It has long been recognized that the amplitude of the seasonal cycle can substantially modify climate features in distinct timescales. This study evaluates the impact of the enhanced seasonality characteristic of the Marine Isotope Stage 31 (MIS31) on the El Niño–Southern Oscillation (ENSO). Based upon coupled climate simulations driven by present-day (CTR) and MIS31 boundary conditions, we demonstrate that the CTR simulation shows a significant concentration of power in the 3–7-year band and on the multidecadal timescale between 15 and 30 years. However, the MIS31 simulation shows drastically modified temporal variability of the ENSO, with stronger power spectrum at interannual timescales but the absence of decadal periodicity. Increased meridional gradient of sea surface temperature (SST) and wind stress in the Northern Hemisphere subtropics are revealed to be the primary candidates responsible for changes in the equatorial variability. The oceanic response to the MIS31 ENSO extends to the extratropics, and fits nicely with SST anomalies delivered by paleoreconstructions. The implementation of the MIS31 conditions results in a distinct global monsoon system and its link to the ENSO in respect to current conditions. In particular, the Indian monsoon intensified but no correlation with ENSO is found in the MIS31 climate, diverging from conditions delivered by our current climate in which this monsoon is significantly correlated with the NIÑO34 index. This indicates that monsoonal precipitation for this interglacial is more closely connected to hemispherical features than to the tropical–extratropical climate interaction.
The Marine Isotope Stage 31 (MIS31; early Pleistocene 1085–1055 ka) is a key
paleoclimate period to simulate and analyze the global environmental response to a
significantly modified climate forcing
Indeed, changes in the physical and dynamical characteristics of the ENSO
have been related to seasonal and interannual global-climate disturbances
The effect of ocean dynamics also modifies the tropical–extratropical interaction due to
different ENSO flavors
The far-reaching effect of equatorial dynamics on climate has also been found
by
Based on paleoreconstruction of wind and precipitation on the Chinese Loess
Plateau,
Significantly modified periodicity and amplitude of past ENSO regimes, and their global
influence, shed light on the potential effect of human-induced climate change on the
equatorial Pacific, and consequently on future ENSO-like climate. Furthermore, it should
be argued that disagreement in the magnitude of cooling or warming among coupled climate
models and paleoreconstructions may be related to the local responses of temperature and
precipitation elicited by distinct ENSO in both spatial and temporal variability
Thus, understanding of the air–sea interaction related to the equatorial
Pacific and its climate response at interannual and multidecadal timescales
in distinct epochs, such as those investigated in the present study, is vital
to understand past interglacial intervals that are characterized by depleted
ice sheets. Verifying the potential effect of atmospheric
Climate simulations have been performed with the International Centre for
Theoretical Physics – Coupled Global Climate Model
The atmospheric component runs at T30 horizontal resolution, and there are eight levels
in the vertical. NEMO is a primitive equation
Two simulations are evaluated: a modern climate driven by present-day
boundary conditions (CTR) and a
second experiment for the MIS31 forcing. The CTR simulation was run to
equilibrium for 2000 years, and our modern climate is the time average of the
last 500 years of the CTR simulation. The CTR is run under present-day
orbital forcing and
The implementation of MIS31 Antarctic topography differs from the CTR counterpart
primarily by the absence of the WAIS, which, according to
Table
Averaged surface temperatures (
When viewing the global distribution of surface temperature (Fig. S1 in the Supplement), it is demonstrated that the ICTP-CGCM is able to reproduce the main features of the ERA-I data. The ICTP-CGCM performs fairly in reproducing the monthly variability in temperatures as shown by the standard deviation (SD). Higher values of SD are over Asia and North America related to the high seasonality due to the effect of continentality. Larger values are also observed over oceanic regions along the storm track. However, due to the model resolution, a limitation is noted over steep topographies such as the Tibetan Plateau, Andes, and Rocky Mountains.
Temperature differences between the MIS31 and the CTR show that most warming occurs in
the boreal summer, reaching 1.2
Figure S2 shows the monthly averaged hemispheric pattern for surface solar radiation (SSR) and surface temperatures delivered by MIS31 and CTR simulations. This figure demonstrates an interhemispheric seesaw emphasizing the substantial increase in the boreal SSR during the summer season in the MIS31 experiment, and a similar situation occurs in the Southern Hemisphere during DJF in the extratropics. It has to be argued that the reason for larger seasonality in the SH for the CTR run is related to the excess of SSR in DJF but deficit in JJA as compared to the NH (Fig. S2). Thus, much warmer summer conditions and colder winter/spring conditions in the SH increase the annual amplitude.
Due to astronomically driven reduced sea ice, larger changes are located in
the NH extratropics (see
The inclusion of distinct astronomical forcing leads to NH peak summer
(June/July) insolation, with an opposite effect in the SH due to the
interhemispheric seesaw relationship of the precession cycle
The inclusion of MIS31 boundary conditions also results in changes in sea level pressure
(SLP) and the vertical structure of the atmosphere. Figure
Moreover, this strategy is important because changes in circulation are dictated by
changes to the gradient of geopotential rather than absolute magnitude anomalies
According to
These changes in the stationary wave induce substantial modifications in the wind stress
and SST near-surface air temperature features delivered by the MIS31 climate. Indeed,
Fig.
Modification in the near-surface atmospheric circulation can also modify the oceanic
vertical characteristics affecting the thermocline depth and ENSO
The ICTP-CGCM properly reproduces the equatorial thermocline depth (using the
depth of maximum vertical temperature gradient) compared to the Levitus
dataset
A deeper thermocline, however, is observed in part of the NIÑO3 region
(Fig.
Over the western Pacific, stronger equatorward winds (Fig.
The intensification of the Sverdrup transport by up to 6 Sv between
20–40
The use of harmonic analysis allows for the identification of dominant climate signals in
the space–time domain, separating small- and high-frequency processes (e.g., diurnal
cycle) from large-scale features (e.g., seasonal). Analyses conducted on the frequency
domain can capture and differentiate the contribution of all timescales. Thus, different
climate regimes and transition regions can be characterized. The first harmonic shows the
dominance of the annual cycle when most of the variance is represented by this harmonic.
It has to be stressed that investigations based upon area-averaged time series are
embedded with small- and large-scale processes dictated by distinct periodicity; this in
turn hampers the identification of periodic climatic signals in the space–time domain
Thus, further evaluation on modifications of the annual and semiannual cycle in the MIS31
and CTR simulations are provided below. Changes in the harmonic variance and amplitude
are highly correlated with the amount of incoming shortwave radiation (SSR) in the MIS31
climate, as shown by differences in the first harmonic (Fig.
Differences in the first harmonic variance between the MIS31 and CTR run for
Figure
Figure
This structure is not seen in the equatorial Atlantic where variance differences between
the MIS31 and the CTR are meridional. In fact, under CTR conditions this can be
interpreted as the tropical Atlantic variability (TAV) related to the continental monsoon
forcing, wind stress, and air–sea interaction
The SLP differences are more complex, showing a pattern that differs from zonal or
meridional features (Fig.
It is expected that those changes in the atmospheric zonal and meridional circulations
and the wind-driven oceanic flow can result in modifying ENSO frequency and power.
Moreover, a shallow thermocline as delivered by the MIS31 simulation indicates reduced
upper-ocean heat content that may intensify the high-frequency cycle in the equatorial
region, in particular the interannual ENSO variability
Multi-taper power spectrum of NIÑO3, NIÑO34, and NIÑO4.
Panels
The following explores the influence of the MIS31 forcing on ENSO indices. Among several
mechanisms related to ENSO dynamics, the magnitude of the seasonal cycle in the
equatorial region characterizes its onset, intensity, and frequency
Figure
All periodicities mentioned below are significant at the 95 % confidence level.
Compared with the power spectrum delivered by the HadISST, the ICTP-CGCM shows sharper
peaks in the 3–7-year band for all NIÑO indices (Fig.
The reason for this slight shift to higher frequency in the NIÑO4 region is not clear; however, because the NIÑO4 is located much closer to the Tropical Warm Pool region, which is dominated by a weak seasonal cycle with the first harmonic explaining about 30 % of the total variance, it may indicate that higher-order harmonics play a role to induce some power at higher frequency.
The weakening of decadal variability in the NIÑO4 region may be related to wind
variability in the off-equatorial tropics as proposed by
The incorporation of MIS31 boundary conditions drastically modifies the temporal
variability of the interglacial ENSO (Fig.
The opposite is delivered by the MIS31 simulation, a fact that usefully serves to support
the assumption of weaker decadal air–sea interaction during this interglacial. Indeed,
for the MIS31 simulation, correlation values between the NIÑO34 index and the first
principal component (PC1) of wind stress computed at 0–20
In fact, the decadal variability found in the CTR NIÑO34 power spectrum fits nicely
with the proposed mechanism raised by
Turning to the regression patterns induced by the NIÑO34 indices, Fig.
Differences in reconstructed SST and Lake E temperatures (
Panel
Lag correlations between the Asian and Australia monsoon domains and the
NIÑO34 index (black for CTR and red for MIS31 simulations).
The impact of NIÑO34 on SLP (Fig.
This is achieved by comparing the modeled SST anomalies for JJA to SSTs differences
between the MIS31 and CTR delivered by the regression pattern related to the NIÑO34
index (
As shown in Fig.
To further investigate the MIS31 climate features, the correlation between
precipitation computed over regional Asia and Australia monsoonal domains, as defined by
Estimates of changes in precipitation for past interglacials are still scarce, but our
MIS31 simulation agrees with other studies showing enhanced Asian summer monsoon during
interglacials
Turning to individual monsoonal domains, it is demonstrated that during the MIS31 the
link between the NIÑO34 and the Asian and Australian monsoons is weakened with respect
to the CTR characteristics (Fig.
In order to verify the impact of decadal variability on the link between the NIÑO34 and
the monsoon, Fig.
This investigation centered on a comparison between present-day conditions (CTR) and those characteristics of a super-interglacial epoch, the Marine Isotope Stage 31 (MIS31). Using coupled global climate model simulations (ICTP-CGCM), we have first demonstrated significant changes in the spatial patterns and seasonality of sea-level pressure, sea-surface temperatures, and heat fluxes during the MIS31 climate compared to present-day conditions, and these changes have a significant impact on the main modes of variability. Anomalous equatorial wind stress associated with a modified seasonal cycle in the MIS31 simulation leads to stronger ENSO variability compared to the present-day climate. Moreover, the decadal variability differs dramatically in the MIS31 simulation from that characteristic of present-day conditions. This decadal variability also differs greatly across the ENSO diversity spectrum with off-equatorial atmospheric circulation playing a significant role in inducing decadal variability.
Evaluation between paleoreconstruction and modeling results is a complex task, because reconstructions depict dominant signals in a particular time interval and locale. Thus, they cannot be assumed to geographically represent large-scale domains, and their ability to reproduce long-term environmental conditions should be considered with care.
Discrepancies between modeling results and paleoreconstructions for the MIS31 climate, which occurred under very particular conditions and high seasonality, may unfortunately be expected. The MIS31 may have been dominated for instance by vegetation patterns drastically different than today. This modifies the global evapotranspiration rates and the hydrological cycle, producing precipitation that can differ greatly from model results. This suggests that uncertainties in the model may be reduced when including more realistic boundary conditions that are currently not available.
All data used in the present study are available upon request to Flavio Justino (fjustino@ufv.br).
The supplement related to this article is available online at:
FJ designed the study, wrote large portions of the paper, and performed data processing and plotting. DL and FK performed all model simulations. All authors substantially contributed to the interpretation of the results.
The authors declare that they have no conflict of interest.
This work was supported by the Brazilian National Research Council project 306181/2016-9. The first author also thanks the ICTP for providing the necessary infrastructure.
This paper was edited by Qiuzhen Yin and reviewed by two anonymous referees.