Articles | Volume 10, issue 10
Research article
24 Oct 2017
Research article |  | 24 Oct 2017

The importance of atmospheric correction for airborne hyperspectral remote sensing of shallow waters: application to depth estimation

Elena Castillo-López, Jose Antonio Dominguez, Raúl Pereda, Julio Manuel de Luis, Ruben Pérez, and Felipe Piña

Abstract. Accurate determination of water depth is indispensable in multiple aspects of civil engineering (dock construction, dikes, submarines outfalls, trench control, etc.). To determine the type of atmospheric correction most appropriate for the depth estimation, different accuracies are required. Accuracy in bathymetric information is highly dependent on the atmospheric correction made to the imagery. The reduction of effects such as glint and cross-track illumination in homogeneous shallow-water areas improves the results of the depth estimations. The aim of this work is to assess the best atmospheric correction method for the estimation of depth in shallow waters, considering that reflectance values cannot be greater than 1.5 % because otherwise the background would not be seen. This paper addresses the use of hyperspectral imagery to quantitative bathymetric mapping and explores one of the most common problems when attempting to extract depth information in conditions of variable water types and bottom reflectances. The current work assesses the accuracy of some classical bathymetric algorithms (Polcyn–Lyzenga, Philpot, Benny–Dawson, Hamilton, principal component analysis) when four different atmospheric correction methods are applied and water depth is derived. No atmospheric correction is valid for all type of coastal waters, but in heterogeneous shallow water the model of atmospheric correction 6S offers good results.

Short summary
This work is part of a project funded by the government of Spain, whose objective was to develop a methodology that would allow the grain size and heavy metals estimation in the sediments of the intertidal zone (Bay of Santander) and depth estimation in the subtidal area, using information (VNIR) captured by the hyperspectral sensor, CASI-2, a spectroradiometer ASD-FR (350–2500 nm) in field and laboratory and classical and robust statistic algorithms.