Possibilities of optical remote sensing of dispersed oil in coastal waters

Abstract

This study is intended to be one of the first steps in assessing the feasibility of remote sensing of dispersed oil in seawater. All optically active seawater constituents, including oil droplets, shape the water-leaving light flux and contribute to a commonly measured apparent optical variable known as remote sensing reflectance (Rrs). Radiative transfer simulations were performed in visible bands for natural seawater in the coastal zone of the Southern Baltic Sea and for a model of seawater polluted by dispersed Petrobaltic crude oil characterised by different droplet size distributions. Our model was supplied by simultaneous in situ measurements of inherent optical properties and the Rrs of natural seawater. The optical description of dispersed oil was based on previous experiments and new application of the Mie solution to the Petrobaltic crude oil of a log-normal size distribution characterised by peak diameters ranging from 0.5 to 500 μm. The results of radiative transfer modelling showed that the typically considered concentration of 1 ppm of oil droplets can locally affect the remote sensing reflectance, causing up to a 6-fold increase or 2-fold decrease, depending on the droplet size distribution. It was demonstrated that the optically significant oil droplet sizes (giving at least 5% contribution to the total scattering coefficient) are <100 μm, as long as oil concentration does not exceed 5 ppm. Moreover, we discussed the influence of dispersed oil droplets on the performance of remote sensing algorithms based on absolute Rrs values or band ratios. Oil dispersion that consists mostly of submicron droplets in a concentration of 1 ppm had the ability to increase blue/green Rrs ratios up to 32%, whereas oil dispersions dominated by micrometre-sized droplets tended to decrease such ratios by up to 18%. Blue/red Rrs ratios were most strongly affected by dispersed Petrobaltic oil, causing a 9%–54% decrease for droplet size distributions characterised by a peak diameter of up to 100 μm. These findings demonstrate why these parameters are considered to be the most useful in the potential remote sensing algorithms.