Section: Software

DatIce

Participants : Bénédicte Lemieux-Dudon, Habib Toye Mahamadou Kele.

The Datice code ( [76] , [77] ) is designed to estimate consistent chronologies of several deep ice cores (i.e., depth-age relationships of the ice matrix and trapped gas). A cost function derived from Bayes theorem puts in competition the chronological constraints brought by heterogeneous observations (stratigraphic links between cores, gas and ice age markers, delta-depth markers, etc.), and the background dating scenarios simulated with glaciological models (firn densification and ice flow models). The minimization of the cost function provides optimal estimations of three key quantities from which dating scenarios can be derived: the past accumulation rate, the close-off depth which is the depth where the gas is trapped into ice, and the total thinning function. Uncertainties of the analysed dating scenarios (key quantities and chronologies) are assessed on the basis of the Bayesian formulation. This approach is innovative because:

• it relies on data assimilation techniques to calculate ice core chronologies and uncertainties;

• it applies to a large number of heterogeneous observations;

• it ensures consistency between the chronologies of several cores and the consistency between the gas and ice age scales.

The code has been used in several recent publications (see [68] , [87] for example).