Current pre-exascale computing systems, and the strong push towards exascale warrant substantial efforts to improve the geoscientific software infrastructure used for Earth System Model (ESM) development, data analysis, and storage. The Exascale era opens a range of opportunities, including increased domain size, simulation duration, model resolution, large ensembles, and new physics. This session will discuss challenges and solutions involving domain scientists, applied mathematicians, computer scientists, HPC, and compression experts.
Contributions address challenges and advances to achieve exascale-readiness geoscience disciplines, methods, and technologies.
Pangeo (pangeo.io) is a community of researchers and developers that tackle these issues in a collaborative manner using a growing Python ecosystem whose core tools include xarray, Iris, DASK, Jupyter, Zarr and INTAKE. Many contributors to this session will share novel tools within the Pangeo ecosystem devoted to Atmosphere, Ocean and Land Models, Satellite Observations, HPC, Cloud computing, Machine Learning, and Scalable scientific computing.
This session also considers how geoscientists can shift towards greener computing by adopting modern data compression techniques including, though not limited to: algorithmic advances, assessments of data storage sustainability, compression efficiency and speed in software and/or hardware, interoperability issues, remote sensing applications, and support in widely used languages (e.g., C/C++, Fortran, Java, Python), data storage formats (e.g., HDF, netCDF, Zarr), and Open Source workflows (e.g., CDO, NCO, Pangeo, Ruby, Xarray).
All authors in this session have the option to submit Jupyter notebooks of their work; the best five will be selected as part of the Pangeo applications gallery of EGU22. Examples of previous galleries are at http://gallery.pangeo.io.