MPASdiag: Data Processing, Visualization and Analysis Toolkit for MPAS from Native Unstructured Mesh

Model for Prediction Across Scales (MPAS) takes advantage of unstructured meshes and variable-resolution grids to provide accurate and efficient simulations, however, datasets from native unstructured meshes can be challenging to analyze and visualize using traditional tools designed for structured grids. In this talk, we present MPASdiag, an open-source Python package developed to facilitate the processing, analysis and visualization of MPAS model output from native unstructured meshes. Besides a powerful data processing engine powered by xarray and uxarray, MPASdiag provides an advanced visualization framework built on top of Matplotlib and Cartopy. MPASdiag visualization capabilities currently include the generation of 2-D surface plots, vertical cross sections, zonal and meridional averages, Hovmöller diagrams, and time series plots. The visualization framework is designed to be highly customizable, allowing users to tailor plots to their specific needs. To demonstrate the capabilities of MPASdiag, we will showcase several case studies including the visualization of 2-D and 3-D atmospheric variables from high-resolution MPAS simulations, batch processing of cycling MPAS forecast data to demonstrate the efficiency of command line interface, and the scalability of parallel processing engine to handle kilometer-scale MPAS data analysis from native unstructured meshes.