Meteorological Tools for Assessing Mantle Flow-related Dynamic Topography Maps

Construction of robust mantle flow trajectories plays an important role in understanding the parameters that govern mantle convection and relating them to geologic observables. In the past, the assessment of constructed trajectories focused majorly on metrics targeted at the mantle volume whilst neglecting the surface manifestations of mantle convection in the form of dynamic topography. However, an increasing amount of interest is being built around linking convection to surface effects, including dynamic topography. As such, it is vital to study ways in which mantle flow trajectories can be assessed via their dynamic topography predictions. Commonly used assessment and comparison metrics such as root-mean-square errors, suffer from the so-called double penalty problem --- a dynamic topography prediction that does not match a reference observation one-to-one is penalized twice: first as a miss and second as a false alarm. It is therefore attractive to investigate metrics that overcome this problem. Here, we introduce an object-based approach, first applied in meteorology, and show that this approach is not only amenable to studying dynamic topography, but that it also overcomes the double penalty problem whilst providing accurate model assessment.