Amore 2.0: A New and Improved Algorithm for the Reduction of Atmospheric Oxidation Mechanisms
FORWOOD WISER, V. Faye McNeill, Siddhartha Sen, Daniel Westervelt, Benjamin Yang, Arlene Fiore, Daven Henze, Arijit Chakraborty,
Columbia University Abstract Number: 416
Working Group: Aerosol Chemistry
AbstractAtmospheric chemical models are limited in size due to the computational constraints of atmospheric simulations. However, through experimental and computational methods, a large number of high fidelity large mechanisms have been developed for the oxidation of atmospheric organics. These mechanisms often have limited application in atmospheric simulations due to their size. To address this problem, we have developed a set of algorithms for the reduction of atmospheric oxidation mechanisms. Our first algorithm, AMORE 1.0 was used to create a highly accurate reduced isoprene oxidation mechanism with 12 species from a starting point of 404 species. This approach required manual input to finalize the mechanism, and was only able to create very small mechanisms. Here we present AMORE 2.0, a new and improved algorithm for the high fidelity reduction of atmospheric oxidation mechanisms. The AMORE 2.0 algorithm is rooted in graph theory, and produces accurate mechanisms that are far smaller than the input full mechanisms. In contrast to the AMORE 1.0 algorithm, which selected optimal reduced pathways for the reduced mechanism, the AMORE 2.0 algorithm ranks species in order of their importance to the mechanism, and removes them in that order. Unlike other methodologies, the reactions involving removed species are rerouted so that species removal has minimal impact on other parts of the mechanism. This algorithm requires no manual adjustments, and reduces mechanisms in a stepwise fashion, allowing users to select the desired mechanism size, with increased accuracy for larger mechanisms. In addition, as a rules-based algorithm, the results are highly interpretable. Examples are given for the isoprene, furans, and alpha pinene mechanisms.