GACP Projects
Remote Sensing of Aerosol over Land with AVHRR
Larry L. Stowe, PI
Abstract:
The objectives of this proposal are to:
1) develop an automated
algorithm for the retrieval of aerosol optical thickness, A, over land where
an
aerosol signal is present in AVHRR cloud-free reflectance data;
2) validate
this algorithm using regional spectral A datasets from NASA's AERONET surface
sun-photometer/sky-radiometers;
3) explore ways of extending to other grid
cells
and times away from AERONET; and
4) evaluate the performance of the algorithm
when implemented into a global AVHRR reprocessing system.
With the successful execution of reprocessing of AVHRR data back to 1981 by the AVHRR Pathfinder Atmosphere (PATMOS) project at NOAA/NESDIS, cloud-free radiance statistics exist for each day in a global 110 km grid for all five channels of AVHRR. Ocean grid cell data have been processed using the NOAA operational aerosol retrieval algorithm to create the most extensive record of A ever compiled. These data are currently in use by over 60 national and international scientific teams, and will be a baseline upon which future satellite measurements will be compared. However, as theoretical climate model studies indicate that the most significant concentrations and radiative effects of aerosols occur over land.
We propose to develop a A retrieval algorithm over land, comparable to the one over oceans, where and when an aerosol signal is present in the 0.63 micron reflectance channel of AVHRR. In a remote sensing approach analogous to the one successfully developed over oceans by NESDIS, observed radiances will be associated with an aerosol optical thickness which yields that radiance in a radiative transfer look-up table (LUT). Results will be validated against surface sun-photometer ground truth sites chosen from AERONET sites distributed over most continents to measure aerosol properties, including spectral A. Land surface reflectance characteristics will be derived empirically by forcing the regression line from AERONET comparisons through zero. The aerosol model will be adjusted to force the regression line from AERONET intercomparisions to have a slope of 1.0. Methods will be explored to extend this algorithm to other periods and locations without AERONET data, using climatologies of the surface and aerosol characteristics, and possibly using visibility measurements for validation. These modifications will be incorporated into the radiative transfer model used to produce LUTs, for each region and time period with measurable aerosol reflectance signals. A final set of LUTs and selection criteria will be developed for use by an AVHRR reprocessing center to help in the construction of a global aerosol climatology back to 1981.