| Summary: | In the context of global warming, monitoring the thickness and mechanical properties of sea ice is a major challenge in modern climatology. In particular, the heavy logistical constraints of polar environments, and the lack of accuracy of satellite remote monitoring methods, are obstacles to improving climate models. As a result, the decline of sea ice, which has been accelerating over the last four decades, is difficult to predict on short or longer time scales. For example, while only 10 years ago, the Arctic was expected to be ice-free in summer from the 2050s, the latest forecasts indicate that this could happen as early as the 2030s. Accurate and regular measurements of pack ice properties are crucial to better anticipate future changes. In this presentation, we introduce methods to demonstrate that it is possible to monitor sea ice passively, based on the ambient seismic field recorded continuously in situ. In particular, we introduce analysis methods based on: - seismic noise interferometry to extract the Green's function of guided waves in ice - deep learning algorithms to classify the recorded signals - guided wave dispersion for recovering the thickness, Young's modulus, Poisson's ratio, and density of the ice pack, via Bayesian inference Based on these analyses, we demonstrate that it is possible to monitor the temporal and spatial evolution of these parameters at the scale of a few kilometers, with a temporal resolution of a few hours.
|
|---|