| Summary: | This thesis describes a new algorithm based on the optimal estimation approach for the retrieval of atmospheric aerosol and surface properties from the Advanced Along- Track Scanning Radiometer (AATSR). This algorithm is a further development on the Oxford-RAL Retrieval of Aerosol and Cloud (ORAC). The new algorithm is set up to use both visible and infrared channels of AATSR to retrieve aerosol optical depth (AOD), effective radius, white sky albedo at four wavelengths (550, 670, 870, and 1600 nm), surface temperature and aerosol layer height. This thesis can be divided into three main parts : 1) the development of the new ORAC algorithm, 2) comparisons of the retrieved AOD with the aerosol products from visible-channel ORAC retrieval: GlobAEROSOL, and with the measurements from AErosol RObotic NETwork (AERONET), and 3) validations of the retrieved sea surface temperature (SST) with the measurements from ship-based radiometers (Infrared Sea surface temperature Autonomous Radiometer, ISAR) and the measurements from drifting buoys. In this thesis aerosols are assigned to four classes, marine clean at two different relative humidities, spherical dust and non-spherical dust. The estimated retrieval error is 0.012 in AOD and 0.083 K in SST. Comparing with the GlobAEROSOL products, the new algorithm (denoted by ORAC) retrieves lower AOD (0.071 ± 0.012) (median ± RMS) and higher sea surface albedo globally (0.067 ± 0.006). The lower AOD, which also occurs in regional scales, is a promising result as previous studies showed GlobAEROSOL overestimated AOD especially over open ocean. The comparison with ground-based measurements (AERONET) shows a good agreement between ORAC AOD and AERONET AOD over ocean, the correlation is 0.820 at 550 nm and 0.807 at 870 nm, and the differences in AOD between the two datasets are 0.067 ± 0.214 for 550 nm and 0.064 ± 0.167 for 870 nm. In contrast weaker corrections, 0.312 at 550 nm and 0.275 at 870 nm, are found over land, and the median difference between the two datasets are nearly 0.2 ...
|
|---|