ISeeSnow v1.0 – a pilot study for snow avalanche model intercomparison of thickness-integrated shallow flow approaches and beyond

Abstract. We present a pilot study for intercomparison of snow avalanche flow simulation tools. Thirteen different groups participated in the study, whose models are categorized into a core group using thickness/depth-integrated shallow flow approaches, and an extended group including one with a differing shallow water approach, and one depth-resolved 3D approach. The intercomparison is performed for three simple test cases representative of typical applications: snow avalanche flow over an idealized and a real topography with a release area of constant thickness based on a Voellmy and pure Coulomb friction relation and prescribed values for the friction parameters. The aim of this pilot study is to analyse the spread in simulation results and discuss potential sources of the observed differences. A quantitative assessment of the variability is based on the distribution of scalar measures like runout length, runout angle and maximum values of flow thickness and flow velocity. Within the core group of thickness/depth-integrated shallow flow approaches, simulation results for the Voellmy test cases (idealized and real topography), excluding outliers, show a spread of roughly 55 m in derived runout length with an interquartile range of about 30/33 m, referring to total runout lengths of 2310/2082 m (median values for idealized and real topography). Runout length is for all three test cases constrained by an abrupt change in slope angle. Maximum peak flow thickness and velocity show a generally larger spread for both, the idealized and real topography test cases. Excluding outliers, differences between the simulation results are more pronounced for the real topography test case compared to the idealized topography case. The largest differences arise if only Coulomb friction is considered, as indicated by an interquartile range of 92 m in runout length. The Coulomb test case also shows a considerably larger number of outliers. This is partly due to the pronounced effect of curvature effects in this case, which is not accounted for by all participating simulations tools. Focusing on the core group, this analysis serves as a first assessment of the uncertainty introduced by the different implementation workflows (e.g., numerical schemes, ad-hoc treatments, geo-data handling, curvature treatment, etc.). However, actual attribution of variability to individual sources is beyond the scope of this preliminary study and will necessitate further testing. Performing this pilot study allowed us to identify common issues, gather information on respective requirements regarding input data and problem definition, which will help to optimize the design of a future, more comprehensive model intercomparison study.