No abstract

No abstract

Numerical models of sea ice play an important role in understanding the changing Arctic and allow researchers to predict the dynamic response of sea ice to different environmental conditions. High resolution forecasts from predictive models are also becoming increasingly important due to increased human activity in the Arctic. The recent decline in Arctic sea ice has lead to more traffic in the Arctic Ocean for fishing, resource extraction, tourism, cargo shipping, and military purposes. Sea ice models that can explicitly capture small discontinuities and fractures in the ice are particularly valuable for navigation. For example, IICWG (2019) lists high resolution information about compression and pressure ridges as one of the most important things missing in current operational ice products.Many sea ice models, such as those used in global climate models, employ continuum approaches where the sea ice is discretized with an Eulerian mesh and the ice is modeled with constitutive models such as viscous-plastic (VP) or elastic-viscous-plastic (EVP) rheologies (Hibler, 1979;Hunke & Dukowicz, 1997). Recent studies, such as (Bouchat & Tremblay, 2017) and (Hutter & Losch, 2020), have shown that VP/EVP rheologies can capture important statistics about largescale sea ice deformation. On smaller scales however, it has been shown that

As Arctic conditions rapidly change, human activity in the Arctic will continue to increase and so will the need for high‐resolution observations of sea ice. While satellite imagery can provide high spatial resolution, it is temporally sparse and significant ice deformation can occur between observations. This makes it difficult to apply feature tracking or image correlation techniques that require persistent features to exist between images. With this in mind, we propose a technique based on optimal transport, which is commonly used to measure differences between probability distributions. When little ice enters or leaves the image scene, we show that regularized optimal transport can be used to quantitatively estimate ice deformation. We discuss the motivation for our approach and describe efficient computational implementations. Results are provided on a combination of synthetic and satellite imagery to demonstrate the ability of our approach to estimate dynamics properties at the original image resolution.