Abstract
The Ocean Colour Climate Change Initiative (OC-CCI) has produced a climate-quality, error characterised, dataset of ocean-colour products (a designated Essential Climate Variable or ‘ECV’). The OC-CCI project uses an optical classification scheme based on fuzzy logic (Moore et al. 2001), to assign product uncertainties on a pixel-by-pixel basis. In this study we show that the pre-existing set of optical water classes derived from in-water remote-sensing reflectance data are insufficient to classify all Rrs spectra present in satellite data at the global scale, particularly in oligotrophic regions. We generate a new set of optical water classes from millions of satellite-derived ocean-colour spectra, providing an improvement in distribution of cumulative class membership values. The use of these classes for uncertainty assignment are demonstrated for chlorophyll-a, utilising a large in situ database of measurements. In addition to being used for uncertainty assignment, performance of multiple chlorophyll algorithms is assessed within each of the classes and a method for blending algorithms while avoiding sharp boundaries, in order to improve final product quality, using class membership is illustrated.
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have