Efficient testing strategies for modernizing the 4-mode Modal Aerosol Model (MAM4) for global kilometer-scale aerosol simulations

PRESENTERS:
To attach your poster or presentation:
E-mail your file for upload

Authors

Lead Presenter

Balwinder Singh - Atmospheric, Climate, & Earth Sciences Division Pacific Northwest National Laboratory Richland WA USA

Co-Author

Oscar Diaz-Ibarra - Center for Computing Research, Sandia National Laboratories, Albuquerque, NM, USA

Jeffrey Johnson - Cohere Consulting LLC

Jerome Fast - Atmospheric, Climate, & Earth Sciences Division Pacific Northwest National Laboratory Richland WA USA

James Overfelt - Center for Computing Research, Sandia National Laboratories, Albuquerque, NM, USA

Michael Schmidt - Center for Computing Research, Sandia National Laboratories, Albuquerque, NM, USA

Brian Gaudet - Atmospheric, Climate, & Earth Sciences Division Pacific Northwest National Laboratory Richland WA USA

Meng Huang - Atmospheric, Climate, & Earth Sciences Division Pacific Northwest National Laboratory Richland WA USA

Chandru Dhandapani - Atmospheric, Climate, & Earth Sciences Division Pacific Northwest National Laboratory Richland WA USA

Kai Zhang - Atmospheric, Climate, & Earth Sciences Division Pacific Northwest National Laboratory Richland WA USA

Mingxuan Wu - Atmospheric, Climate, & Earth Sciences Division Pacific Northwest National Laboratory Richland WA USA

Kyle Pressel - Atmospheric, Climate, & Earth Sciences Division Pacific Northwest National Laboratory Richland WA USA

Jaelyn Litzinger - Pacific Northwest National Laboratory, Richland, Washington

Cameron Rutherford - Pacific Northwest National Laboratory, Richland, Washington

Hui Wan - Atmospheric, Climate, & Earth Sciences Division Pacific Northwest National Laboratory Richland WA USA

Peter Bosler - Center for Computing Research, Sandia National Laboratories, Albuquerque, NM, USA

Po-Lun Ma - Atmospheric, Climate, & Earth Sciences Division Pacific Northwest National Laboratory Richland WA USA

Shuaiqi Tang - State Key Joint Laboratory of Environmental Simulation and Pollution Control, School of Environment, Tsinghua University, Beijing, China.

Abstract

The EAGLES (Enabling Aerosol-cloud interactions at GLobal convection-permitting scalES) project is focused on advancing the understanding and predictability of aerosol and aerosol-cloud interactions in the Earth system. To this end, EAGLES is modernizing the 4-mode Modal Aerosol Module (MAM4) software infrastructure within E3SMv4 (version 4 of Energy Exascale Earth System Model) to perform global kilometer-scale simulations by harnessing the computing powers of the exascale machines with GPUs.

Software engineers collaborating with domain knowledge experts have completely rewritten the MAM4 aerosol package. This modernization effort has led to the development of the MAM4xx library, written in C++ using the Kokkos performance portable library. The team employed efficient workflows and robust testing strategies to increase the trustworthiness of the ported code. The key emphasis was on the use of end-to-end testing techniques during the Fortran refactoring, porting, and EAMxx (E3SMv4’s Atmospheric Model) integration to ensure correctness. The aerosol sub-processes are protected with multiple unit tests during code refactoring, translation, and E3SMv4 integration. The aerosol processes that are integrated into the EAMxx framework can simulate aerosols at global kilometer-scale resolution using GPUs.

In this presentation, I will present the efficient workflows and testing strategies that we employed to successfully prepare the Fortran codes to port to the C++/Kokkos framework. Additionally, I will share recent results from global simulations using MAM4xx within its host model E3SMv4.

Related Meeting/Event

EESM PI Meeting 2024

Presentations

Lead Presenter

Co-Author