| description abstract | Since 1988, the National Oceanic and Atmospheric Administration (NOAA) has provided operational aerosol observations (AEROBS) from the Advanced Very High Resolution Radiometer (AVHRR/2) on board the afternoon NOAA satellites [nominal equator crossing time, (EXT) ?1330]. Aerosol optical depth (AOD) has been retrieved over oceans from channel 1 of AVHRR/2 on board NOAA-11 (1988?94) and -14 (1995?2000) using the first- and second-generation algorithms, respectively. With the launch of the NOAA-KLM series of satellites, in particular NOAA-16 (L) in September 2000 (EXT ?1400), and NOAA-17 (M) in June 2002 (EXT ?1000), an extended and improved third-generation algorithm was enabled. Like its predecessors, this algorithm continues to employ a single-channel methodology, by which all parameters in the retrieval algorithm (excluding AOD) are set globally as nonvariables. But now, in addition to AOD from channel 1, τ1 (?1 = 0.63 ?m), the algorithm also retrieves τ2 and τ3 in AVHRR/3 channels 2 (?2 = 0.83 ?m) and 3A (?3 = 1.61 ?m). The retrievals are made with more accurate and flexible, satellite- and channel-specific lookup tables generated with the Second Simulation of the Satellite Signal in the Solar Spectrum (6S) radiative transfer code. From pairs of τi and τj, the Ångstrom exponent (AE) parameters can then be determined as αij = ?ln(τi/τj)/ln(?i/?j). This paper describes the AEROBS processing and gives examples of aerosol products, along with a preliminary diagnostics of their quality using some of the previously developed self-consistency checks. Interconsistency between the NOAA-16 and -17 aerosol retrievals is also checked. The AODs are largely coherent but distorted by the AVHRR calibration uncertainties, and subject to noise and outliers. These τ errors, unavoidable in real-time AVHRR processing, severely impact the derived AE, demonstrating a fundamental instability in estimating the aerosol model under typical maritime conditions from AVHRR. Consequently, it is concluded that the robust single-channel retrievals should be continued in the AEROBS operations in the KLM era. The more sophisticated multichannel techniques may be tested while reprocessing historical AVHRR data, only after the data quality issues have been resolved (viz., calibration uncertainties constrained, outliers removed, and noise suppressed by spatial averaging). | |
