Development and Validation of Multiple-Satellite Data Sets for Global Aerosol Radiative Forcing
Philip A. Durkee
Abstract:
Aerosol properties can be derived from numerous satellite data
sets. Most aerosol retrieval schemes utilize a single sensor system such as
NOAA AVHRR, CZCS, or Meteosat. This study proposes to combine the benefits
of individual satellite sensors into a "composite" retrieval scheme. In
particular retrievals with NOAA AVHRR nominally provide two daytime views
per day. Although AVHRR provides high spatial resolution (1 km) and
multi-wavelength observations, the temporal resolution is weak. The GOES
Imager provides single-wavelength retrievals with similar spatial resolution
to AVHRR but with lower signal-to-noise ratio (less accuracy). However,
GOES provides observations as often as four times per hour. The first focus
of this study will be to develop combined AVHRR-GOES retrievals that exploit
the advantages of each system. In addition, collaboration with TNO
Netherlands will investigate composite retrievals using AVHRR, GOES and the
ERS-2 ATSR. ATSR adds multiple view angles to the mix of tools for
retrieval of aerosol properties. ATSR data is available through TNO and
this collaboration will make this data set available to the Science Team
formed through this NRA.
The second focus of this study will be to advocate the preparation of
comprehensive data sets for validation of retrievals prepared by the Science
Team. Comparison of retrieval schemes as well as validation of an aerosol
climatology will require intercomparison with common in situ, radiation and
satellite data sets. Aircraft, surface (including ship), and satellite data
will be used to test not only the absolute result of a retrieval (such as
aerosol optical depth), but also the individual assumptions imbedded in the
retrieval methods (such as aerosol model size distribution, scattering phase
function parameterization, surface reflectance, etc.). These data sets
would become standard, community data sets that all algorithms could be
validated against.
