David Lary (PI)
University of Maryland Baltimore County
dlary@gmao.gsfc.nasa.gov

The Swift Neurological Acceleration of Atmospheric Photochemistry and Aerosol Calculations [SNAPA]: Development and Application to Atmospheric Chemistry and Aerosol Simulations With a Focus on UT/LS Ozone

An adequate photochemical mechanism to describe the evolution of ozone in the upper troposphere and lower stratosphere (UT/LS) in a computational model involves reactive nitrogen, hydrogen, halogens, hydrocarbons, and interactions with aerosols.  This complex mechanism is computationally expensive, and applications are limited by the computational burden.  Simulations are made tractable by using coarse horizontal resolution (4º latitude x 5º longitude or greater) or by greatly reducing the interactions accounted for in the photochemical mechanism, but these compromises limit the scientific applications. Neural networks, multi-variate nonlinear machine learning algorithms, offer a means to obtain a fast, accurate solution to the stiff ordinary differential equations that comprise the photochemical calculations, thus making high resolution simulations including the complete photochemical mechanism possible.  This proposal will implement advanced neural networks to speed up the photochemical solution and also two other key elements of the Global Modeling Initiative (GMI) Chemistry and Transport Model (CTM), the photolysis calculation and aerosol microphysical calculations.  In addition to demonstrating that the neural networks provide improved computational efficiency without sacrificing accuracy, we will exploit the increased computational efficiency to examine the ozone budget in the upper troposphere and lower stratosphere (500-50 mb).

+ Back to Participant Listing