Sensitivity metric approach for retrieval of aerosol properties from multiangular and multispectral polarized radiances

摘要

Linearly polarized radiation is sensitive to the microphysical properties of aerosols, namely, to the particle-size distribution and refractive index. The discriminating power of polarized radiation increases strongly with the increasing range of scattering angles and the addition of multiple wavelengths. The polarization and directionality of the Earth's reflectances (POLDER) missions demonstrate that some aerosol properties can be successfully derived from spaceborne polarimetric, multiangular measurements at two visible wavelengths. We extend the concept to analyze the retrieval capabilities of a spaceborne instrument with six polarimetric channels at 412, 445, 555, 865, 1250, and 2250 nm, measuring approximately 100 scattering angles covering a range between 50 and 150 deg. Our focus is development of an analysis methodology that can help quantify the benefits of such multiangular and multispectral polarimetric measurements. To that goal we employ a sensitivity metric approach in a framework of the principal-component analysis. The radiances and noise used to construct the sensitivity metric are calculated with the realistic solar flux for representative orbital viewing geometries, accounting for surface reflection from the ground, and statistical and calibration errors of a notional instrument. Spherical aerosol particles covering a range of representative microphysical properties (effective radius, effective variance, real and imaginary parts of the refractive index, single-scattering albedo) are considered in the calculations. We find that there is a limiting threshold for the effective size (approximately 0.7 (mu)m), below which the weak scattering intensity results in a decreased signal-to-noise ratio and minimal polarization sensitivity, precluding reliable aerosol retrievals. For such small particles, close to the Rayleigh scattering limit, the total intensity provides a much stronger aerosol signature than the linear polarization, inspiring retrieval when the combined signals of intensities and the polarization fraction are used. We also find a strong correlation between aerosol parameters, in particular between the effective size and the variance, which forces one to simultaneously retrieve at least these two parameters.