[1] A retrieval technique has been developed to simultaneously determine column ozone amounts and aerosol optical properties using surface observations of solar ultraviolet direct normal and diffuse horizontal irradiance from a multifilter rotating shadowband radiometer. The retrieval consists of a Bayesian scheme involving a tropospheric ultraviolet radiative transfer model. The technique was tested using cloud-free observations collected during a Mexico City Metropolitan Area air pollution field campaign from April to May 2003. Retrieval results compared favorably to those of independent techniques, including ozone amounts from a direct-Sun method, Langleyderived aerosol optical depths, and aerosol single-scattering albedos from a direct-todiffuse irradiance ratio technique. Further comparisons were performed between the measurements and model simulations when using the retrieval results as inputs, both from the proposed technique and the combined independent methods. Simulations using the results of the new method were found to agree with the observations within the assumed limits of measurement and model uncertainty. It is anticipated that the technique will be applied across a 33-site network of radiometers maintained by the U.S. Department of Agriculture UV-B Monitoring and Research Program for development of aerosol climatologies and for providing ground validation for satellite measurements.

[1] This paper describes a number of practical considerations concerning the optimization and operational implementation of an algorithm used to characterize the optical properties of aerosols across part of the ultraviolet (UV) spectrum. The algorithm estimates values of aerosol optical depth (AOD) and aerosol single scattering albedo (SSA) at seven wavelengths in the UV, as well as total column ozone (TOC) and wavelength-independent asymmetry factor (g) using direct and diffuse irradiances measured with a UV multifilter rotating shadowband radiometer (UV-MFRSR). A novel method for cloud screening the irradiance data set is introduced, as well as several improvements and optimizations to the retrieval scheme which yield a more realistic physical model for the inversion and increase the efficiency of the algorithm. Introduction of a wavelength-dependent retrieval error budget generated from rigorous forward model analysis as well as broadened covariances on the a priori values of AOD, SSA and g and tightened covariances of TOC allows sufficient retrieval sensitivity and resolution to obtain unique solutions of aerosol optical properties as demonstrated by synthetic retrievals. Analysis of a cloud screened data set (May 2003) from Panther Junction, Texas, demonstrates that the algorithm produces realistic values of the optical properties that compare favorably with pseudoindependent methods for AOD, TOC and calculated Å ngstrom exponents. Retrieval errors of all parameters (except TOC) are shown to be negatively correlated to AOD, while the Shannon information content is positively correlated, indicating that retrieval skill improves with increasing atmospheric turbidity. When implemented operationally on more than thirty instruments in the Ultraviolet Monitoring and Research Program's (UVMRP) network, this retrieval algorithm will provide a comprehensive and internally consistent climatology of ground-based aerosol properties in the UV spectral range that can be used for both validation of satellite measurements as well as regional aerosol and ultraviolet transmission studies.

No abstract