Student Seminar Series
Department of Atmospheric & Oceanic Sciences
a talk by
Spatial Verification of Sea Ice Edges Using Distance-Based Metrics
Sea ice models are used to forecast sea ice edges, which are useful for determining safe shipping routes. Traditional verification of sea ice models have typically relied on grid-point to grid-point comparisons of observations against forecasts, which ignores spatial structure and leads to a double-penalty in assessing forecast skill. Ice edge forecasts are well suited for spatial verification, as interest is in where the forecast was wrong, and by how far. Common spatial verification methods for sea ice edges include the Integrated Ice Edge Error (IIEE) and various distance-based metrics. For a distance-based metric to be considered a true metric, it must fulfill four criteria: positivity, separation, symmetry, and the triangle inequality. Common distance metrics currently used by the sea ice community fail to fulfill the symmetry principle. In this talk, I will present some commonly used sea ice edge spatial verification methods, explain how they are not symmetric, and conclude by proposing a new method that is symmetric.
Wednesday Nov 06/ 2.30 PM/ Room 934 Burnside Hall