Скачать или смотреть Enabling Petabyte-scale Ocean Data Analytics- Thomas Nicholas, Julius Busecke | SciPy 2022

The largest oceanographic simulations available are petabyte-scale, presenting a huge challenge when computing even relatively simple analysis functions. Furthermore, doing vector calculus with data living on a staggered grid requires specific numerical schemes to guarantee correctness.

XGCM solves both these problems - it applies common vector calculus operations in a grid-aware manner, and now uses the minimum number of dask tasks. This works by applying operations atomically as a “grid ufunc”, using a protocol which generalises numpy ufuncs to specify the input and output grids for a given computation. Using this approach in conjunction with another package, xhistogram, we computed vorticity-strain histograms in different ocean basins globally.