Project Topics
Get Complete Project Material File(s) Now! »
Use of paleoclimate records to understand the natural variability
To better understand the long-term climate variability, we rely on the estimation of cli-mate parameters based on the measurement of proxies. Indeed, the physical, chemical com-position or ecological assemblage in an archive – a lake, ice, coral or any natural material that was made at a given time – depends on the conditions during its formation. By analogy to what we observe today, under the uniformitarian principle, the composition of the archive may be used to infer the past conditions during the formation of the archive: its composition acts as a proxy. Archives that contain information about past climate are named paleoclimate archives. The Intergovernmental Panel on Climate Change (IPCC) summarized the knowledge acquired from paleoclimate in the 2013 report (Masson-Delmotte et al., 2013). They exposed the forcings that drive climate changes, such as solar irradiance, volcanic aer-osol, and greenhouse gases. These parameters are important because they determine the total amount of energy the Earth exchanges with the Sun and space, by absorbing or reflecting short-waved electromagnetic radiation, or emitting and reabsorbing long-waved radiation. Masson-Delmotte et al. (2013) also point out the importance of internal variability, as energy redistribution may change due to restructuration of atmospheric and oceanic circulation, such as in El-Niño events.
Figure 1.1 Regional reconstructions since 1500 CE (colored lines) with 15-year (thin black lines) and 50-year (thick black lines) Gaussian smoothing, shown alongside the median time of onset for sustained, significant industrial-era warming assessed across 15–50-year filter widths (vertical black bars). Grey 1°C scale bar denotes the y-axis scale of each regional tem-perature reconstruction. Figure from Abram et al.
The last 2000 years are especially important as it provides a context of climate variability to better understand the range of natural variability and the contribution of anthropogenic forcing to the recent warming. This timescale is relevant for our society, and relatively well preserved in a wide variety of archives around the world. The Past Global Changes 2k Net-work was specifically created to respond to the raising interest around this time-frame of 2000 years. Climate scientists in this network create new records and gather existing ones to build a database of past climate changes. Temperature is one of the most important parameter of the climate system and was reviewed in detail on land (PAGES 2k Consortium, 2013) and oceans (Tierney et al., 2015). Abram et al. (2016) used this database to provide insight on the recent warming in different regions of the globe: they showed that the onset of recent warming was not simultaneous in all regions, with an early response of tropical oceans and Northern Hemisphere continents, but a warming starting later in the Southern Hemisphere (figure 1.1). Of all the regions presented, Antarctica is the only region where a warming trend during the 20th century is not detected at a continental scale.
Regional differences in Antarctic temperature records
Stenni et al. (2017) present in further detail the temperature reconstructions of the past 2000 years in Antarctica, with a division of this continent into seven regions. Stenni et al. (2017) show that the Antarctic continent was generally cooling from 0 to 1900 CE (Common Era), but that no common trend is standing out for the last 100 years. While the Antarctic Penin-sula, West Antarctica and the coastal Dronning Maud Land show a recent warming trend, other regions and among which the large East Antarctic Plateau show no significant trend, resulting in the absence of consistent trend for the most recent 100 years at the continental scale (figure 1.2). The database used by Stenni et al. (2017) relies largely on the measurement of water stable isotopes in ice cores, as a proxy for temperature (the measurement and inter-pretation of water stable isotopes will be detailed in Chapter 2 of this work). There are two limitations to this: (1) ice cores included in this database are unevenly spatially distributed, especially the records covering the full 2000-year window, as ice cores are often drilled on domes or divides to avoid the glaciological advection of ice, leaving the vast areas between coastal domes and the plateau summits untouched and (2) the temperature reconstruction relies almost only on water-stable isotopes, which also depend on other parameters, particu-larly the conditions at the moisture source during evaporation (Jouzel et al., 2003; further details in Chapter 2). The existing isotope data would benefit from a better understanding of processes influencing the recorded signal. The regional climate would be better understood with temperature records from new locations, to increase the spatial coverage (Christiansen & Ljungqvist, 2017). Finally, the temperature reconstructions in the Antarctic continent could include more diverse temperature proxies, here borehole temperature inversion was only used to calibrate the water isotope-temperature reconstruction in West Antarctica.
The work of Stenni et al. (2017) has shown differing temperature trends between the sub-regions, based on water stable isotope analysis. The spatial heterogeneity of water stable iso-topes was detailed by Masson-Delmotte et al. (2008): they used a database of surface snow analyses to discuss the changes of isotopic composition with elevation, and local average temperature. Masson-Delmotte et al. (2008) highlight the difference between coastal and in-land sites, for which the moisture paths are very different, with a more local source supplying water for the coastal precipitations compared to plateau sites for which the precipitation comes from the free-tropospheric flow. This work shows the importance of understanding the atmospheric circulation which may explain some spatial discrepancies, because the spe-cific atmospheric processes that can influence a given site differ between regions. To fully capture the atmospheric variability and its influence on local climate, we should study a va-riety of locations representative of their regions. The database used by Masson-Delmotte et al. (2008) covers many areas in the Antarctic continent, but there is no equivalent coverage for paleoclimate records, because the logistics are much heavier to retrieve ice cores. Increas-ing the spatial coverage of paleoclimate records seems essential to capture the past variability of atmospheric processes across Antarctica.
Sensitivity of Antarctic climate to atmospheric modes of variability
Ice core water stable isotopes are proxies of temperature initially formed by snow pre-cipitations. For Antarctica, the sensitivity of temperature and precipitation to atmospheric variability has been described by Marshall & Thompson (2016), and Marshall et al. (2017). They define the main modes of variability in the southern hemisphere south of 30°S, as the empirical orthogonal functions (EOF) of atmospheric parameters (eddy kinetic energy and geopotential height). EOF are principal component analysis technique applied to geograph-ically weighted mapped variables, and help understand the most influent sources of variance. The most described mode in the literature on the Southern Hemisphere is the Southern An-nular Mode (SAM), sometimes named as Antarctic Oscillation. Marshall & Thompson (2016) define the SAM as the first EOF of geopotential height, an analog of pressure varia-bility. In a positive SAM phase, the latitudinal pressure gradient is strengthened, pushing the westerly jet poleward, with enhanced zonal circulation and reduced meridional flow. Conse-quently, the Antarctic continent is colder and more isolated as the advection of warm air from mid latitude ocean is reduced (Van Den Broeke & Van Lipzig, 2003). Because of these two effects, a SAM positive phase has a cold signature in most of Antarctic surface temperature (figure 1.3, Marshall & Thompson, 2016), and is associated with reduced precipitation espe-cially on the East Antarctic Plateau (Marshall et al., 2017). As the SAM influences both tem-perature and precipitations simultaneously, a detailed impact assessment is necessary for a better understanding of how water isotopes in Antarctica respond to the main variability mode in the Southern Hemisphere.
Turner et al. (2019) described the precipitation variability in Antarctica. They evidenced that just a few extreme precipitation events dominate the annual precipitation budget, and they also contribute to most of the interannual variability. There is a spatial diversity in the relative contribution of extreme precipitation events. The highest parts of the plateau are less affected by precipitation caused by marine air intrusions, which are responsible for the ex-treme precipitations, and receive a substantial contribution from clear-sky precipitation (Stenni et al., 2016). Conversely, coastal areas are much more sensitive to precipitation caused by synoptic events, and extreme precipitation events contribute largely to the total. Therefore, Turner et al. (2019) show that locations in between the coast and the summit dis-play a range of sensitivity to extreme precipitations. This study further demonstrates that precipitations, in which the initial isotopic composition of snow is acquired, are not neces-sarily representative of the average conditions, and precipitation events may bias temperature reconstructions based on water isotope proxies. The direct impact of precipitation intermit-tency on water isotopes was presented by Persson et al. (2011) for Greenland, but the diver-sity of precipitation regimes and contribution of extreme events at each site calls for careful evaluation of the impacts at each ice coring location, in order to estimate the potential biases in the water stable isotope record.
Diversity of temperature proxies
We discussed the limits of the precipitation-borne proxy that is the isotopic composition of water, related to the variability of precipitation. In addition, the snow deposition is associ-ated with processes of wind mixing, redeposition, patchy accumulation, and vapor-ice ex-changes, resulting in further changes in the measured isotopes (Casado et al., 2018; Jones et al., 2017; Picard et al., 2019). The added variability is superimposed on the climate signal in the isotopes, and can be considered as noise. The deposition noise modifies the isotopic signal at a short spatial scale of a few meters, even though the climatic influence is the same (Münch et al., 2016), and is particularly important in low accumulation areas (Casado et al., 2020). Consequently, it is difficult to evaluate the climate variability from the water isotopic signal of a single core, and averaging multiple cores helps to smooth out the deposition noise (Münch & Laepple, 2018). However, drilling the long ice-cores needed to assess past climate variability is technically costly and difficult: even in the areas with low resolution and high time span, the 2000-year-old ice is buried at a depth of about 100 m, and coastal areas with high accumulation would require cores of several hundred meters. As a result, drilling mul-tiple cores at a single location is not always feasible. For instance, the 3,400 m core retrieved from the West Antarctic ice sheet divide took five seasons to complete drilling (WAIS Divide Ice Core, 2011), as most of the field work is restricted to summer months. In the absence of replicated time series, the limitations on water stable isotopes can be mitigated by comple-menting water isotope data with other proxies for temperature reconstruction in Antarctica.
The temperature of the ice sheet itself is controlled by the diffusion and advection of heat in the polar ice sheet, with the ground temperature and the atmospheric surface temperature as boundary conditions (Ritz, 1987). Changes in annual average surface temperature, and advection of ice by accumulation of new snow are the main sources of variability as the ground temperature is controlled primarily by the geothermal heat flux, which is stable under ice sheets. Using a model of ice advection and heat diffusion, Dahl-Jensen et al. (1998) pro-duced temperature profiles in the ice sheet, from different history scenarios for the tempera-ture. They then used a Monte-Carlo method to match the modelled temperature profiles with measurements of temperature in the borehole left after drilling the GRIP ice core in Green-land, and have a range of compatible temperature histories. Dahl-Jensen et al. (1998) used the temperature in ice to quantify past changes in annual surface temperature in the snow over the last 50,000 years, with a low resolution. This method uses the diffusion properties of temperature in the ice, which can provide a direct quantification of temperature. However, the estimate quickly loses in temporal resolution as the diffusion smears the variations over time. In general, the glacial-interglacial temperature change can be accurately retrieved by this method, at sites with high enough accumulation (Cuffey et al., 1995, 2016; D. Dahl Jensen et al., 1998). Large features of the late Holocene, like the “little ice age” temperature minima can also be reconstructed (Dahl-Jensen et al., 1998; Orsi et al., 2012), and the most recent 50 year changes in temperature trends can be retrieved at more locations, because it is less diffused. For instance, Muto et al. (2011) reconstructed the recent trends at four sites in Dronning Maud Land, East Antarctica. They found both warming and cooling trends, and attributed the divergence of the results to the different altitude and surface slope of their sites, with warming temperature trends on the crest and cooling or no trend in the slope. This method can be very useful to estimate the stability of the local temperature in recent decades, and provide an absolute reference for the temperature calibration of other proxies. It is also precious to establish longer term trends that may be too weak to be captured by other meth-ods.
In addition to the borehole temperature proxy, another temperature estimate can be de-rived from the gases that diffuse in the porous snow at the top of an ice cap (Severinghaus et al., 1998). The porous snow layer is named the firn and is usually 60 to 100 m deep, at which depth the pressure of the snow loaded above forces the closure of porosity and traps gases in bubbles within the ice matrix. Severinghaus et al. (1998) have shown that the isotopic com-position of gases in the firn not only depends on the gravitational settling of heavy isotopes, but is also sensitive to the temperature gradient in the firn. Severinghaus et al. (2001) pre-cisely quantified the thermal fractionation of isotopes for different gases in the firn, allowing Kobashi et al. (2008) to reconstruct past temperature changes in Greenland. Kobashi et al. (2008) used series of nitrogen and argon isotopes to produce a temperature signal independ-ent of the gravitational fractionation. Improving from this study, Orsi et al. (2014) used the same isotope pairs and a temperature diffusion model in the ice, to reconstruct a temperature history, similarly to what has been done from borehole temperature (figure 1.4). Contrary to borehole temperature inversion, the gases isotope inversion allows for a much higher tem-poral resolution, and is useful to quantify changes of temperature at a decadal to centennial scale. Orsi (2013) combined the information from gases and from borehole temperature, and used an ice-advection temperature diffusion model to fit both datasets to provide information on the past temperature in West Antarctica. This method has not yet been used to reconstruct millennial-length temperature records in East Antarctica.
Previous works have highlighted the uniqueness of Antarctic climate. While the climate in most regions has warmed as a response to anthropogenic forcing, the high variability of Antarctic surface temperature does not allow us to distinguish a general trend for Antarctica (Abram et al., 2016). The different regions of Antarctica respond differently to changes in Southern Hemisphere climate, owing to atmospheric pathways that are specific to each re-gion. Paleoclimate reconstructions, and especially high resolution temperature records cov-ering the last 2000 years, are needed to complement the sparse records available in Antarc-tica. So far, ice cores have been preferentially drilled near the coast or at high elevation, and the few ice cores in between usually cover a much shorter time span.
When drilling ice cores with the aim to reconstruct temperature, the main proxy used is traditionally the water isotopic composition of the ice. However, it is highly sensitive to pre-cipitation regimes. It is thus necessary to evaluate the sensitivity of the water isotopes to the climate where the core is drilled.
Finally, temperature reconstructions are mostly reliant on the water stable isotopes. Other existing proxies such as borehole temperature and gases isotopes should be further applied, as they could complement the water isotopes and provide a quantitative estimate of temper-ature changes.
Here we propose to study the last 2000 years of temperature variability on the lower East Antarctic Plateau, using both water stable isotopes and inert gases stable isotopes in the Au-rora Basin North ice core. This core was recently drilled in East Antarctica to increase the spatial coverage of paleo-temperature records, at a site midway between the coast and the summits.
Organization of the manuscript
After detailing the commonly used methods (Chapter 2), we will determine the local climate conditions at Aurora Basin North, with a focus on the temperature and precipitation (Chapter 3). We will evaluate the impacts of the precipitation regimes on the water stable isotopes in recent snow (Chapter 4). Then we will describe the main core that was drilled, and how we used the measurements made by our collaborators to reconstruct the temperature from water stable isotopes (Chapter 5). We will detail the measurements of stable isotopes in gases and how they were used to produce a second temperature record (Chapter 6). Finally, we discuss the temperature records and how they inform us on the climate evolution of Ant-arctica in the last 2000 years (Chapter 7), and give our conclusions (Chapter 8).
Table of contents :
1 Past and present Antarctic Climate, a search for temperature changes
1.1 Importance of Antarctic temperature in the global climate
1.2 Use of paleoclimate records to understand the natural variability
1.3 Regional differences in Antarctic temperature records
1.4 Sensitivity of Antarctic climate to atmospheric modes of variability
1.5 Diversity of temperature proxies
1.6 Organization of the manuscript
2 Tools for understanding temperature and climate from ice core records
2.1 Characteristics of ice sheets allowing for ice coring
2.2 General information on stable isotopes
2.2.1 Definition of stable isotopes
2.2.2 Partitioning of stable isotopes during fractionating processes
2.2.2.1 Equilibrium fractionation
2.2.2.2 Kinetic fractionation
2.3 Water stable isotopes
2.3.1 Isotopes in polar precipitation
2.3.1.1 Evaporation at moisture source
2.3.1.2 Transport and condensation of precipitation
2.3.2 Mitigation of the climate signal related to snow deposition
2.3.2.1 Deposition
2.3.2.2 Snow-vapor exchanges
2.3.2.3 Diffusion
2.3.3 Calibration of isotope – temperature slope
2.3.4 Analytical methods
2.3.4.1 Cavity Ring-Down Spectroscopy
2.3.4.2 Fluorination and mass spectrometry
2.3.4.3 Working standards
2.3.5 Summary of water isotope signal in ice cores
2.4 Gases Stable isotopes
2.4.1 Structure of the firn
2.4.2 Firn fractionation processes
2.4.2.1 Gravitational fractionation
2.4.2.2 Thermal Fractionation
2.4.2.3 Convective disequilibrium
2.4.3 Fractionation during close-off
2.4.4 Analytical methods
2.5 Firn model and temperature reconstructions
2.5.1 Forward diffusion-advection model
2.5.2 Inversion of the model for temperature reconstruction
2.6 Borehole temperature
2.7 Atmospheric models and reanalysis
2.8 Conclusion
3 Evaluation of the climate variability at Aurora Basin North, using atmospheric climate models
3.1 Introduction
3.2 Methods
3.2.1 Site description
3.2.2 The regional climate model MAR
3.3 ABN as a tracer of East Antarctic Climate
3.4 Climatology of the precipitation in the MAR model
3.4.1 Distribution of precipitation events
3.4.2 Seasonality and variability of precipitation
3.4.3 Synoptic conditions driving snowfall
3.4.4 Temperature anomaly associated with precipitation events.
3.5 Conclusions
4 Warm synoptic event control of the snow isotope variability in East Antarctic Plateau evidenced from snow record and isotope-enabled atmospheric circulation model
4.1 Introduction
4.2 Material and methods
4.2.1 The isotope-enabled general circulation model ECHAM5-wiso
4.2.2 Snow isotope records
4.3 Annual dating of snow
4.3.1 Snow record age models
4.3.2 Comparison of accumulation records
4.4 Water stable isotopes in ECHAM5-wiso
4.4.1 Sensitivity to temperature and precipitation
4.4.2 Spatial correlation of isotope signal
4.5 Comparison of isotope records and climate from 2005 to 2014
4.5.1 Effect of post-deposition processes
4.5.2 Isotopic signature of the winters with warm events
4.5.3 Influence of the Southern Annular Mode
4.5 Calibration of the isotope – temperature slope
4.6 Conclusions
Appendix 4.A – Computation of weighted statistics
5 The Aurora Basin North main core
5.1 Motivation for ice coring in Aurora Basin North
5.2 Dating the ice
5.3 Accumulation history
5.4 Borehole temperature
5.5 Ice rheology
5.5.1 Ground penetrating radar
5.5.2 Elevation changes
5.5.3 Temperature changes
5.6 Water stable isotopes
5.6.1 Measurements of δ18O and δD
5.6.2 Flow-correction with spatial δ18O calibration
5.6.3 Conversion to temperature
5.6.4 Deuterium excess
5.6.5 17O-excess
5.7 Gas concentrations
5.8 Conclusion
6 Reconstruction of temperature from δ40Ar, δ15N, and borehole temperature
6.1 Measurement of δ15N and δ40Ar from air trapped in ice
6.1.1 Wet extraction of Nitrogen and Argon gases
6.1.1.1 Ice preparation
6.1.1.2 Line preparation
6.1.1.3 Ice melt, oxygen removal, and transfer
6.1.2 Preparation of Air samples from free Atmosphere and firn
6.1.2.1 Air standards
6.1.2.2 Firn air samples
6.1.3 Mass Spectrometry
6.1.3.1 Description of the mass spectrometer
6.1.3.2 Introduction of gases
6.1.3.3 Measurement
6.1.4 Data calibration and correction
6.1.4.1 Pressure imbalance correction
6.1.4.2 Chemical Slope correction
6.1.4.3 Drift correction
6.1.4.4 Normalization to atmosphere
6.1.4.5 Argon gas loss
6.1.4.6 Detection of Outliers
6.1.4.7 Multiple-sample-based smoothing
6.2 Reconstruction of the temperature history
6.2.1 Quantification of gravitational and thermal fractionation of gases
6.2.2 Convection in the shallow firn and evolution of the lock-in depth
6.2.3 Gas Age model
6.2.3.1 Firn compaction
6.2.3.2 Adjustments with methane
6.2.3.3 Implications for the Lock-in Depth
6.2.4 Modeling of temperature diffusion in the firn and ice
6.2.5 Linearization and inversion of temperature history
6.2.5.1 Simulation of multiple scenarios bi of temperature history: forward model
6.2.5.2 Recombination of temperature history scenarios to match the temperature gradients estimated from gases: inversion and determination of xi
6.2.6 Sensitivity of temperature inversion
6.2.6.1 Influence of Borehole Temperature in the temperature inversion
6.2.6.2 Initial temperature hypothesis
6.2.6.3 Constant delta-temperature shift to firn-ice δ40Ar correction
6.2.6.4 Seasonal temperature cycle amplitude
6.2.6.5 Accumulation
6.2.6.6 Lock in depth
6.2.6.7 Summary
6.2.7 Glaciological correction
6.3 Conclusions on the temperature reconstruction
7 Climate interpretations of the Aurora Basin North data
7.1 Climate at ABN
7.1.1 Summary of the temperature records
7.1.2 Possible causes for the divergence of temperature reconstructions
7.1.2.1 Moisture source effect on the δ18O
7.1.2.2 Boundary layer changes
7.1.2.3 Change in seasonality of the precipitation events
7.1.2.4 Post deposition effects on water stable isotopes
7.2 Teleconnections and climate variability in the southern hemisphere during the last 2000 years
7.2.1 Comparison with other ice cores in the region
7.2.2 Influence of the Southern Annular Mode
7.2.3 Relationship with sea ice records
7.2.4 Relationship with Pacific South America patterns
7.2.5. Inter-hemispheric coupling
7.3 Conclusions
8 Conclusion and perspectives
8.1 The Aurora Basin North temperature record
8.2 Perspectives
8.2.1 Explore the influence of other climate modes on ABN temperature
8.2.2 Confirm our interpretations with further analyses
8.2.3 Further explore moisture source variability
8.2.4 Reproduce the 15Nexcess and borehole temperature inversion at other ice core locations
