Development and Application of Coupled Atmospheric Chemistry Models

Abstract

Comprehensive and accurate representation of atmospheric chemistry in Earth System Models is a major priority in geoscientific modeling. GEOS-Chem is a state-of-the-science atmospheric chemistry model used by hundreds of research groups worldwide. It has been restructured to operate as a chemical module in weather and climate models while sharing the same scientific codebase as the standalone chemical transport model. The Community Earth System Model version 2 (CESM2) is a state-of-science Earth System Model that has participated in multiple model intercomparison activities. CESM2 includes a comprehensive chemistry option, CAM-chem, with a different development heritage from GEOS-Chem. The two models show large differences in model skill in comparison to observations in different regions, implying differences in sensitivity to perturbations. Previous model intercomparisons generally compared entire modeling systems, introducing first-order differences complicating the attribution of particular strengths in reproducing observations to specific representations of processes in each model. Implementation of GEOS-Chem within CESM2 as an alternative chemistry option to CAM-chem allows for detailed process-based intercomparison between two models, informing more accurate representation of atmospheric chemistry, and contributing to the vision of a modular chemistry-climate model. In this work, we develop a multi-model emissions and data tool through an updated version of the Harmonized Emissions Component (HEMCO) and couple it to the CESM2 model. This serves as the foundation for implementation of GEOS-Chem within CESM2 and intercomparison with CAM-chem (Chapter 1). Because of the increasing chemical complexity in atmospheric chemistry models resulting in increased computational burden, which is particularly relevant in coupled model environments, we develop an adaptive, auto-reduction solver for chemical kinetics that can reduce the complexity of the chemical mechanism depending on local conditions. It can provide a ∼30% speed-up in the chemical solver while introducing ∼1% error in the troposphere. We evaluate the adaptive solver in the offline GEOS-Chem chemical transport model at a 2◦ × 2.5◦ resolution, but it has been implemented in the Kinetic Pre-Processor (KPP) software package making it easily extensible beyond GEOS-Chem (Chapter 2). We then apply the coupled chemistry-climate model GEOS-Chem within CESM2 to compare its representation of tropospheric oxidant chemistry against the default CAM-chem chemistry driven by the same dynamics, physics, and emissions. We use ozonesondes and aircraft observations from the ATom-1 and KORUS-AQ campaigns to identify and attribute major differences between the two models. (Chapter 3).