A33D-0189
A reliable light scattering computing for black carbon-containing particles: Hybrid discrete dipole approximation (h-DDA)

Wednesday, 16 December 2015
Poster Hall (Moscone South)
Nobuhiro Moteki, University of Tokyo, Bunkyo-ku, Japan
Abstract:
Black carbon (BC) is a light-absorbing carbonaceous aerosol emitted from combustions of fossil fuels and biomasses and is estimated as the second most important contributor to positive climate forcing after the carbon dioxide. In the atmosphere, the fractal aggregate of BC-spherules may be mixed with non-absorbing (or weakly absorbing) compounds that forms morphologically complex “BC-containing particle”. A reliable scattering code for BC-containing particles is necessary for predicting mass absorption efficiency of BC and designing/evaluating optical techniques for estimating microphysical properties (i.e., size distribution, mixing state, shape, refractive index) of BC-containing particles. The computational methods that derived from the volume-integral form of the Maxwell equation, such as discrete dipole approximation (DDA), are method of choice for morphologically complex object like BC-containing particles. In ordinary DDA, the entire particle volume is approximated as a collection of tiny cubical dipoles (with side length d) placed on a 3D cubic lattice. For several model BC-containing particles, the comparisons with numerically exact T-matrix method reveals that the ordinary DDA suffered from persistent positive systematic error (up to +30%) in absorption even under d <<λ. The cause of this DDA error is identified to be the shape error in BC-spherules. To eliminate the shape error in BC-spherules, we propose a new DDA methodology which may be called hybrid DDA (h-DDA): each primary BC sphere is assumed as a spherical dipole, while remaining particle volume of coating material is approximated by a collection of tiny cubical dipoles on a 3D cubic lattice. Positive absorption bias up to +30% in ordinary DDA is suppressed to within 3% in h-DDA. In h-DDA code, an efficient FFT-based algorithm for solving the matrix equation has been implemented, by utilizing the multilevel block-Toeplitz property of the submatrix corresponding to inter-dipole interaction within coating material.