AAAR 36th Annual Conference October 16 - October 20, 2017 Raleigh Convention Center Raleigh, North Carolina, USA
Abstract View
A Stability-based Inverse Model Applied to the General Dynamic Equation
DANA MCGUFFIN, Peter Adams, Erik B. Ydstie, Carnegie Mellon University
Abstract Number: 349 Working Group: Regional and Global Air Quality and Climate Modeling
Abstract The goal of this work is to use measured aerosol size distributions to constrain atmospheric box model simulations of aerosol microphysics based on the general dynamic equation. We use a zero-dimensional version of the TOMAS microphysical code to model size-resolved aerosols as they undergo emission, nucleation, condensation, coagulation, and deposition to the surface. Such measurements of aerosol mass and number distributions are available at several ground sites. Measured size distributions implicitly contain a great deal of information about microphysical processes such as nucleation, primary emissions, and condensational growth rates and are generally used to validate models. This is usually done through sensitivity analyses in forward models in which tuning factors or model parameters are systematically adjusted until the model reproduces a measured distribution of aerosols. However, a more robust way to do this is to generate and run an inverse model. In general, inverse models are used to constrain uncertain processes based on a set of measurements. Inverse models usually minimize a cost function related to the least-squares error by linearizing the forward model. However, inverse modeling of aerosol microphysics is challenging since it involves nonlinearities that span large spatial and temporal scales. This work develops a new, stability-based inverse modeling technique that is not computationally intensive and rigorously accounts for nonlinear dynamics.
Here, we present the development of the inverse modeling technique and preliminary applications to representative aerosol size distributions. Parameters tuned by the inverse model include those that are highly uncertain in atmospheric applications: nucleation rates, growth rates due to organic condensation, and primary emissions. The methodology developed here will facilitate future work that estimates uncertain parameters in a global model based on a measurement network.