Simultaneous retrieval of aerosol and surface properties from a combination of AERONET and satellite data

Abstract

We present a method for simultaneously retrieving aerosol and surface parameters from ground-based and satellite observations collocated in space and time. We show that a combination of down and up-looking observations provides sufficient measurement constraints for characterizing both aerosol and surface properties with minimal assumptions. In order to employ this concept in AERONET processing, the standard inverse algorithm [Dubovik, O. & King, M. D. (2000), A flexible inversion algorithm for retrieval of aerosol optical properties from sun and sky radiance measurements. Journal of Geophysical Research, 105, 20673–20696] has been modified to retrieve surface reflectance in addition to aerosol parameters when co-incident satellite measurements are available.

The method was applied to observations of smoke and desert dust over the Mongu (Zambia) and Solar Village (Saudi Arabia) AERONET sites respectively. The AERONET data were complemented by available observations from the MISR, MODIS, and POLDER-2 satellite sensors. The retrieved bidirectional reflectance factor (BRF) and surface albedo comparison shows good agreement between results obtained using observations from different satellites.

The robustness of the method is tested by analyzing surface albedo time series retrieved during periods of high aerosol optical depth variability and low seasonal changes in surface reflectance. The analysis shows that the performance of retrieval algorithm is stable under different aerosol loadings. It is shown that much of the observed surface albedo temporal variability could be attributed mostly to the combined uncertainty in satellite radiometric calibration and aerosol vertical distribution for Mongu and to differences in satellite angular sampling on different days for Solar Village.

The sensitivity of surface retrievals to assumptions on aerosol vertical distribution and aerosol particle shape are analyzed. It is found that the maximum error in retrieved surface albedo at 0.44 μm is 0.035 for aerosol optical depth 0.85 at 0.44 μm. For aerosol optical depths lower than ∼ 0.7 the error in retrieved surface albedo is less than 0.02. Analysis of particle shape assumptions on surface retrievals showed that aerosol particle non-sphericity significantly affects the angular shape of BRF, but not the surface albedo.

Finally, the sensitivity of AERONET aerosol retrievals to uncertainty in assumed surface reflectance is analyzed by comparing aerosol retrievals obtained with different surface assumptions. It is found that the uncertainty in surface reflectance model employed in the version 1 AERONET operational algorithm is larger than was previously assumed in [Dubovik, O., Smirnov, A., Holben, B. N., King, M. D., Kaufman, Y. J., Eck, T. F., & Slutsker, I. (2000), Accuracy assessment of aerosol optical properties retrieved from AERONET sun and sky radiance measurements. Journal of Geophysical Research, 105, 9791–9806] and may have more significant effect on the retrieved aerosol properties than was documented in that work. In particular, larger errors were encountered for the real part of the refractive index (∼ 0.05–0.07 increase) and maximum of the particle size distribution (∼ 20% decrease) retrievals for the Mongu case, when the aerosol optical depth was relatively small (∼ 0.4 at 0.44 μm). The retrieved single scattering albedo uncertainties were within the error bars (0.03) estimated in [Dubovik, O., Smirnov, A., Holben, B. N., King, M. D., Kaufman, Y. J., Eck, T. F., & Slutsker, I. (2000), Accuracy assessment of aerosol optical properties retrieved from AERONET sun and sky radiance measurements. Journal of Geophysical Research, 105, 9791–9806], with the exception of the 0.44 μm retrievals for the desert dust case when they increased by ∼ 0.09 and 0.07 for low and high aerosol loadings respectively.