Validation of the atmospheric correction of Landsat OLI imagery and turbidity retrievals using AERONET-OC data from the Bahia Blanca site

ARENA, MAXIMILIANO; DELGADO, ANA LAURA; PRATOLONGO, PAULA; CELLERI, CARLA; VITALE, ALEJANDRO

2021 19th Workshop on Information Processing and Control, RPIC 2021

Institute of Electrical and Electronics Engineers Inc.

Lugar: Nashville; Año: 2021

AERONET-OC collects data from worldwide distributed autonomous CE-318 sun-photometers adapted to provide measurements of the radiance emerging from the sea (i.e., water-leaving radiance). The network supports satellite ocean color validation activities through standardized measurements performed at different sites with a single measuring system and protocol. In January 2020 the Bahia Blanca site was established at the mouth of the main navigation channel of the Bahia Blanca Estuary, a complex system where widespread erosion and strong tidal currents are responsible for the typically high suspended loads in the channel. In this work we present a validation of three different algorithms for the atmospheric correction of Landsat-OLI scenes over the area, based on AERONET-OC data and in situ measurements obtained with a hand-held spectroradiometer. Surface remote sensing reflectance was used to retrieve satellite turbidity and the algorithm performance was tested using turbidity measurements obtained from water samples collected simultaneously with satellite overpasses. According to our results, the best fit was obtained with the atmospheric correction algorithm based on the black-pixel assumption, with the aerosol type estimated using the two OLI-SWIR bands and allowed to vary spatially (per pixel variable epsilon). The aerosol type changed along the estuary, with an epsilon parameter that commonly decreased from the inner section through the mouth. The regression line of in situ measurements with satellite turbidity showed a good performance (R2 = 0.97, n=20), with a high accuracy (RMSE = 4.03, NMAE= 0.16) and no tendency to over or underestimation (BIAS = 0.63). The application of properly validated algorithms to satellite data from high spatial resolution images allowed for the identification of detailed turbidity features in the estuary, resulting from re-suspension phenomena over banks and turbulent currents.