Satellite and Correlative Measurements of the Stratospheric Aerosol. III: Comparison of Measurements by SAM II, SAGE, Dustsondes, Filters, Impactors and Lidar

Publisher Website
Google Scholar
Abstract
A large satellite validation experiment was conducted at Poker Flat, Alaska, 16–19 July 1979. Instruments included the SAM II and SAGE satellite sensors, dustsondes impactors, a fitter collector and an airborne lidar. We show that the extinction profiles that were measured independently by SAM II and SAGE agree with each other. We then use a generalized optical model (which agrees with the Poker Flat optical absorption and relative size distribution measurements) to derive extinction profiles from the other measurements. Extinction profiles thus derived from the dustsonde, fitter and lidar measurements agree with the satellite-measured extinction profiles to within the combined uncertainties. (Individual 1 σ uncertainties are, at most heights, roughly 7 to 20% each for the satellite, dustsonde and filter measurements, 30 to 60% for the lidar measurements, and 10 to 20% for the process of converting one measured parameter to another using the optical model.) The wire impactor-derived results are also consistent with the other results, but the comparison is coarse because of the relatively large uncertainties (±35% to a factor of 4) in impactor-derived mass, extinction, N0.15 and N0.25 (Nx is the number of particles per unit volume with radius greater than x μm.) These uncertainties apply to background stratospheric aerosol size distributions, and result primarily from relatively small uncertainties (±8 to ±20% for confidence limits of 95%) in radii assigned to impacted particles, combined with the steepness of background size distributions in the radius range that contributes most to mass, extinction, N0.15 and N0.25. Polar nephelometer-measured asymmetry parameters (0.4 to 0.6) agree with a previous balloon photometer inference, but are significantly less than the value (∼0.7) obtained from Mie scattering calculations assuming either model or measured size distributions.