Ozone and aerosol optical properties from ground based ultra-violet irradiance measurements
Date
2006
Authors
Slusser, James, author
Taylor, Thomas E., author
Stephens, Graeme L., 1952-, author
Department of Atmospheric Science, Colorado State University, publisher
Journal Title
Journal ISSN
Volume Title
Abstract
Error analysis and characterization of an optimal estimation retrieval algorithm are described. The algorithm, which was initially developed by Goering et al., estimates values of aerosol optical properties (aerosol optical depths, AOD, and aerosol single scattering albedo, SSA) at 7 wavelengths in the ultraviolet (UV) spectral region as well as total column ozone (TOC) Goering et al. (2005). The measurements used in the retrieval algorithm are obtained from a UV Multi-Filter Rotating Shadow-Band Radiometer (UVMFRSR), which measures diffuse and direct irradiances at 300-, 305-, 311-, 317-, 325-, 332- and 368-nm Bigelow et al. (1998). These radiometers are deployed as part of the United States Department of Agriculture's (USDA) Ultra-Violet Monitoring and Research Program (UVMRP) network of 33 surface stations distributed around the USA. The Tropospheric Ultraviolet/Visible (TUV) radiative transfer model Madronich (1993) is employed as the forward model in the retrieval algorithm. Several improvements to the original retrieval algorithm were developed. The wavelength independent asymmetry factor, ɡ, was added as an explicit retrieval parameter and a priori error covariances were incorporated as a way of adding information to the retrieval. More careful evaluation of the retrieval was performed, including sensitivity studies of the radiative transfer solvers used by the forward model. Model sensitivities to TOC, ɡ and surface albedo over domains of AOD and SSA were used to create a realistic model error budget. Realistic wavelength-dependent surface albedos were incorporated into the model using analytical expressions created from measurements Doda and Green (1981). The new algorithm also uses a more rigorous UV-MFRSR measurement error budget determined from previous research Krotkov et al. (2005) as well as specified model error derived from the sensitivity analysis. The new version also benefits from several modifications that significantly speed up the retrieval algorithm, making it potentially valuable for routine operational use. Synthetic testing of the retrieval algorithm for various atmospheric conditions was performed to determine the conditions under which the algorithm produced accurate and statistically significant results. Retrieval results were found to improve with increasing atmospheric turbidity (AOD) and with increasing particle scattering properties (SSA and ɡ). The sensitivity of the results to solar zenith angle was determined to be minimal. Solution spaces of the retrieval in the AOD-SSA domain space were studied to show that the retrieval has a uni-modal solution with respect to these two parameters and that the model has sufficient sensitivity and resolution to these parameters to make reasonable estimates of their values. The optimal estimation retrieval was performed on a cloud screened data set from May, 2003 at the Panther Junction station in Big Bend National Park, Texas. The monthly time series of AOD, SSA, TOC and ɡ were filtered using the chi-squared value at the 95% significance level before the estimated errors and diagnostics were analyzed to interpret the usefulness of the retrieved parameters. Daily intercomparisons of 368nm AOD with the Langley method Harrison and Michalsky (1994) and TOC with the direct sun method Gao et al. (2001) were then made for May 12 and May 22, which were high and low turbidity cases, respectively. Overall, the retrieval results were shown to be physically consistent. As with the synthetic testing, the retrieval tends to yield more useful results, i.e., lower estimated error and better diagnostic values, with increasing turbidity and scattering properties. The results are used to establish apriori and posteriori boundary conditions which can be used in an automated fashion for determining successful retrievals. The results indicate that it is feasible to use the algorithm operationally on the data collected by the UVMRP network to build a data base of TOC and aerosol optical properties. When implemented on the network of ~30 instruments it will provide a comprehensive and internally consistent climatology of ground-based aerosol properties in the UV spectral range, which can be used for both validation of satellite measurements as well as for regional aerosol and ultra-violet transmission studies.
Description
Spring, 2006.
Includes bibliographical references (pages 162-170).
Includes bibliographical references (pages 162-170).
Rights Access
Subject
Atmospheric aerosols
Atmospheric ozone
Ultraviolet radiation -- Measurement