 Okay, I think we are ready to start. Good afternoon everyone. My name is Sandro Scandal. I'm here on behalf of the director of the STP without in town Fernando Quevedo. I'd like to welcome all of you to this colloquium today. Not only those who are present here in the room, but also those who are following through our live stream on Facebook and YouTube. Those of you are there, hello. Anna Pirani will be following the feed and so you can actually ask questions not only from the audience here, but also from your live stream. So we have an outstanding speaker today. Actually this month we really have a lineup of outstanding speakers including a Nobel Prize winner next week. But I will actually ask my colleague Erika to introduce today's speaker. I just want to mention the rules of the game. So at the end of the colloquium there are refreshments outside for everyone. The diploma students as usual they will be asked to come down closer to the speaker and they have a private session with the speaker and we'll make sure that there are some food left for them at the end. So now ask, where is Erika? You go ahead. Erika Coppola. Hi everybody. So we have today here we have Dr. Valérie Mazzona-Dalmond from the University of Saquille, Paris. And why she is visiting us is visiting us because she is the co-chair of working group one of the assessment report six that is just starting from the IPCC. So we are hosting here this week the IPCC expert meeting on assessing climate information on region. And the reason why we are hosting here is because for the first time this new assessment report is focusing on regional climate and the ESP has the same focus. So this was a nice hosting. And for those of you that don't know what IPCC is, I will say just a few words. So the IPCC has been founded in 1988 was established by the UNAP, so the United Nations Environmental Program together with the World Meteorological Organization. It was supposed was established to provide information on climate change to the and on climate change, the effect on climate change, a potential effect on environment and such economic impact. So this is a very peculiar organization because it's made from scientists, scientists work in this in this report but it speaks to the society, so to the government and to the politician. So the the way in which the IPCC structure is it's it produces an assessment report every four or five years and now we are at the beginning of the next one that will be the sixth one. There are several working group in the IPCC and for each working group there are several chapter and each chapter has a lead outer and I think it's worth to mention because student a year that for this assessment report that Valerie is trying we have two students from ESP section that have been nominated as leading outer one in chapter two and one in chapter third in the third chapter. Okay so with this let's see so Valerie is the co-chair as I mentioned already as she was she research was a focus on quantifying understanding past change in climate and atmospheric water cycle using analysis from ice core in Greenland, Antarctica and Tibet analysis from the tree ring and from tree ring and from present day monitoring and for climate modeling for the past and for the future. So she is active also in outreach for children and for genetical public as well and there are several recognition that she got for excellent work. One is the European Union Decratus Prize from EPICA project the women scientist Irena Curie Prize in 2013 the Tinker Muse Prize for science and policies in Antarctica in 2015 and also she's a highly excited scientist since 2014. So welcome and the floor is yours. Good afternoon thank you very much for this opportunity to share some of the research I've been doing with many colleagues around the world and many young scientists and especially one colleague of mine who has worked for a long time in the University of Trieste Barbarasteni she has been the closest person I've been working with and now she's a professor in Venice University. So what I would like to share with you is the interest of using natural archives and insights on past climates before the instrumental records. Share with you examples of the methods we are using with the work I've been doing with water stabilizer tops introducing the interest of understanding better the parameters that we can measure in these natural archives and technological progress opportunities to better understand what we call proxies. So surrogates to climate variables from measurements we have in natural archives and give you examples of the challenges we have with respect to past climates in polar regions so with illustrations from Greenland and from Antarctica. So I will give a brief introduction I will focus on some of the most recent results we have obtained from present-day monitoring it is motivated by the wish to better understand what we measure in ice cores but at the scale of where the process is and it's also bringing new knowledge and opening new questions about how well do we understand today's water cycle. Instead of focusing on the longest ice core records and the largest signals we can find I will show some of the challenges we face when we look at the last centuries to millennia where we try very hard to be able to detect small signals within the noise associated with climate variability but also the way our records are produced in the snow in polar regions and the challenge is associated with the understanding of the drivers of these signals and finally I will focus on some of the largest changes the response of the climate system to changes in the earth's orbit the astronomical forcing it is well known and thus it is an opportunity to try to disentangle how the climate system is responding to this well-known forcing and I will focus on the last interglacial period where we have strong signals so this introduction will focus on briefly why do we study past climates we study past climates because it allows us to make use of what we can see as natural experiments on the earth's climates natural experiments mean that this climate system has been responding to natural perturbations I will show some of the work we've been doing with volcanic forcing and I'm trying to decipher the magnitude of the climate response to volcanic forcing in the last thousand years I will also show as I said before what we are trying to do to understand our ability to correctly model the response of the climate system to the well known changes in the earth's position around the sun so the astronomical forcing during interglacial periods past climates are also important to decipher the full range of natural climate variability in some places of the world especially polar regions our satellite records are about 30 years old our instrumental records may be longer but only in some scarce places in Antarctica mostly back to the international geophysical year 1957 so our direct records of climate variability are very short and we need to make use of whatever exists to better understand climate variability over longer time scales multiple decades centuries or millennia I will not show what we have learned about past abrupt changes but this has been a quite strong focus of my own research trying to characterize better the full magnitude of the most rapid changes of climate during past cold phases ice ages and I will illustrate how we use a diversity of indicators to benchmark climate models so we cannot do it directly comparing temperature to temperature but because some of the parameters we measure in archives can be implemented in models of the atmosphere we can compare directly what we measure in ice cores the ratio of heavy to light water isotopes water isotopes we can compare that directly to climate models if they have been equipped with a multiplicity of water cycles for each of these water molecules so I will briefly illustrate that finally I want to show that ice cores are formidable time machines when working with ice cores and you can see a small picture of a segment of Antarctic ice from a depth of about a thousand meters you can see water and the parameters I've been working on mostly are measurements of the abundance of heavy to light water molecules in that water they are traces of past temperature but also integrated traces of any phase change from evaporation to condensation and storage in the ice matrix they give insights for instance about past changes in moisture sources supplying moisture to polar regions when you have drilled an ice core the hole provides you with a record of past temperature just due to the diffusion of temperature in the ice matrix so basically by inverting the heat diffusion equation you have access to an a smoothed information on past temperature at the surface mostly for the last decades in some cases back to the last ice age but quite smooth you don't see that there are also aerosols deposited in ice cores and these aerosols provide information on regional climate for instance the deposition of dust provides information about dryness over surrounding continents the deposition of sea salts provides information about for instance sea ice extent or turbulence and transport of sea salts towards polar regions deposition of volcanic aerosols in ice cores is extremely precious because these are key to reconstruct the amount of for instance sulfur aerosols in the atmosphere and they afford to understand the climate effects of various explosive volcanic eruptions in the past the radium 10 the cosmogenic isotopes deposited in ice cores is used to reconstruct past activity of the sun through its interplay with the high atmosphere and we also have direct records of pollutants in various places of the world and finally we have bubbles you can see them here at the depth of 100 meters you don't see bubbles these bubbles are trapped and isolated from the ambient air in between the surface and these depths where there is a cloud close of process you can also extract a wealth of information from this fern densification process because it is also linked to local climate changes in the amount of snow precipitation changes in local temperature and when you understand these diffusion processes in the fern you can also make use of the atmospheric composition signal and ice cores are the unique direct way to access to past changes in well-mixed greenhouse gases in the atmosphere and this is a global signal since the 1950s people have developed and improved logistical and technological methods to access to the inland polar regions and drill ice cores of the best possible quality the deepest record is about 4 000 meter long it comes from west Antarctica and I will show in a minute the longest record obtained from ice cores it is from the French Italian station of Concordia also known as Domsi on the map of Antarctica and it covers 800 000 years so a succession of ice ages and brief warm interglacial periods in Greenland on the oldest ice core retrieved so far is located at the Neen site on the northwest of Greenland and it allows us to go back to the last warm period before the last ice age about 125 000 years ago records are shorter in Greenland because there's each year a larger amount of snow precipitation compared to Antarctica which is colder and drier the ability to retrieve very old ice is linked to the glaciological ice flow thinning and deformation the conditions below the ice thickness especially geothermal heat flow or the lack of geothermal heat flow and the snow precipitation amount today and through time there are two main challenges with ice core records trying to get as far back in time as possible but also trying to have the highest possible resolution for the very recent climate at the regional scale so that's the reason why you may also see that we have a matrix of ice cores also in coastal regions where we can reach sometimes a few tens of years in the back in back in time sometimes a few thousand years but in many places we can do that almost with seasonal resolution so we can identify summer and winter layers and thus we can provide climate information with a very precise chronological framework and information from layer counting in ice core records is used as a reference also for calibrating other methods to date natural archives it can be done with a year precision back to a thousand years it can be done with a few tens of years of precision back to about 10 000 years and of course our age uncertainty increases back especially when we go into cold conditions where it is drier and thus annual layers gets thinner and thinner and it's also decreasing when we get further back in time because of the thinning of ice layers due to glaciological processes the best resolution of our records that we can obtain is a few decades for let's say the last interglacial period a thousand twenty five thousand years ago a hundred twenty five thousand years ago and a few hundred years for some of the oldest warm phases so my favorite tool is the ratio between light and heavy isotopes of the water molecule so basically we have a reference standards that's the mean composition of the ocean water and using different techniques we measure the relative abundance of oxygen 18 to oxygen 16 in samples and we compare that and we scale that to the standard mid ocean water using a unit called a delta unit which is very common in geochemistry we do that for oxygen 18 we do that for hydrogen so deuterium isotopes as well this has been made possible since the 1940s and the development of mass spectrometers and now we use also optical methods laser spectrometers that are cheaper easy to deploy in the field that can provide continuous records and are reaching about the same accuracy as the usual laboratory mass spectrometers so some of the key issues linked with water stable isotopes are phase changes when you have evaporation condensation you have fractionation between the solid or the liquid and the vapor phases so if you look at water vapor form that's the ocean surface it contains less heavy isotopes than in seawater and when air masses are transported from warm regions to the poles they undergo a distillation process so when the air masses get colder you have rain or snow and each time you have condensation you remove a little bit of the heavy molecules in the liquid or in the solid phase compared to the vapor phase so gradually the vapor is losing heavy isotopes the oxygen 18 delta unit gets lower and lower and this is today spatially related to temperature this discovery was made in the 1960s both for Greenland Antarctica and precipitation around the world and it's it it led to the use of the measurement of the ratio of heavy to light molecules as an indicator of past temperature and this can be done in ice cores but this can be done when you have indirect records of precipitation for instance in carbonates in lake sediments or in cellulose in truings so this is a widespread tool it is used in paleoclimatology it is also used in hydrology to track the source of groundwater for instance or sometimes for monitoring purposes to check the origin of olive oil or wine or water from multiple sources this is a simple story but of course there are many open questions I spoke about equilibrium fractionation during phase changes but we also face kinetic fractionation especially during evaporation processes we can have due to different diffusivities of molecules kinetic fractionation processes taking place that are basically linked to how fast is evaporation occurring and this is dependent on sea surface temperature and relative humidity above the sea surface we also have kinetic fractionation when you have droplet revaporation or when you have condensation on ice crystals these are processes that happen out of equilibrium at the molecular scale so we also make use of this second order kinetic fractionation effect by combining the different isotopes of water and if you look at global mean a global database of precipitation samples most of them align on a line with a slope of eight which is the mean effect of equilibrium fractionation and thus we are interested by the deviation from this global mean so-called meteoric water line and we use second order parameters like deuterium excess defined as the content the delta deuterium minus the average relationships factor of eight oxygen 18 and by doing that we can get rid of this distillation temperature driven effect and we can identify a signal that is dominated by kinetic fractionation why is that helpful it is helpful because by measuring this type of parameter in polar ice you can have access to an information that is at first order linked to the evaporation conditions at the initial moisture source of the water vapor so it provides a distant information on the water cycle beyond an information on local climate and distillation effects just to give you an idea of progress through time since the 1950s we know the link between isotopes and temperatures spatially these are pioneer studies for Antarctica it was obtained with maybe a hundred points by Claude Lourius a French glaciologist I repeated this analysis in 2008 with more than 2,000 data points and I obtained the same result however thanks to high resolution records we have also used water stabilizer tops to identify annual layers together with aerosols that are deposited at different seasons this is an example from Greenland ice cores at a higher resolution and temporarily by doing that we have identified challenges the relationship between oxygenating and temperature that we see spatially that is easy to explain with relay distillation processes is not stable through time it is not valid at the seasonal scale it is not valid at the inter annual scale and this is because at these various scales you have changes in the origin of moisture and you have changes in the timing of precipitation and the covariance between temperature and precipitation even at the day-to-day scale can distort the value of the relationship between isotopes and temperature so it is a key challenge for quantitative reconstructions and we have worked a lot on that for instance including atmospheric general circulation models equipped with water stable isotopes the first implementation was done by SidVision Som in the early 1980s in the French atmospheric general circulation model we have about now 10 atmospheric general circulation models that have been equipped with water stable isotopes oxygenating deuterium and allowing us also to look at second order parameters this is an example of a comparison between the information from precipitation data globally and the outputs of one of these atmospheric general circulation models but in a way there is a sort of circular reasoning because there are still a couple of open questions with respect to the quantitative understanding some of the fractionation coefficients are not fully known in the full range of temperature variations there are open questions about parameterizations associated with these isotopic processes and thus the these models are usually tuned to what is available and what is available is this data set of today's precipitation samples and this calls for independent information to be able to really assess model skills independently of what is used to tune them why is it so important because we use these models to understand our records we can run these climate models for the last 50 years and look at the isotope temperature relationship and deconstruct what is driving this relationship we can run these models in response to volcanic eruptions we can run them in response to orbital forcing or ice age conditions and by doing that we have a two-way process we use model outputs to understand our records what are the physical drivers of the signal we get in the perfect model world and at the same time we also use the data completely independent from the model development to assess whether we get the right signal and maybe if we get it right for the right reason with respect to the water cycle and since the the last decade basically using laser instruments we are now able to monitor water vapor isotopes and in the past it was totally invisible in between precipitation events we are able to model to monitor that continuously on a sub-hour basis in the field in polar regions but also thanks to satellite measurements or remote sensing from the ground also obtain integrated information on the troposphere water stable isotope composition and this is relevant for key processes in the climate system in relationship for instance with convection and water vapor profiles in the atmosphere and it is also relevant for processes at the interface between the surface of snow and the atmosphere before going into the most detailed results i just want to share with you the longest records that we have obtained from ice cores 800 000 years so you can see information on the composition of the atmosphere and the content of methane in the atmosphere in the upper panel with the recent period and a number of people used this type of records to try to find the term to illustrate how much human activity is disturbed the atmospheric composition for instance using the term Anthropocene you can look at the concentration of carbon dioxide in the atmosphere through time and today's perturbation we are now I think at 410 ppm in the atmosphere from the monoloa record this is a colorful estimate of Antarctic temperature changes based on a quantitative interpretation of water stable isotopes from the oldest available ice core locally you can see that the range of temperature change is about 10 degrees between warm phases and cold phases in Antarctica it is about twice larger that that denotes the global mean illustrating a process described as polar amplification compared to global change in average temperature you can also see the close interplay between changes in temperature in Antarctica and changes in concentrations in greenhouse gases in the atmosphere and this closing interplay goes in multiple ways um Antarctic temperature is sensitive to changes in greenhouse effect obviously the Antarctic climate is closely linked to what is happening in the southern ocean which plays a key role on glacial interglacial changes in the global carbon budget and in atmospheric suit to concentration changes in polar climate plays a significant control at the global scale given the larger amplitude of temperature changes and the driver of these changes lies in changes in the distribution of incoming solar radiation through seasons and through latitudes and it is illustrated here by just one dimension of this astronomical forcing the well-known changes in northern hemisphere summer insulation I always like to refer to that in terms of the importance of conceptual approaches in climate sciences the first to calculate accurately changes in the distribution of incoming solar radiation was Milutin Milakovich a mathematician from Serbia in the 1940s and he suggested that the driver of ice ages would be major shifts in the amount of solar insulation in the northern hemisphere in summer his reasoning was that for an ice sheet to survive to to to be created you need the snow formed in winter not to melt in summer so you need cool summers as a driver of ice ages so he proposed the timing for past insceptions of ice ages and this was in fact verified decades after that using deep sea sediment records for instance allowing to estimate past changes in sea level or from the Antarctic records starting in the 1980s over these time periods and in between this driver this indicator of the driver and the full earth system responses you have the close interplay between the orbital forcing changes in the cryosphere expansion of ice sheets climate feedbacks carbon cycle feedbacks and this illustrates why it is so important to test our understanding of the feedbacks within the earth system including ice sheets including the carbon cycle against the largest major changes that have been experienced so now I'm going to focus on the tools that I've been using and what we can learn from today's monitoring of water stabilizer tops these are places where we have been running monitoring sometimes during a season in Antarctica just one month in the summer in the warm summer season where it's easier to have field work and where we have enough moisture in the air to be able to deploy these field measurements given today's technologies and for the Arctic we've been working in Iceland, Svalbard, South Greenland above the Greenland ice sheet and on the coast of Greenland trying to match with a major pathways of moisture transport by the atmosphere towards the Arctic as illustrated by these arrows which come from atmospheric reanalysis what have we learned from this monitoring we have seen that they are by doing a parallel sampling of precipitation surface water vapor and surface snow that there are day-to-day changes in surface vapor and precipitation driven by synoptic weather as you have new air masses arriving they bring with them an isotopic fingerprint linked to phase changes during atmospheric transport it can change quite quickly from day to day and it is well recorded in water stable isotopes especially second order parameters like deuterum excess in summer we have evidence journal variations which are located which are related to two things evaporation from the snow surface or sublimation but also movements of course of the boundary layer along daytime and nighttime even during the polar continuous day we have evidence that they are related to snow air exchanges which are caused by a process described as snow metamorphism this process is extremely important this transformation of the snow grains acts on the reflective capacity of the snow surface it's albedo it is important for deformation from snow to ice and even glaciological properties of the ice so we have shown that water stable isotopes are a tool to provide a quantitative information about the snow metamorphism processes and this is now gradually being used by the cryosphere community with the goal for instance to have snow models including water stable isotopes to be able to test whether they represent correctly the water cycle aspects associated with water vapor in the snow we have also shown that there are changes in surface snow isotopic composition even when there is no snowfall what is happening is that changes in air masses affect the vapor isotopic composition and during the exchanges between water vapor molecules in the air and in the snow part of this signal is picked up in the surface snow this is a challenge to the usual interpretation of ice core records our life was easier when we thought that the signal was just created by successive snowfall events with smoothing processes in between snowfall events now we understand that our records might be more continuous than we thought also capturing an atmospheric signal even when there is no snowfall at the moment we are not yet able to quantify the importance of this process but I wanted to share it with you because it is important for the interpretations of ice core records that I will show next so the implications are the following snow surface isotopic composition is of course influenced by snowfall especially in large amounts but it is not just a precipitation weighted signal we have an open question about what controls and what is the exact magnitude of the post-deposition processes related to water vapor snow exchanges and for that we need more data sets in different regions so the young scientist who has done some of the work in Antarctica is now in Tibet trying to obtain this type of measurements under high summer or spring insulation in Tibet glaciers to try to see if the processes that we have observed in Greenland and Antarctica are enhanced when you have a higher amount of insulation in a different context and the Davos snow laboratory has also done a number of laboratory experiments to try to quantify and isolate the processes at play and there's work underway to implement snow models with isotopes now I'm going to try to show what we can learn from punctual measurements that is in fact valuable at a larger scale first thing that we have done in July 2012 we were extremely lucky to capture water vapor isotopes during a major extra tropical storm so-called an atmospheric river event that event was remarkable in that it caused widespread melting at the surface of the Greenland ice sheet and we had measurements in Bermuda we had measurements in South Greenland and measurements above the Greenland ice sheet of water stable isotopes and thanks to this luck in a way for the first time we were able to confirm that the isotopic composition of the vapor from the marine boundary layer where most of the evaporation was occurring feeding this extra tropical storm the deuterium excess signal was preserved during transportation above South Greenland and above the Greenland ice sheet this has been had been hypothesized by physicists since the 1960s based on theoretical distillation models this was supported by analysis of atmospheric general circulation models equipped with water stable isotopes and for the first time we were able to demonstrate that this initial water vapor flag or signal was preserved the second thing that we are doing now is using local measurements from from for instance South Greenland from Iceland or from Svalbard and every day we have our measurements we combine them with calculations of atmospheric back trajectories and we project the source signal to the starting point of the air mass moisture transport we are using diagnosis of moisture transport in back trajectories developed by Harald Sodman in Norway and when we do that we of course can see a spatial pattern emerging beyond punctual local measurements and what we see is a pattern where we have lower deuterium excess when we have evapotranspiration above the continents or in the temperate oceanic regions and we have high deuterium excess values when we go closer to polar regions it has been suggested that this is driven by kinetic processes occurring with fast evaporation especially at the margins of sea ice where you can have air masses with a quite low humidity levels provoking in fact intense evaporation but associated with a strong kinetic effect and a flag that is then captured why am i focusing on that because we are also using our data to test if today's atmospheric general circulation models get these spatial gradients right so the data are independent of any data set used to develop the representation of the water cycle in models they are independent of the precipitation data used to adjust the representation of water stabilisotopes in atmospheric models and if you look this is the german atmospheric model ihan in its fifth version one of the models that has the best skills for water stable isotopes in renon and in ontartica and that we are using very frequently to better understand the signals we have in ice course but as all the other atmospheric models this model fails to capture correctly spatial patterns in deuterium excess glacial interglacial changes in deuterium excess so in the second order isotopic parameters and we have long been wondering why we had this challenge by looking at today's boundary layer water vapor isotopic composition oxygen 18 here deuterium excess here we can see that some of the problems emerge from the inability to represent today's patterns we don't get the gradients right for oxygen 18 from the temperature regions to the arctic region and for deuterium excess we completely fail to capture the fact that there are higher values in the arctic and lower values in the subtropics this is an open question for those involved in the development of atmospheric models in the parameterization of evaporation in the representation of water stable isotopes in the evaporative flux to understand why we do not get the signal right it can be linked to the representation of soup soup grid processes at the sea ice margin that's a possibility but it may not be the only one and therefore there's a way here to be able to better test atmospheric general circulation models for the input flux the evaporation flux at the ocean surface so the take home message is that we have model skills for entrances on all or summer to summer variations in oxygen 18 but not for the mean level and the variance of deuterium excess and this is linked to the inability to resolve the spatial structure of the initial marine boundary layer isotopic composition so i'm a paleoclimatologist i'm interested in understanding past climates but through these measurements that we have recently implemented in order to better understand our core tool we open also new possibilities and new applications of the use of that tool in other communities so as a take home message using surface water vapor monitoring we can have access to journal or weather scale processes there's a potential to quantify snow metamorphism processes we need further monitoring coordinated from multiple sites to be able to evaluate atmospheric models and we confirm that some of the key assumptions related to the interpretation of ice core records such as preserving a moisture source signal makes sense i'm now going to show a few examples of recent results obtained from ice cores in Greenland and Antarctica and i go from the present to the past and focus on the warm periods so a recent example of a work we've been doing has been to focus on the most recent trends in Greenland the last 30 years and we've looked at instrumental records and as you can see they mostly come from coastal regions and some of the some of the camps we've looked at reanalysis for instance the european weather forecast reanalysis and we have also looked at the outputs of a regional atmospheric model specifically equipped for polar regions developed in belgium driven by the same reanalysis and compares them with the the very precise reconstructions of temperature trends that we have obtained from central green on ice core records combining water stable isotopes with borehole temperature profiles so you can see for instance at NIM that's the trend in the last about 30 years has been more than two and a half degrees during that time period it's amongst the largest in the world and the amount of this trend in annual mean temperature is underestimated if you use reanalysis for a surrogate of missing observations you can also see that you may get a better pattern sometimes when you have downscaling approaches using dynamical regional models but some of the mismatches between our observations and model outputs may be also linked with the representation of albedo processes linked with snow metamorphism processes that are most of the time not implemented at all the other example that i want to give about what we can learn from ice core records on the recent past are associated with the climate response to major volcanic eruptions this is a reconstruction of the volcanic of the radiative forcing so the perturbation of radiation budget at the top of the atmosphere associated with the major volcanic eruptions of the last 2 500 years and i will focus on the the last eruption that's led to the largest sulfur deposition in polo region and it is associated with a volcano called samalas in indonesia and its eruption occurred in the 13th century just a few comments on this long record the average recurrence time of pinatubo type pinatubo like events like the one in 1991 is between 4 and 137 years with a median recurrence time of about 40 years we can see a remarkable event so the samalas one but also the tambora eruption which occurred at the beginning of the 19th century and led remarkable prints in prints in historical records it was called the year following that event was called in europe and other places of the world the year without the summer because of the cold conditions so using this be decadal variability with volcanic events acting as pacemakers for such events this is in fact quite important if you want to look at trends in the north Atlantic ocean circulation and you have to deconvolve the effect of volcanic eruption in order to detect what could be a human in use trend in circulation as well so there's really a potential to further explore the predictability following major eruptions informed by insights from past climates i'm now going to show the key questions that we have for recent Antarctic climate if you look at the assessment of the IPCC in the last report and key uncertainties were associated with recent changes in Antarctica it is one of the places where it is hard to detect trends against variability it is one of the places where climate can be affected by the loss of the stratospheric ozone layer in addition to greenhouse gas emissions and it is one of the regions where there was no key conclusion about greenhouse induced influence on recent trends so what we tried to do was to work on the longer context not just satellite records not just instrumental temperature records but what do we have for the scale of the last 200 years what do we have for the scale of the last 2000 years and we are really at the limit of what we can do with ice core records so these are the noisy observed trends from satellite records or instrumental records since 1979 so you can see a warming trend in the central part of eastern Tartica a strong warming trend in western Tartica contrasting trends with some areas with decreased sea ice extent and other areas with a trend of increased sea ice extent and you also have contrasted trends in sea surface temperature opposite in various sectors around Antarctica now I would like to show what we see when we place these last 30 years in a context of 200 years so we have done that by compiling ice core records borehole temperature records water stable isotope records and also some records from a coastal deep sea records around Antarctica so it's really trying to make use of any single information that we have that can help us characterize temperature variations around Antarctica in the last 200 years so what do we see we see that in fact the trend in the Antarctic peninsula stands out against a background variability from the previous centuries but this is not the case in the other regions in most of the other regions recent trends are just part of multi decadal or longer variability that we just do not sample when we focus on 30 years so the only signal that we see is a trend of the southern annular mode especially in summer and fall and we clearly show that 30 years of observations in this region are insufficient to detect and attribute temperature trends which is by itself as a new information we then try to expand that not just to the last 200 years but to the last not just to the last 200 years that's 2000 years if we want to go back to 2000 years we are losing some of the records that have very high resolution so instead of trying to work year by year and season by season we worked on timescales of five and ten years because we are using records that do not allow that accurate seasonal records we looked at the coherency of temperature across regions from year to year and we defined a number of regions that are known to have coherent temperature variations from the short instrumental records and from regional climate modeling so we have all together seven different regions so I will just drop that and show that for the last 2000 years you may not see it but the timescale is from zero to year 2000 common era and you see the patterns of the records obtained for each region from high score records only this time the key findings us is that the only signal coming out of the background natural variability is the warming signal of the Antarctic peninsula for all the other regions we are still within natural variability and the second key signal that we see is a significant decreasing trend in several regions most of them actually and the exact reason for that decreasing trend is not fully understood we need to better understand how Antarctic climate is responding to changes in frequencies of volcanic eruptions this is still a very uncertain topic and we need to understand what part of this trend is a trend associated with the response of the Antarctic climate to orbital forcing over multiple thousand years the problem that we have is that today's climate models do not capture such a trend for most of them so we are limited both by the tools that we have to model Antarctic climates but also by the understanding of the key drivers of Antarctic climate trends at this time scales so at this time scale we see the emergence of unusual trends for the southern annular mode for the Antarctic peninsula warming but not circum-antharctic recent warming we have no clear signal following major eruptions with the records that we have that are not accurate enough and we have this robust and significant cooling of Antarctica over the time scale of the last basically 2000 years but we do not yet understand the cause and the mechanisms so not to finish with a frustrating note I would like now to focus on one of the strongest signal that we have and it is the signal of the last warm period about 125 000 years ago where we have a strong change in the distribution of insulation at the earth's surface and a strong signal this period is relevant because at the global scale temperature was between one and two degrees above pre-industrial levels from the synthesis of available information from all climate paleoclimate records sea level was about five to ten meters above present day so there is a clear indication that during this last interglacial period ice sheets reacted to warmer polar regions we don't yet know exactly the amount of ice lost by the Greenland ice sheet by the Antarctic ice sheets and it is extremely frustrating because it is the single benchmark we have to test the ability of our ice sheet models to actually represent a known episode of significant melt of Greenland and or Antarctic ice sheets so I will try to explain what we have done with our ice core records this is the current interglacial period so I just showed you a minute ago a zoom on the last 20000 years with a decreasing trend this is the current interglacial period the last 12 000 years in Antarctica and you can see small variations in Antarctica a small warm phase that was locally one degree above pre-industrial levels we are now moving before the last ice age to the previous warm phase about 125 000 years ago and you can see in Antarctica compared to today's level for each record that we have higher oxygenating values that are understood to reflect warmer conditions and our understanding of the warming suggests that it is between three and five degrees depending on the site we don't have any climate model able to produce such a warming or such an isotopic amplitude in direct response to orbital forcing so it is a major challenge to understand why we are not able to reproduce that with our models if we look at Greenland Greenland was also warmer at that time because of changes in summer insulation and this is well understood this is our estimate of Greenland temperature changes more than five degrees compared to pre-industrial levels over several thousand years and because we still had ice at that place in Greenland and we have an estimate of the changes in elevation at that place in Greenland we can use that to say combined with ice sheet models that the Greenland ice sheet did not lose more than the equivalent of one and a half to four point three meters of equivalent sea level I told you before that the estimated range of average sea level increase is five to ten meters so this implies a significant loss of ice from the Greenland ice sheet and it is the only information that we have at this stage to be able to provide any indirect quantitative constraint on how much the Antarctic ice sheet lost in terms of ice during that period if you look at the average of global climate models driven by changes in the Earth's orbit for this period this is from the last IPCC report you can see that they get a small amplitude of warming in annual mean in Greenland they really face difficulties in getting to the amplitude that we have in our data whether you use temperature whether you use oxygenating we really do not get the amplitude neither in annual mean nor in precipitation weighted signals for Antarctica most of the climate models show in fact no signal during that period while it is the warmest signal we get in all our Antarctic ice cores of the last 800 000 years so this is an unknown point and several people have tried to explore that for instance maybe these simulations are not right because the Antarctic ice sheet was reduced and this was not accounted for so people tried for instance to make simulations with a strongly reduced West Antarctic ice sheet or a West Antarctic ice sheet replaced by an ocean and they look at the oxygenating that is simulated by their atmospheric model in response to these stronger conditions they never get the amplitude that we have in our ice core records so even by lowering totally making the West Antarctic ice sheet flat does not allow to reconcile climate models with the ice core data one hint came recently from the same author Max Holloway from Bristol during his PhD thesis where he tried the simulation by forcing the sea ice around Antarctica to significantly retreat by more than half compared to today in winter in his coupled climate model simulation and only by doing that was he able to reach an oxygenating magnitude in his simulation or a level of warming that is compatible with our ice core records why do you need artificially to reduce Antarctic sea ice to be able to get a signal comparable to observations does it mean that coupled climate models driven only by orbital forcing fail to simulate correctly processes linked with the southern ocean circulation and Antarctic sea ice are there issues linked with the length of the simulations are there issues linked with the interplay between melt of the green on ice sheet changes in ocean circulation and the interplay with what's happening around Antarctica these are still open questions but they are important because this is the last case when we know the Antarctic climate was significantly warmer than today and we know that's the Antarctic ice sheet was significantly reduced so I just try to illustrate what we learn when looking at a big signal the largest warming we have in Greenland and Antarctica there's evidence that warmer polar conditions are associated with significant retreat of both ice sheets at the moment isotopic anomalies and warming are underestimated by state of the earth climate models ice sheet retreat and reduced sea ice extension may be able to explain the ice core signals but we need more work in that direction and thus ice core records in that way could provide unique constraints on past changes in ice sheet topography and sea ice extent so that we can test our ability to simulate the associated earth system feedbacks so to finish I try to illustrate what how do we make progress in understanding the signals that we archive in ice ice cores where improved knowledge comes from but also what are some of our knowledge gaps especially in terms of having confidence in our modeling tools and the ability to resolve regional changes in the last centuries to millennia especially for natural variability and to decipher the first response to volcanic eruptions in this case I illustrated an example of a case where we have very strong polar warming during the last interglacial period we can provide refined temperature reconstructions we have a potential to constrain past changes in sea ice extent and in ice sheet topography so there's a potential to learn more from these past natural experiments on the earth system to inform confidence to the tools that we use for projections and finally I didn't show that but we also have a full database of past abrupt change and the bipolar dimensions of these abrupt changes during the last ice age with very high resolution and there are still a number of open questions about the exact causes of these abrupt shifts linked with the reorganization of the atlantic overturning circulation and what we can learn from these past abrupt changes in terms of thresholds in the earth system for today and for the future thank you very much for your attention thanks Marie for this really nice talk and travel to the past the floor is open for question the first one is there thanks Valérie for this wonderful talk so I have a very naive question so you have shown us that the new data set challenge the classical interpretation of the ice record has just precipitation weighted signals so could that be also a challenge to our ability to use them as a quantitative isotopic thermometer or maybe naively I mean when you when we when we when you show us the temperature for instance in the last ice ice age period what are the error bar on the temperature we should put on top of this these are excellent questions so you understand that I wanted to show these issues before showing the reconstructions in terms of temperature reconstructions we for Antarctica we have evidence that our uncertainty is about 30 percent on the coefficients relating isotopes to temperature at the glacial interglacial scale there are some simulations done with atmospheric models trying to explore the validity of the isotope temperature relationship in climate warmer than today but we don't get them in response to orbital forcing so we signed from the British Antarctic Survey looked at warmer worlds in response to higher CO2 concentrations and and she suggested that using the classical isotope temperature relationship would lead to an underestimation of the magnitude of temperature changes so it's not a symmetric error bar it might be skewed for cold periods we have quite we are quite convinced that we are within 30 percent for warmer than today periods we might under estimate the amplitude of temperature changes in Antarctica in Greenland it is a little different because in Greenland we have more constraints from other temperature methods and in Greenland so we have borehole temperature data we have methods using firm measurements or gas isotope measurements so we know that when we use today's isotope temperature relationship we underestimate past changes for sure and for Greenland our conversion to temperature is informed by these other methods so it is much more robust so that's just for quantification so we have confidence that it is a reliable temperature signal we we need to refine the effect link to snow metamorphism we are not yet equipped with the possibility to quantify this effect we have that floating when we compare ice core records and climate model output because we are comparing the precipitation signal simulated by climate models to an ice core record that might be influenced by post deposition effects so we are aware of that we write that very clearly in our work now and we need tools that resolve these processes to be able to go further could this explain part of your third point no no we we are convinced that it cannot explain the mismatches that I showed especially because when we are looking at warmer than today's periods we know that the rate of precipitation was larger so the frequency of snowfall for instance and it leaves less space for potential snow air exchanges in between snowfall events hi thanks a lot I have a question on on modeling okay in the sense that you know in the in the 70s people put a lot of effort to you know learn and improve the modeling of the atmosphere different processes the ocean was coupled we got to hundreds of years and there seems to be a major barrier in including ice the dynamic modeling of ice I mean if we want to merge daytime models in some sort of data simulation whatever this will be on this scale where do you feel how do you see the perspective of basically going beyond the millennial timescale with the non toy models because of course we can reproduce the ice ages and so on with toy models but what about something looking like a gcm or should we think of gcm since the right tools to start it this sort of timescales it's not obvious in my opinion in a way that's why I added the abrupt changes as the last point because I think at the moment what is missing is really exploring what we know in terms of abrupt changes I may just let me just show a slide for that so this is just this is a it's not the same orientation of the axis as in the past figures this is the geological way of showing time so the present is here and the past there and so you can see here the current warm period and then the ice age and and this is the recording green on the abrupt changes and this is the Antarctic counterpart and it reflects a bipolar seesaw mechanism so what I just wanted to say is that we have not yet fully exploited this whole database of abrupt events because we don't have the modeling tools with the the resolution of the key processes what exists at the moment is for instance an american simulation of the last deglaciation about 10 000 years where people used an ocean sea ice atmosphere model and they used the sort of try and test and revise approach so that they basically injected freshwater fluxes in the north Atlantic ocean so that's the outcome of the simulation would be as consistent as possible with some of the key records that we have of ocean circulation of green on temperature changes and this is an interesting framework because it's like trying to do that simulation in a way not in a fully comprehensive physical or statistical way trying to have a comprehensive climate model framework to also help better understand the processes at play but by doing that you are killing the conservation of water for instance because you don't have the ice sheet response so it's one step into this direction but I think there's a potential here to really exploit the non-linear behavior of a regional climate around the north Atlantic to better understand the threshold processes over multiple events in various ice sheet contexts in various orbital contexts in various greenhouses concentration contexts so basically various gradients of density context in the ocean it's like trying to produce a best guess but there's a potential especially with faster models there's a potential really to better exploit this variability very nice talk Valery and I have a I'm curious curious about one of the figures that you showed the longest record in Antarctica because there is this conspicuous period to the right of 400,000 years ago yes where the insulation was very low but sea level and temperature just to the right of 400,000 years everything is very high but the insulation is very low what may have happened there so behind that pattern of change in insulation you have the characteristics of the astronomical variations of the earth's orbit so it is a time when the eccentricity of the earth was quite small and therefore the effect of the other orbital parameters is tuned okay so that happens with a pace of about 400,000 years as you can see for the near future as well so what happened here was that you didn't have a strong orbital forcing in the magnitude of changes in summer insulation in the northern hemisphere so you could have a long warm phase more than the usual 10,000 years so that warm phase about 400,000 years ago lasted about 30,000 years in a low orbital forcing context basically and this period is really interesting because there are issues about the ice sheet response to polar warming whether what is the questions are whether what is important whether the amplitude of the warming is key or the duration of a warm period is key to trigger for instance ice sheet instability so this is one of the periods that is also a focus of studies for what we learn from past warm worlds but we are out of reach of many accurate dating methods and most of the records are associated with more uncertainty than for the most recent warm phase that would affect the polar jet the waving of the polar jets and etc for that I cannot answer but we have evidence from pollen records we have evidence from lake records from green on ice street records for mild conditions in the high latitudes of the northern hemisphere we have evidence for a significant higher sea level compared to today as well but we have no constraint on the our very limited constraints on the contributions from green on the formant arctica and we don't have a green on the ice core record either for that period hi roberto buitza from the european center so i'm not surprised about what you say about the quality of our modus i mean we are now running weather prediction using couple modus and dynamical sea ice and we know for example that in ocean with a 25 kilometer resolution is not good enough to correctly simulate the gulf currents with all the implications so clearly our modus even at resolution we're running for nwp are not good enough there's no modus don't have enough layers and type of snows so i'm not surprised that we are not able to go back and simulate this event so perfect comment while i want to make another point which is we have been despite this we have been trying to go back further with our reanalysis so we have done a couple reanalysis down to 1900 couple ocean land atmosphere sea ice with an ensemble of components so that we can estimate the uncertainty in the reanalysis and we are actually thinking to go back further even to 1800 or so there are clearly a lot of difficulties because the data are less and less but i think it would be interesting to test and compare reanalysis down at least to 1900 not just 1970s or so on than with your entry because maybe maybe we can learn about the performance of our model and we can get some understanding of the events say from 1900 onwards so i think it's a good opportunity to share a dream my dream would be that operational models like this one used for weather forecast but also for reanalysis would be equipped with things like water stable isotopes because that would be extremely interesting in having a data set for benchmarking them completely independent of what is assimilated given the interest in water cycle variables and the correct representation of moisture sources and it goes across scales it's important for feedbacks it's important for a number of dimensions and with models now equipped with water stabilizer top some people in japan in the us are exploring the possibility to assimilate these indirect archives before the instrumental period so they use classical assimilation methods to noisy records it is really work on the progress but i think it has the potential to be able to provide us with the best guest of the state of the atmosphere and maybe some other aspects prior to the instrumental period especially for ontartica which in fact might be important for having a right initial state for the southern ocean as well i'm fred i'm working in the climate section at icdp also i have one just i want to what you'd like to ask your theory about the little ice age because i've seen you've shown some records that that covered that period i wanted to ask you do you think it's real or just anecdote based is it and what is the most likely forcing mechanism is it solar or volcanic or internal climate variability what is it yeah so this is just an illustration of what we are talking about so these are different estimates of changes in global mean temperature in black we see the instrumental warming and in purple some estimates of global mean temperature changes on the scale of the last thousand years the scale of the last ten thousand years and glacial interglacial changes so when we speak about the little ice age we are speaking of small scale variations this is from ocean records so small scale variations at the global scale looking at northern hemisphere surface temperatures something that might be half a degree temperature anomaly with a cooling studied in the 14th century until the 19th century compared to previous periods the ipcc last assessment reports provided an an assessment of the amount of temperature changes and an assessment of at the time so back to 2012 what was understood to be key drivers of this anomaly and it highlighted the importance of the frequency of major volcanic eruptions and the challenge in attributing this mild cooling anomaly to a single driver because there is covariance in between the occurrence of major volcanic eruptions and a minima in solar activity so you cannot it is not trivial to separate the role of each of them and that's also a reason why the paleoclimate research community is also interested in broadening that back to the last two and a half two and a half thousand years so that we can look at more centennial variations and more periods with more or less active volcanic eruptions or solar activity with the hope to be able to have a better understanding than focusing on one single period and I try to be quite objective and I personally think that there's a key role of volcanic activity in the beginning of the warm period and in the ocean heat content anomalies as well from the work I've been doing on volcanic eruptions hello thanks great talk a question from an ignorant my understanding is that the extent and and the geometry of the ice sheet has a considerable implication on your isotopic response and then this is affecting your signal and then the deposition and both the positions are contributing are adding errors to the system is there a way by which for instance you extend your data sets to sedimentological records by which you could maybe this would give you a little bit more off so for the ice core records they are as you say potentially influenced by changes in the ice sheet topography and geometry either just locally because your sampling site is moving in haze or by indirect effects of changes in ice sheet topography affecting regional atmospheric circulation and transport processes so that's true and when looking at comparisons between climate model outputs and past climates for specific periods we actually use everything available so we use deep sea records and estimates of sea surface temperatures especially for the growing season of organisms used to reconstruct these temperatures we are using lake sediment records we are using insights from pollen within international efforts for instance from the paleoclimate modeling intercopyason project or for the past global change project the key issue is that when you focus on the diversity of records each of them is associated with a specific uncertainty or specific recording process and an age scale uncertainty and it is just to stress the fact that when we work on Greenland and Antarctica records only we have a coherent time scale because we use a number of markers common between Greenland and Antarctica to align our chronologies so when we compare with other records from other archives we have more uncertainty and it is particularly important for the last interglacial period because the warmest periods did not occur simultaneously in various regions of the world and some of the initial estimates of global mean temperature were biased because the initial work was just done by trying to take the warmest signal everywhere assuming it would be synchronous and we now know that it was not if there are no more questions I think we need to leave the floor to the student and to thank Valerie first and then we