GRASP: a versatile algorithm for characterizing properties of the atmospheric aerosol and underlying surface

The GRASP (Generalized Retrieval of Aerosol and Surface Properties) algorithm has been developed for enhanced characterization of the properties of both aerosol and land surface from diverse remote sensing observations. The overall concept of the algorithm is described by Dubovik et al. (2014), while the detailed are given in the paper is by Dubovik et al. (2011). The algorithm is based on highly advanced statistically optimized fitting implemented as Multi-Term Least Square minimization (Dubovik, 2004) and deduces nearly 50 unknowns for each observed site. The algorithm derives a set of aerosol parameters similar to that derived by AERONET including detailed particle size distribution, the spectral dependence on the complex index of refraction and the fraction of non-spherical particles. The algorithm uses detailed aerosol and surface models and fully accounts for all multiple interactions of scattered solar light with aerosol, gases and the underlying surface. All calculations are done on-line without using traditional look-up tables. In addition, the algorithm can use the new multi-pixel concept - a simultaneous fitting of a large group of pixels with additional constraints limiting the time variability of surface properties and spatial variability of aerosol properties. This principle provides a possibility to improve retrieval for multiple observations even if the observations are not exactly co-incident or co-located. Significant efforts have been spent for optimization and speedup of the GRASP computer routine and retrievals from satellite observations. For example, the routine has been adapted for running at GPGPUs accelerators. GRASP inherits many aspects used in AERONET retrieval. At first GRASP has been developed for POLDER/PARASOL multi-viewing imager and later adapted to a number of other satellite sensors such as METEOSAT/MERIS at polar-orbiting platform and COCI/GOMS geostationary observations. It can be equally applied to ground-based AERONET and lidar observations. The results of numerical tests and results of applications to real data will be presented.

Go to webcast

Lecture slides