Aerosol retrieval over land using background composites of geostationary satellite data

Abstract

Over the past decade, aerosol research has seen substantial growth in knowledge and remote sensing capabilities. Aerosols - especially those influenced anthropogenically - are now seen as a significant source of uncertainty in the global radiative balance. Also, aerosol remote sensing will soon be able to monitor aerosols more accurately with the launch of the Terra satellite and its suite of instruments, i.e. the Moderate Resolution Imaging Spectrometer (MODIS). While this and other satellites will provide measurements of aerosol properties over data-sparse regions, the diurnal characteristics will remain unknown because the instruments are limited to once-per-day observations from polar-orbiting satellites. However, observations from the Eighth Geostationary Operational Environmental Satellite (GOES-8) visible channel allow measurement of the aerosol diurnal cycle. Herein, a method is developed which uses GOES-8 to retrieve aerosol optical depth (AOD). A background composite is compiled for each satellite observation time from imagery during the course of a month and is then corrected for atmospheric extinction to obtain a surface reflectance. Optical depths are then retrieved using a radiative transfer model. However, satellite remote sensing requires accurate radiance measurements, so the GOES-8 visible sensor was calibrated from August 1995 through 1999. Results suggest a 5.6% annual degradation in sensor response. These results were used to retrieve AOD over South America in 1995 and 1998. The biomass burning regions provide an area with strong aerosol signal and the recent Smoke, Clouds and Radiation- Brazil (SCAR-B) field experiment provided a relatively well-described biomass burning aerosol. The GOES AOD retrieval was highly correlated with ground truth sites in South America. An increase in the GOES imagery availability in 1998 provided a comparison of daily average values, which resulted in correlation coefficients of 0.91 and 0.97. Retrievals from 1995 also compared well with retrievals from aircraft imagery. Ten retrievals were collocated with AOD estimates from the MODIS airborne simulator, of which, eight were highly correlated. The 1998 retrievals were also compared to satellite radiative flux estimates. In general, correlation was high when the number of comparison points were high. Finally, methods for estimating aerosol mass emission rates from biomass burning are discussed and compared.