Abstract View
Integrating Aerosol Size Distribution Measurements with a 3D Chemical Transport Model
DANA MCGUFFIN, Erik B. Ydstie, Peter Adams, Carnegie Mellon University
Abstract Number: 390
Working Group: Remote and Regional Atmospheric Aerosol
Abstract
Accurately modeling aerosol dynamics is required to obtain an understanding of the ambient size distribution, cloud condensation nuclei (CCN) activity, and therefore aerosol indirect effects. Any uncertain processes or uncertain model inputs will lead to uncertainty in the predicted concentration fields. Key uncertainties in predicting CCN concentration fields include formation of primary particles due to aerosol emissions and the nucleation of condensable vapors as well as formation of secondary particles due to the condensation of volatile organic compounds.
The goal of this work is to improve 3D chemical transport model (CTM) predictions by constraining several uncertain processes with a network of ground-based size distribution measurements. The CTM utilized here is GEOS-Chem TOMAS driven by GEOS meteorological fields. We aim to constrain the primary organic aerosol (POA) emissions, nucleation rate, and secondary organic aerosol (SOA) production rates over this region to improve predicted concentration of CCN.
We use a novel inverse method previously developed for a box model. This method transforms the full size distribution into three aerosol properties, each of which are sensitive to one of the uncertain process rates we aim to constrain. In this work, we distribute the inversion technique among each grid block in the 3D CTM that contains a measurement station. We utilize measurements from 13 stations in Western Europe, of which seven stations are used as a training dataset to constrain the three uncertain processes and six stations are used as a testing dataset.
Applying the inverse method reduces bias in the training dataset for the three aerosol properties. For particle number concentration between 3 and 6 nm, the model-measurement bias decreases by over a factor of two and the correlation improves. The constrained nucleation rate is, on average, scaled down during local noon from the rate predicted by GEOS-Chem TOMAS.