Avaliação de métodos de correção atmosférica para Sentinel-2 em um sistema estuarino-lagunar tropical

Abstract

Continental aquatic ecosystems are subject to natural and anthropogenic environmental pressures. The understanding of the consequences of these pressures on those environments is limited by the absence of consistent water quality data, measured systematically and at sufficient temporal and spatial resolutions. Execution of in situ measurements are proving to be increasingly challenging considering complex interactions among physical, biological and chemical processes on these environments, aside from other adversities such as difficult access to the monitored area and costs for field campaigns and laboratory analysis. Thus, remote sensing proves to be a useful tool for continuous and wide-ranging monitoring. Sentinel-2 mission, with its multispectral sensors, has radiometric and spectral characteristics suitable for optical applications on inland waters, transition environments and oceans. In addition to the spatial resolution of 10, 20 and 60 m and temporal resolution of 5 days at the equator, these characteristics are favorable to water quality monitoring of continental aquatic ecosystems such as Mundaú-Manguaba Estuarine-Lagoon System (MMELS), located at the state of Alagoas, which receives input from pollution sources such as sanitary effluents and from sugarcane industry. A gap is noticeable in the evaluation of atmospheric correction methods on tropical estuarine systems, especially using a sensor equipped with high spatial resolution and strategic spectral bands for use in bio-optical models and with a perspective of continuity for the next years. As the mission’s multispectral sensors undergo strong influence of atmospheric scattering caused by gases and aerosols, specular reflection, adjacency effects and others, radiance, a parameter which is measured by sensors, needs to be corrected. Therefore, this work proposes to evaluate the performance of different algorithms for atmospheric correction of images from the Sentinel-2 Mission applied to a tropical estuarine-lagoon system, in this case, MMELS. On Manguaba lagoon, iCOR and Sen2Cor performed better in general, despite undergoing adjacency effects caused by clouds. For the Mundaú lagoon, the correction methods showed great variability among R² values and big discrepancies from water column spectral signature. Regarding MMELS evaluated in terms of median values, in general iCOR has showed better general correlation considering in situ measurements. High cloud coverage on the region made it difficult to apply the algorithms, either undermining aerosol removal or causing scattering itself, in addition to the characteristics of land use and occupation around MMELS, which lead to differences in the performance of the algorithms between the Manguaba and Mundaú lagoons.