Uncertainties in remote sensing reflectance at various spatial scales

Abstract

Uncertainties in the retrieval of remote sensing reflectance, Rrs, from Ocean Color (OC) satellite sensors have a strong impact on performance of algorithms for the estimation of chlorophyll-a concentrations and inherent optical properties (IOPs). Uncertainties are highest in the blue bands, especially in coastal waters with low blue-band Rrs values. We recently showed that the main uncertainty contributions when observing at sun glint-optimized geometries are due to two components: variability of in-water parameters and skylight reflected from the water surface. Sunlight propagates to the water and back to the top of the atmosphere (TOA), capturing the instantaneous state of in-water conditions and sky light reflected from the wind-roughened wave facets. Both processes are averaged with the spatial resolution of the sensor. This results in the satellite measured TOA radiance spectrum, which is typically different from vector radiative transfer simulations that are based on the mean values of sea surface reflectance coefficient. Preliminary analysis shows that these two uncertainty components are spatially highly variable. Using the recently released provisional Aquatic Reflectance product for Landsat 8, we analyzed spatial scales of these components for multiple scenes in the open ocean and coastal waters at spatial resolutions ranging from 30 m to several kilometers.