Development and Performance of the Modularized, High-performance Computing and Hybrid-architecture Capable GEOS-Chem Chemical Transport Model

Wednesday, 17 December 2014
Michael S Long1, Robert Yantosca2, Jon Nielsen3, John C Linford4, Christoph Andrea Keller1, Melissa Payer Sulprizio2 and Daniel J. Jacob1, (1)Harvard University, School of Engineering and Applied Sciences, Cambridge, MA, United States, (2)Harvard University, Cambridge, MA, United States, (3)Global Modeling and Assimilation Office, Greenbelt, MD, United States, (4)ParaTools, Inc., Baltimore, MD, United States
The GEOS-Chem global chemical transport model (CTM), used by a large atmospheric chemistry research community, has been reengineered to serve as a platform for a range of computational atmospheric chemistry science foci and applications. Development included modularization for coupling to general circulation and Earth system models (ESMs) and the adoption of co-processor capable atmospheric chemistry solvers. This was done using an Earth System Modeling Framework (ESMF) interface that operates independently of GEOS-Chem scientific code to permit seamless transition from the GEOS-Chem stand-alone serial CTM to deployment as a coupled ESM module. In this manner, the continual stream of updates contributed by the CTM user community is automatically available for broader applications, which remain state-of-science and directly referenceable to the latest version of the standard GEOS-Chem CTM. These developments are now available as part of the standard version of the GEOS-Chem CTM. The system has been implemented as an atmospheric chemistry module within the NASA GEOS-5 ESM. The coupled GEOS-5/GEOS-Chem system was tested for weak and strong scalability and performance with a tropospheric oxidant-aerosol simulation. Results confirm that the GEOS-Chem chemical operator scales efficiently for any number of processes. Although inclusion of atmospheric chemistry in ESMs is computationally expensive, the excellent scalability of the chemical operator means that the relative cost goes down with increasing number of processes, making fine-scale resolution simulations possible.