Aerosol optical depth assimilation for a size-resolved sectional model: impacts of observationally constrained, multi-wavelength and fine mode retrievals on regional scale analyses and forecasts

- Indicates paper has been withdrawn from meeting
- Indicates an Award Winner
Wednesday, 5 February 2014: 9:30 AM
Room C206 (The Georgia World Congress Center )
Pablo E. Saide, University of Iowa, Iowa City, IA; and G. Carmichael, Z. Liu, C. S. Schwartz, H. C. Lin, A. Da Silva, and E. J. Hyer

An aerosol optical depth (AOD) three-dimensional variational data assimilation technique is developed for the Gridpoint Statistical Interpolation (GSI) system when WRF-Chem forecasts are performed with a detailed sectional model (MOSAIC). Within GSI, forward AOD and adjoint sensitivities are performed using Mie computations from the WRF-Chem optical properties module providing consistency with the forecast. GSI tools such as recursive filters and weak constraints are used to provide correlation within aerosol size bins and upper and lower bounds for the optimization. The system is used to perform assimilation experiments with fine vertical structure and no data thinning or re-gridding on a 12-km horizontal grid over the region of California, USA. A first set of simulations is performed comparing the assimilation impacts of operational MODIS dark target retrievals to observationally constrained ones (i.e. calibrated with AERONET data), the latter ones showing higher error reductions and increased fraction of improved PM2.5 and AOD ground-based monitors. A second set of experiments reveals that the use of fine mode fraction AOD and ocean multi-wavelength retrievals can improve the representation of the aerosol size distribution, while assimilating only 550nm AOD retrievals produces no or at times degraded impact. A demonstration of these tools will also be presented for operational forecast for flight planning of the SEAC4RS campaign. Figure attached: Left panels: May 2010 average maps of operational MODIS Terra (top-left), NASA-NNR (top-middle) product for the same MODIS Terra data, non-assimilated model (bottom-left) and assimilated using NASA NNR (bottom-middle). Monthly averages of the raw and observationally constrained retrievals can be quite different, and assimilation brings model estimates closer to the observation. Right panels: 550-870nm Angstrom exponent fractional error reductions from non-assimilated to assimilated model computed with respect to Aqua retrievals. Figure on top assimilates only MODIS 550nm AOD while the one on the bottom assimilates MODIS 550, 660, 870, and 1240nm over ocean and only 550nm over land. Higher error reductions on angstrom exponent when using multi-wavelength data point towards improvements in aerosol size distributions.