QuSI-DEM version 1.0: A polygon-based discrete element sea ice model in spherical coordinates. Part I: Model description
Abstract. Discrete element modeling (DEM) provides a methodology for simulating the dynamics of sea ice from fracture through post-failure granular behavior. Because the approach is designed explicitly to represent the cohesive or frictional forces between distinct floes, it may offer a better representation than continuum-based approaches for some phenomena. As such, discrete element models (DEMs) provide a complementary approach to sea ice modeling at geophysical scales. However, many uncertainties remain regarding best practices in the application of the method. These range from choosing element shape, element scale, contact physics, and numerical implementation. Here we describe a novel polygon-based discrete element sea ice model, developed with careful attention to the implementation details and the fundamental response of interacting element-pairs. The crack patterns that develop as a result of sea ice fracture depend critically on the elastic and dissipative response of the material. Propagation of cracks can depend as much on the characteristics of post-failure element contacts as intact ones. So, distinct parameters are introduced to the contact physics in this model to provide control of the failure rate and timing of transitions in contact response and to constrain the amount of energy and momentum transfer in the contact failure process. Stresses can be communicated over great distances in sea ice due to its rigidity. This can present difficulty in specifying boundary conditions for simulations of localized processes as remote influences may be large. With this in mind, other like other models in its class, this model is formulated in spherical coordinates, and borrows from finite element mesh generation and refinement techniques that make it possible to simulate across scales up to the planetary without excessinve computational cost.
Overview:
Durski et al. presents a detailed description of the new sea ice Discrete Element Model. While several sea ice DEM models have been developed previously, this model has a number of unique features such as solving sea ice momentum natively on the sphere using quaternions, and having a energy conserving contact model. This paper described the model, while a part II companion paper describes verification and validation of the model and application to ice arching. The paper does a mostly excellent job of describing the model, although some additional commenting on its utility and applicability is in order (see below). Its also unclear what benefit dividing the paper into two parts does. Neither part could exist independently as a full paper, and the authors should consider merging into a single paper.
Comments:
The applicability of the model should be discussed in more detail in the introduction. There is a very short section in the discussion but this should be significantly expanded in the introduction. The authors should be clearer about what temporal scales the model is appropriate for as well as discuss the effect on applicability of the neglect of ridging and its consequent strengthening of the sea ice. The authors should also discuss the expected ability of the model to represent linear kinematic features on basin scales.
l47: The authors state that polygons have an advantage over disks in that they can represent 100% coverage. Firstly, Turner et al. 2022 developed a method to allow a disk based sea ice DEM to represent 100% sea ice coverage, although a polygonal DEM is able to much more simply. Secondly, even a polygonal DEM after the elements break apart and separate will not have a purely 100% ice coverage, either possessing small slivers of open water between elements or overlap. The 100% coverage only really applies at the start. Please clarify in the text.
l180: Do unbonded elements near each other not resist relative rotation? Contact compression forces are not necessarily directed along the line connecting centers of mass and would produce a torque? I guess here the authors are talking about individual contact points?
Section 3.4: Would extreme overlap not represent ridging, resulting in thickening strengthening ice that better resists further compression? Weakening in this situation seems counter intuitive, especially if it allows elements to pass through each other, surely not something really represented in nature?
Section 4.1.1: I found the description of how quaternions work and relate to the model in this section inadequate. This is one of the unique aspects of the model and the authors should improve the representation here, specifically, how to map between them and more familiar coordinate systems. I also didn't understand what an oriented-vector is. Aren't all vectors oriented? This specific section on orientation especially needs improvement.
l454: Why use the COM of one of the elements for the tangent plane and not somewhere in the middle? Would seem to result in larger mapping errors for the other element. Is it just for computational efficiency and doesn't matter too much?
l496: Does the distortion of elements potentially result in preferred directions? Or is the effect quite small?
Minor corrections:
l28: extra space after "Feltham, 2011)"
l34: commas around "however"
l337: inn -> in
eqn 42: two commas
l467, 549: Don't start a sentence with "So, "