Development of High-Order DDM Sensitivity Analysis for Particulate Matter in Multidimensinal Air Quality Models

WENXIAN ZHANG (1), Shannon Capps (2), Athanasios Nenes (3), Armistead Russell (4)

(1) (2) (3) (4) Georgia Institute of Technology

     Abstract Number: 413
     Last modified: May 12, 2010

Abstract
The decoupled direct method (DDM) sensitivity analysis is an approach to efficiently quantify the impacts of model inputs and parameters on simulated pollutant concentrations. For example, DDM can be used to conduct emission source impact analysis. In the past decade, this approach has been applied to evaluate emission control strategies and has been implemented in air quality models to compute first- and second-order gas sensitivities as well as first-order aerosol sensitivities. The formation of secondary aerosol undergoes complicated chemical and physical processes (e.g., thermodynamic equilibrium reactions; aqueous-phase oxidation of precursor gases), so the responses of aerosol concentrations to source emissions cannot be completely captured by first order sensitivity. In order to improve the prediction of model responses of particulate matter (PM) to emissions, this work has implemented high-order DDM sensitivity analysis of particulate matter (HO-DDM/PM) in the Community Multiscale Air Quality (CMAQ) model.

This new implementation has been tested using a week-long winter episode over the US. Results compare well with the traditional brute force sensitivities, and DDM performs better in locations where the brute force results are demonstrably inaccurate. The magnitude of second-order PM sensitivities typically are small compared to first-order ones, but do elucidate the non-linearities of the system. Both sulfate and nitrate aerosols show noticeable non-linear responses to their precursors (e.g., sulfate to NOx, nitrate to SOx). A detailed source apportionment of PM is performed based on HO-DDM/PM.