Publications

Rubin, J. I.; Collins, W. D. (2014). Global simulations of aerosol amount and size using MODIS observations assimilated with an Ensemble Kalman Filter. JOURNAL OF GEOPHYSICAL RESEARCH-ATMOSPHERES, 119(22), 12780-12806.

Abstract
A global assimilation that uses an Ensemble Kalman Filter and a set of derived scaling equations is presented for jointly adjusting the amount of atmospheric aerosol and the relative contribution of fine and coarse aerosols. The assimilation uses Department of Energy and National Science Foundation's Community Atmosphere Model (CAM) model and aerosol optical depth (AOD) and Angstrom exponent (AE) retrievals from NASA's Moderate Resolution Imaging Spectroradiometer (MODIS) instrument. Aerosol Robotic Network (AERONET) AE retrievals are used to constrain size over land. The presented system includes 60 ensemble members with a daily analysis, incorporating daily-averaged retrievals. A CAM control simulation and a CAM experiment with data assimilation (CAM-DA) are run for the year 2007. Control run comparisons to MODIS observations reveal a persistent negative bias in AOD, indicating an underprediction of the amount of atmospheric aerosol (CAM: 0.09 (+/- 0.06), MODIS:0.16 (+/- 0.09)). The negative bias decreased in the assimilation run with a globally averaged AOD of 0.12 (+/- 0.05). CAM-DA is able to better capture spatial and temporal variations. A comparison of regional time series reveals the greatest reduction in model bias with respect to both aerosol amount and size over the oceans, especially the Southern Ocean. With respect to land regions, good agreement with AERONET AOD is found over the United States, Europe, and East Asia. Additionally, CAM-DA has clear spatial differences from the control with more aerosol and a larger fine contribution in the Northern Hemisphere. The results also demonstrate the utility in assimilation methodologies for identifying systematic model biases, using the data assimilation correction fields as an indicator.

DOI:
10.1002/2014JD021627

ISSN:
2169-897X