Modelling the 3-D evolution of glaciers at regional to global scales: challenges and opportunities

<p>Various techniques exist to model the evolution from glaciers at regional to global scales. Whereas pioneering efforts typically relied on volume-area scaling approximations or parameterizations based on observed glacier changes (retreat parameterization), more recent approaches now also explicitly incorporate ice-dynamical processes. In these latter studies, glaciers are typically represented through central flowlines. Such flowline approaches are particularly suited for mountain glaciers that span over a large elevation range, i.e. valley-glaciers with an elongated shape. However, flowline approaches are not ideal to represent the geometry of ice caps (large glaciers) that generally have a dome-shaped geometry. For ice caps, a model representation that explicitly accounts for the glacier&#8217;s 3D geometry and that allows for the glacier to lose and gain mass in all directions, both through mass balance and ice dynamic processes, is needed.</p><p>Here we present simulations performed with a coupled surface mass balance &#8211; ice flow model that explicitly accounts for the 3D geometry of individual glaciers. The model, written in Python, relies on the shallow ice approximation to describe ice flow, allowing to run large ensembles of simulations. The goal is to simulate the temporal evolution of glaciers with distinct shapes and situated in various climatic regimes, i.e. having a model that allows for an automated intialization and that is suited for regional to global-scale applications.</p><p>In this contribution, we present simulations performed with this new large-scale model for regions with mountain glaciers (e.g. European Alps and Scandinavia), as well as regions with large ice caps (e.g. Iceland). Through this, we highlight various challenges that relate to model initialization or the choice of model settings, for instance. We also explore how simulated glacier evolutions compare to those simulated with a retreat parameterzation and through flowline modelling, thereby shedding light on the need for a 3D modelling approach.</p>