Icechrono1: a probabilistic model to compute a common and optimized chronology for several ice cores

Polar ice cores provide exceptional archives of past environmental conditions. The dating of ice cores and the estimation of the age-scale uncertainty are essential to interpret the climate and environmental records that they contain. It is, however, a complex problem which involves different methods. Here, we present IceChrono1, a new probabilistic model integrating various sources of chronological information to produce a common and optimized chronology for several ice cores, as well as its uncertainty. IceChrono1 is based on the inference of three quantities: the surface accumulation rate, the lock-in depth (LID) of air bubbles and the thinning function. The chronological information integrated into IceChrono1 are modeling scenarios of the sedimentation process (accumulation of snow, densification of snow into ice and air trapping, ice flow), ice- and air-dated horizons, ice and air depth intervals with known durations, delta depth observations (depth shift between synchronous events recorded in the ice and in the air) and finally ice, air or mix stratigraphic links in between ice cores. the inference problem is formulated as a least squares optimisation, implying that all densities of probabilities are assumed to be Gaussian. It is numerically solved using the Levenberg-Marquardt algorithm (thus assuming that the model is almost linear in the vicinity of the solution) and a numerical evaluation of the model’s Jacobian. IceChrono1 is freely available under the General Public License v3 open source license.

Details