How can we identify extreme weather features in next-generation global climate models?
Next-generation global general circulation models (GCMs) can exploit massively parallel computing systems to perform simulations with grid spacings of 30km horizontal resolution and finer. At these resolutions, GCMs can adequately resolve extreme weather, such as tropical cyclones and mesoscale convective systems. To quantify these features in GCM data, algorithmic detection approaches are needed since there is no pointwise observational dataset that can be used to extract discrete storms. However, new high-resolution GCMs utilize unstructured grids and produce large amounts of data, necessitating new tracking techniques from both a scientific and software engineering perspective.
This presentation will discuss recent work regarding the development of a new Lagrangian software package that can be used to detect and track extreme weather features in high-resolution gridded datasets. A sensitivity analysis framework commonly applied in engineering applications will be tested to demonstrate potential methods for assessing uncertainty in tracked extreme storms statistics. Finally, pointwise feature tracking is applied to new 14km variable-resolution datasets using both the Spectral Element (SE) and Model for Prediction Across Scales (MPAS) dycores in the Community Earth System Model (CESM2). Examples of model-simulated severe weather are highlighted.