience anomalously dry conditions in the Group II (low boreal summer insolation) experiments.The most interesting results regarding the tropical rainfall response to astronomical forcing appear in Group III, where the monsoonal precipitation anomalies show opposite signs in North Africa and India.Table 2 summarizes the summer monsoonal rainfall amounts for the North African (20 • W-30 • E; 10-25 • N) and Indian (70-100 • E; 10-25 • N) regions.Highest rainfall in the North African monsoon region occurs in the 9, 125, 504, and 579 kyr BP time slice runs (all Group I) associated with low precession values (Fig. 1).Driest conditions occur at 115, 495, 516, and 609 kyr BP (all Group II) associated with precession maxima (Fig. 1).As in North Africa, Group I (Group II) experiments exhibit anomalously wet (dry) monsoon conditions in India.

Net Primary Production (NPP) anomalies
Vegetation responds to changes in surface temperature and precipitation and, in certain regions, may feedback to the climate (cf.Rachmayani et al., 2015).Figure 6

Climatic effects of obliquity variations during MIS 11 and MIS 13
The MIS 11 time slices 394 and 416 kyr BP show opposite obliquity extremes (at similar precession), as do the MIS 13 time slices 495 and 516 kyr BP (Fig. 1).Insolation differences between the high obliquity (416, 495 kyr BP) and low obliquity (394, 516 kyr BP) cases (i.e.416 minus 394 and 495 minus 516 kyr BP) are displayed in Fig. 7.The effect of high obliquity is to strengthen the seasonal insolation cycle.At low latitudes, the effect of obliquity on insolation is small.For the maximum obliquity time slices (416 and 495 kyr BP) relatively high boreal summer insolation directly translates into positive surface temperature anomalies over Northern Hemisphere continents, except for the monsoon regions where higher rainfall (see below) leads to substantial surface cooling (Fig. 8a  anomaly is extremely high in this experiment (Fig. 7a).During boreal winter, Northern Hemisphere continents show large-scale cooling in response to high obliquity (and hence relatively low insolation), except for the Arctic realm where the summer remnant effect results in substantial positive surface temperature anomalies (Fig. 8c and d).
During the same season (DJF) anomalously high insolation causes surface warming in the Southern Hemisphere in response to high obliquity.
As a general pattern, especially in the annual mean, maximum-minus-minimum obliquity insolation anomalies cause anomalous surface warming at high latitudes and surface cooling at low latitudes.Aside from seasonal insolation anomalies and climate feedbacks (polar summer remnant effect, monsoons) that create this general pattern, it is also consistent with the obliquity effect on annual insolation, which implies higher annual insolation at the poles and lower annual insolation (but to a lesser extent) at the equator.
Despite the weak insolation signal at low latitudes, substantial obliquity-induced changes in tropical precipitation are simulated (Fig. 8e and f).The strongest signal is found in the North African monsoon region in the MIS 11 experiments, where greater JJAS precipitation occurs during maximum obliquity at 416 kyr BP than during the obliquity minimum at 394 kyr BP.A positive Sahel rainfall anomaly is also found in the MIS 13 experiments (495-516 kyr BP), but much weaker than in the MIS 11 case (416-394 kyr BP).We suppose that the obliquity-induced increase in North African monsoonal rainfall is counteracted by the high precession at 495 kyr BP that tends to weaken the monsoon.

Evaluating the climatic effects of astronomical and GHG forcings through correlation maps
In order to evaluate the climatic effects of obliquity, precession and GHG concentrations, linear correlations between the individual forcing parameters and climatic fields (surface temperature, precipitation) were calculated from the 14 time slice experiments (13 interglacial time slices plus PI).To this end, the total radiative forcing from CO 2 , 3082 Introduction

Conclusions References
Tables Figures

Back Close
Full Figure 9 shows the corresponding correlation maps for annual mean, boreal summer, and boreal winter surface temperature.As expected, GHG forcing is positively correlated with surface temperature over most regions of the globe (Fig. 9a), which is particularly pronounced in the annual mean.For the seasonal correlation maps (boreal summer and winter) the correlation coefficients are smaller because of the dominant impact of obliquity and precession forcing.As already described in the previous subsection, the general surface temperature pattern of high obliquity forcing is warming at high latitudes and cooling at low latitudes (Fig. 9b).High precession (northern solstice near aphelion) leads to boreal summer surface cooling over most extratropical regions (Fig. 9c).However, surface warming occurs in tropical regions as a response to weaker monsoons.During boreal winter, anomalously high insolation causes anomalous surface warming except in the Arctic (due to the summer remnant effect) and northern Australia (due to a stronger regional monsoon).
Correlation maps for annual mean, boreal summer, and boreal winter precipitation are shown in Fig. 10.GHG radiative forcing exhibits no clear response in precipitation except for the high latitudes where the hydrologic cycle accelerates with higher GHG concentrations (Fig. 10a).Arctic precipitation is also amplified by high obliquity during summer (Fig. 10b).Obliquity also strengthens the monsoonal rainfall in North Africa (Sahel region), whereas no effect of obliquity can be detected for the Australian monsoon.The most robust response of the hydrologic cycle is found for precession (Fig. 10c).In particular, high precession reduces summer rainfall in the monsoon belt from North Africa to India as well as in the Arctic realm.During boreal winter, the hydrologic cycle strengthens in the Arctic and Antarctic regions, while Southern Hemisphere monsoon systems amplify resulting in enhanced rainfall over South America, southern Africa, and northern Australia in response to high precession.Introduction

Conclusions References
Tables Figures

Back Close
Full

Discussion
While most time slices presented in this study were simulated for the first time using a comprehensive CGCM, the 6, 115 and 125 kyr BP time slices have been studied extensively in previous model studies.In general, the CCSM3 results are in line with these previous studies in terms of large-scale temperature and precipitation patterns.Warm boreal summer conditions (relative to PI) over most parts of the continents and the Arctic are a general feature in paleoclimatic simulations of the mid-Holocene (6 kyr BP), while the North African and South Asian monsoon regions are anomalously cold due to enhanced rainfall (Braconnot et al., 2007).Though evidenced by proxy records (e.g., McClure, 1976;Hoelzmann et al., 1998;Fleitmann et al., 2003), several models fail to simulate wetter mid-Holocene conditions over the Arabian Peninsula (cf.https://pmip3.lsce.ipsl.fr/database/maps/),while CCSM3 simulates not only enhanced rainfall but also greening of the Arabian Desert.The 125 kyr BP surface temperature pattern shows similar features than the 6 kyr BP pattern, but much more pronounced due to the larger orbital eccentricity and hence stronger precessional forcing.However, compared to other simulations of the last interglaciation, our CCSM3 simulation produces a relatively cold MIS 5e surface climate as shown by Lunt et al. (2013).At 115 kyr BP, surface temperature anomalies show the opposite sign with dramatic cooling over the Arctic and the northern continental regions providing ideal conditions for glacial inception (e.g., Khodri et al., 2005;Kaspar and Cubasch , 2007;Jochum et al., 2012).A retreat of the vegetation at high northern latitudes tends to amplify the insolation-induced cooling (cf.Gallimore and Kutzbach, 1996;Meissner et al., 2003).
A recent simulation of the MIS 13 time slice at 506 kyr BP using the CGCM HadCM3 (Muri et al., 2013) can be compared to our 504 kyr BP time slice using CCSM3.Global patterns of surface temperature anomalies (relative to PI) are remarkably similar in the two different simulations with warm anomalies over all continents (except for the North African and South Asian monsoon regions) in boreal summer and worldwide cold anomalies during boreal winter.Moreover, both simulations show anomalously high Introduction

Conclusions References
Tables Figures

Back Close
Full boreal summer precipitation over northern South America, North and central Africa as well as the South Asian monsoon region.
Although our CCSM3 results show general agreement with other model studies, the validation of model results with data is usually not straightforward.The reader is referred to previous work where our CCSM3 simulation of 125 kyr BP (Lunt et al., 2013) as well as the MIS 11 simulations have been extensively compared to proxy data (Milker et al., 2013;Kleinen et al., 2014).Taken together, these and other studies (e.g., Lohmann et al., 2013) indicate that CGCMs tend to produce generally smaller interglacial temperature anomalies than suggested by the proxy records.So far, the reason for these discrepancies is unsolved (cf.Liu et al., 2014), but Hessler et al. (2014) pointed out that uncertainties associated with sea surface temperature reconstructions are generally larger than interglacial temperature anomalies.Thus, currently available surface temperature proxy data cannot serve as a target for benchmarking interglacial model simulations.
Two time slices of MIS 11 (394 vs. 416 kyr BP) and two time slices of MIS 13 (495 vs. 516 kyr BP) allow the investigation of (almost pure) obliquity effects on global climate.As such, the results from these simulations can be compared to previously performed idealized model experiments in which obliquity has been changed from maximum to minimum values (Tuenter et al., 2003;Mantsis et al., 2011;Erb et al., 2013;Bosmans et al., 2015).The common results of those idealized and our experiments can be summarized as follows.High-versus-low obliquity climates are characterized by strong warming over the Northern Hemisphere extratropics and slight cooling in the tropics during boreal summer.During boreal winter, a moderate cooling over large portions of the Northern Hemisphere continents and a strong warming at high southern latitudes is found.The obliquity-induced Northern Hemisphere summer warming appears to be of particular interest for the MIS 13 climate evolution.At 495 kyr BP, precession is at maximum, but the global benthic δ 18 O stack by Lisiecki and Raymo (2005) does not show the expected increase towards heavier values which would indicate colder conditions and Northern Hemisphere cryosphere expansion (Fig. 1).In fact, despite high Introduction

Conclusions References
Tables Figures

Back Close
Full precession, the 495 kyr BP simulation exhibits the warmest Northern Hemisphere summer temperatures from all Group II experiments (Fig. 3), which can be attributed to concomitant high obliquity.We therefore suggest that the Northern Hemisphere summer climate at 495 kyr BP was not cold enough for ice sheets to grow and global ocean δ 18 O to increase.
Moreover, our CCSM3 results as well as the studies by Tuenter et al. (2003) and Bosmans et al. (2015) suggest a significant effect of obliquity on the strength of the West African monsoon despite the weak insolation signal at low latitudes.Bosmans et al. (2015) have shown that obliquity-induced changes in moisture transport towards North Africa result from changes in the meridional insolation gradient (Davis and Brewer, 2009).However, the impact of obliquity on the monsoon also depends on precession.In the 495-516 kyr BP experiment the obliquity-effect on the West African monsoon is minor, as both time slices (495 and 516 kyr BP) are characterized by precession maxima leading to extremely weak monsoonal circulation and rainfall in both cases.The existence of a ∼ 41 kyr cyclicity (in addition to orbital-related ∼ 100 and 19-23 kyr cycles) in reconstructions of North African aridity during the Quaternary has usually been attributed to obliquity-forced Northern Hemisphere cryosphere effects on the monsoon climate (e.g., Bloemendal and deMenocal, 1989;deMenocal et al., 1993;Tiedemann et al., 1994;deMenocal, 1995;Kroon et al., 1998).Our model results along with the studies by Tuenter et al. (2003) and Bosmans et al. (2015) complement this picture, showing that the direct insolation-gradient forcing associated with obliquity can contribute to West African monsoon changes without involving high-latitude remote climate forcing associated with Northern Hemisphere ice sheets.
According to the CCSM3 results, the Indian monsoon is less sensitive to direct obliquity (insolation gradient) forcing than the West African monsoon.This finding is consistent with proxy records from the Arabian Sea that show substantial 41 kyr (obliquity) periodicity only after the onset of Quaternary glacial cycles when waxing and waning of northern ice sheets could have worked as an agent for the transfer of obliquity forcing to the Indian monsoon region (Bloemendal and deMenocal, 1989).In general, it Introduction

Conclusions References
Tables Figures

Back Close
Full is found that the two monsoon systems do not always vary in concert.This is particularly evident in the Group III experiments (394 and 615 kyr BP) where the precipitation anomalies over North Africa and India have opposite signs (Table 2).Considering the annual insolation cycle in these experiments (Fig. 2), the West African monsoon turns out to be forced by summer insolation, whereas spring/early summer insolation is more important for the Indian monsoon.The different responses to specific forcings and the sometimes out-of-phase behaviour of the African and Indian monsoon systems challenge the global monsoon concept -according to which all regional monsoon systems are part of one seasonally varying global-scale atmospheric overturning circulation in the tropics (Trenberth et al., 2000;Wang et al., 2014) -at orbital timescales.

Conclusions
Using  2008) and Muri et al. (2013).But also during late Brunhes interglacial stages, like the Holocene, model studies suggest an influence of changing land ice on the interglacial climate evolution (Renssen et al., 2009;Marzin et al., 2013).
With increasing computer power long-term transient simulations of interglacial climates will become more common.So far, transient CGCM simulations have been performed for the present (e.g., Lorenz and Lohmann , 2004;Varma et al., 2012;Liu et al., 2014) and the last interglacial (e.g., Bakker et al., 2013;Govin et al., 2014).Transient simulations of earlier interglacials will help to develop a significantly deeper understanding of interglacial climate dynamics in future studies.Introduction

Conclusions References
Tables Figures

Back Close
Full   20• W-30 • E and 10-25 • N) and over India (70-100 • E and 10-25 • N) along with anomalies relative to PI. Absolute precipitation values are given with standard error (2σ) based on 100 simulation years of each experiment.

Stage
Time slice North Africa North Africa Anomaly India India Anomaly (kyr BP) (mm day −1 ) (mm day −1 ) (mm day −1 ) (mm day Full  Berger, 1978) for the different interglacials.The points mark the time slices simulated in this study.
400 and 100 kyr (eccentricity), 41 kyr Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | the GHG forcing of all experiments with values based on Lüthi et al. (2008), Loulergue et al. (2008), and Schilt et al. (2010) using the EPICA Dome C timescale EDC3, except for the MIS 1 and MIS 5 experiments, where GHG values were chosen following the PMIP guidelines (see above).We note that due to the uneven Discussion Paper | Discussion Paper | Discussion Paper | Screen / Esc Printer-friendly Version Interactive Discussion Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | shows the simulated changes in NPP, reflecting increase/decrease and expansion/retreat of vegetation covers, relative to PI.In high Arctic latitudes, vegetation advances (NPP increases) in the Group I simulations, except for 405 kyr BP where temperature changes Discussion Paper | Discussion Paper | Discussion Paper | and b) despite minor obliquityinduced local insolation anomalies.By contrast, receiving anomalously low insolation during austral winter, Southern Hemisphere continents exhibit anomalously cold surface temperatures.For the 416-394 kyr BP case, however, the Antarctic continent and the Southern Ocean show large-scale warming during the JJAS season, which can be attributed to a south polar summer remnant effect as the austral summer insolation 3081 Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | CH 4 , and N 2 O in each experiment was calculated based on a simplified expression (IPCC, 2001).
Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | a state-of-the-art CGCM, 13 interglacial time slice experiments were carried out to study global climate variability between and within Quaternary interglacials.The different roles of obliquity, precession and GHG forcing on surface temperature and precipitation patterns have been disentangled.Several similarities with previous idealized orbital-forcing experiments could be identified.In particular, a significant role of obliquity in forcing the West African monsoon was found, whereas the Indian monsoon appears to be less sensitive to obliquity changes.Different responses to specific forcings and the obvious anti-phase behaviour of the African and Indian monsoon systems in the 394 and 615 kyr BP experiments challenge the global monsoon concept.High obliquity can also explain relatively warm Northern Hemisphere high-latitude summer temperatures despite maximum precession at 495 kyr BP (MIS 13) probably preventing a glacial inception at that time.Future studies should include the effects of changing ice sheets and associated meltwater fluxes in shaping interglacial climates.Large Northern Hemisphere ice sheets might have played an important role for regional and global climates especially during early Brunhes interglacials (MIS 13 and before) as suggested by, e.g., Yin et al. Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Tiedemann, R., Sarnthein, M., and Shackleton, N. J.: Astronomic timescale for the Pliocene Atlantic δ 18 O and dust flux records of Ocean Drilling Program Site 659, Paleoceanography, 9, 619-638, 1994.3086 Trenberth, K. E., Stepaniak, D. P., and Caron, J. M.: The global monsoon as seen through the divergent atmospheric circulation, J. Climate, 13, 3969-3993, 2000.3073Discussion Paper | Discussion Paper | Discussion Paper |
are probably too small.By contrast, Arctic NPP declines in the Group II experiments, albeit only in the easternmost part of Siberia in the 495 kyr BP experiment.A substantial decline of Arctic NPP is also simulated for 394 kyr BP (Group III).In the tropical regions, vegetation changes are mostly governed by precipitation.Consequently, enhanced rainfall results in increased NPP over North Africa, the Arabian Peninsula and India in all Group I experiments.In North Africa increased NPP is associated with a northward shift of the Sahel-Sahara boundary.The largest shifts are simulated for 125 and 579 kyr BP in accordance with maximum North African rainfall anomalies.In these experiments, a complete greening of the Arabian Desert is simulated.Opposite NPP anomalies in the tropical monsoon regions are simulated in the Group II experiments.In Group III, NPP increases result from anomalously high rainfall in North Africa (615 kyr BP) or India (394 kyr BP).