The amount of snow on Arctic sea ice impacts the ice mass budget. Wind redistribution of snow into open water in leads is hypothesized to cause significant wintertime snow loss. However, there are no direct measurements of snow loss into Arctic leads. We measured the snow lost in four leads in the Central Arctic in winter 2020. We find, contrary to the general consensus, that under typical winter conditions, minimal snow was lost into leads. However, during a cyclone that delivered warm air temperatures, high winds, and snowfall, 35.0 ± 1.1 cm snow water equivalent (SWE) was lost into a lead (per unit lead area). This corresponded to a removal of 0.7–1.1 cm SWE from the entire surface—∼6–10% of this site’s annual snow precipitation. Warm air temperatures, which increase the length of time that wintertime leads remain unfrozen, may be an underappreciated factor in snow loss into leads.

The discrete element method (DEM) can provide detailed descriptions of sea ice dynamics that explicitly model floes and discontinuities in the ice, which can be challenging to represent accurately with current models. However, floe-scale stresses that inform lead formation in sea ice are difficult to calculate in current DEM implementations. In this paper, we use the ParticLS software library to develop a DEM that models the sea ice as a collection of discrete rigid particles that are initially bonded together using a cohesive beam model that approximates the response of an Euler-Bernoulli beam located between particle centroids. Ice fracture and lead formation are determined based on the value of a non-local Cauchy stress state around each particle and a Mohr-Coulomb fracture model. Therefore, large ice floes are modeled as continuous objects made up of many bonded particles that can interact with each other, deform, and fracture. We generate particle configurations by discretizing the ice in MODIS satellite imagery into polygonal floes that fill the observed ice shape and extent. The model is tested on ice advecting through an idealized channel and through Nares Strait. The results indicate that the bonded DEM model is capable of qualitatively capturing the dynamic sea ice patterns through constrictions such as ice bridges, arch kinematic features, and lead formation. In addition, we apply spatial and temporal scaling analyses to illustrate the model’s ability to capture heterogeneity and intermittency in the simulated ice deformation.

The mechanical deformation of sea ice has substantial influence over large-scale (e.g., > 10 km) ice properties, such as the ice thickness distribution, as well as small-scale (e.g., < 50 m) features, including leads and ridges. The conditions leading to sea ice fracture are frequently studied in the context of a uniform ice sheet. Natural sea ice, however, is highly heterogeneous and riddled with flaws. Failure occurs primarily as brittle fracture localized in space and time where stresses, and strain rates, locally exceed failure criteria. Here we seek to better understand the mechanical deformation and fracture of sea ice under such typical field conditions. In particular, we aim to characterize how forces propagate across an approximately 1 km^2 heterogeneous domain by observing the stress-strain field in an ice floe at resolutions required to capture pre-fracture elastic strains. The combination of instruments deployed allow a detailed view of the formation, propagation, parting, and subsequent shearing of a fracture in natural sea ice, providing field evidence of modes of failure in compressive shear. The relatively low change in stress observed within meters of the fracture location highlights the need for further research into disparities in sea ice strength measurements at laboratory and field scales. The ability of this system to capture strain concentration zones and to detect initial fracture hours prior to lead formation indicates the potential for predicting areas at high risk for fracture in an on-ice operational setting.

The thermodynamic growth of sea ice is a critical factor in the mass balance of Arctic sea ice, which has important implications for Arctic communities and the global climate. However, the magnitude by which snow atop Arctic sea ice limits thermodynamic ice growth is still not fully understood. Prior work has shown that the wind-driven snow redistribution could significantly modify the heat conduction through the snow cover and hence the rate of thermodynamic ice growth. However, the effects of snow redistribution on sea ice growth have not been quantified and are not well represented in climate models. We use observations from the MOSAiC expedition to show how different facets of snow redistribution can enhance or reduce heat conduction through the snow cover for the same mean snow thickness. The net effect depends on ice topography and environmental conditions. For example, snow redistribution onto young ice in April at MOSAiC reduced heat conduction by approximately 5-15%. We quantify the impact winter and springtime snow redistribution events on the heat conduction on deformed, level, and young ice. We explore the implications of these snow redistribution processes in the Community Earth System Model and discuss priorities for improving climate models.