Abstract
An atmospheric correction algorithm for medium-resolution satellite data over general water surfaces (open/coastal, estuarine and inland waters) has been assessed in Australian coastal waters. In situ measurements at four match-up sites were used with 21 Landsat 8 images acquired between 2014 and 2017. Three aerosol sources (AERONET, MODIS ocean aerosol and climatology) were used to test the impact of the selection of aerosol optical depth (AOD) and Ångström coefficient on the retrieved accuracy. The initial results showed that the satellite-derived water-leaving reflectance can have good agreement with the in situ measurements, provided that the sun glint is handled effectively. Although the AERONET aerosol data performed best, the contemporary satellite-derived aerosol information from MODIS or an aerosol climatology could also be as effective, and should be assessed with further in situ measurements. Two sun glint correction strategies were assessed for their ability to remove the glint bias. The most successful one used the average of two shortwave infrared (SWIR) bands to represent sun glint and subtracted it from each band. Using this sun glint correction method, the mean all-band error of the retrieved water-leaving reflectance at the Lucinda Jetty Coastal Observatory (LJCO) in north east Australia was close to 4% and unbiased over 14 acquisitions. A persistent bias in the other strategy was likely due to the sky radiance being non-uniform for the selected images. In regard to future options for an operational sun glint correction, the simple method may be sufficient for clear skies until a physically based method has been established.
Highlights
An atmospheric correction algorithm for medium-resolution satellite data over general water surfaces has been assessed in Australian coastal waters
There has been continuing development of models and methods for high sensitivity and low spatial resolution sensors such as SeaWiFS, the MERIS, MODIS and VIIRS. They have been applied to the high spatial resolution environmental satellites such as Landsat and Sentinel-2 at the scales of concern here. These methods include the various processing options within the SeaWiFS Data Analysis System software (SeaDAS) [7,8,9], image correction for atmospheric effects [10], atmospheric correction for OLI ‘lite’ (ACOLITE) [11], the polynominal-based algorithm applied to MERIS (POLYMER) [12] and the Case 2 Regional
It was found in the early match-up experiments that the results for Landsat band 1 were quite different from the match-up data, and they did not conform with expectations of physical models
Summary
An atmospheric correction algorithm for medium-resolution satellite data over general water surfaces (open/coastal, estuarine and inland waters) has been assessed in Australian coastal waters. Many methods and algorithms have been developed to remove the atmospheric, BRDF and terrain illumination effects in order to obtain consistent and comparable measures of Earth surface reflectance from satellite data [1,2,3] These corrections go beyond providing basic calibrated and geocoded products traditionally distributed by agencies and have significant advantages—especially for time series and data analytics. It is often much more difficult to estimate aerosol load using the satellite data itself because of the contamination by sun and sky glint It seems that many current methods face increasing issues as the waters become more complex and the spatial resolutions increase
Sign-in/Register to view highlights & summary Sign-in/Register to access full text options 
Free) 
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a call
