Regarding the technical innovation for deep ice core analysis side of the project, several PhD projects led to the ongoing development of new analytical and innovative techniques to analyse deep section of ice cores with the highest resolution possible and minimizing the amount of ice needed by allowing several analyses in a row on a same section, with micro-destruction techniques, as well as allowing the analysis of multiple impurities, organic and inorganic particles.
Regarding the quantification of processes responsible for the climate signal in the ice core, the different PhD projects related to this aspect contribute to better understanding the processes that can affect the climate signal, in different parts of the ice (very deep ice, basal ice, firn, surface snow) and for different types of information (interpretation of gas concentration, dating methods, surface and subsurface processes affecting proxies, changes in ice microstructure and air inclusions…). All the projects have been mainly focused on preparing the samples or observations to get the first datasets required to conduct analysis for a better understanding of the ongoing processes.
Concerning the modelling and statistical tools used for ice core-based climate reconstructions, the different PhD projects related to these tools will enable the development of new modeling and statistical tools that will be used for a better interpretation of the data collected through ice core analysis (in particular, water isotopes signal, dating of the ice and age-depth profiles, and glacial dynamics and orbital forcing influencing climate proxies contained in ice cores), and therefore will participate to improve ice core-based climate reconstructions. All the projects have been mainly focused on preparing the statistical tools and models that will be useful for the future analysis of deep ice core data, by either developing new model or adapting existing ones to integrate new functions and better resolution and producing the first sets of data. Some evaluations of the new model results with existing dataset have already been performed.