Arctic sea ice plays a critical role in the climate system. Therefore it needs to be modeled accurately to make precise climate predictions of the current anthropogenic climate change. Most of today’s climate models simulate sea ice motion using a viscous-plastic (VP) rheology with an elliptical yield curve and a normal flow rule [1]. This rheology gives accurate predictions at low resolution but features discrepancies at high resolution when compared to observations. Observations show that narrow lines of deformation dominate the pattern of sea ice motion. Models using the VP rheology feature these fracture lines but overestimate the intersection angles between them [2]. To understand and solve these differences, we study the creation of fracture angles using idealized compression experiments. We test several yield curves and flow rules and investigate their effect on the angles between fracture lines. The results show that: first, the fracture angle depends on the shape of the yield curve as well as the flow rule, in agreement with Roscoe’s angle [3]; second, the elliptical yield curve with normal flow rule cannot create angles lower than 30° in uniaxial compression; finally, implementing a non-normal flow rule can lead to fracture angles as low as 22°. Results also show that sea ice dynamics in models disagree with sea ice observed granular behavior. With the new knowledge gained from these idealized experiments, we can now define new VP rheologies for more accurate sea ice modeling, e.g., with Mohr-Coulomb or teardrop yield curves. REFERENCES: [1] Hibler III WD: A dynamic thermodynamic sea ice model, Journal of physical oceanography. Jul;9(4):815-46, 1979. [2] Hutter, N. and Losch, M.: Feature-based comparison of sea ice deformation in lead-permitting sea ice simulations, The Cryosphere, 14, 93–113, https://doi.org/10.5194/tc-14-93-2020, 2020. [3] Roscoe, K. H.: The Influence of Strains in Soil Mechanics, Géotechnique, 20, 2, 129-170, 1970.