Presenter: Elizabeth C Atwood (PML - Plymouth Marine Laboratory)

Description:
Ocean colour and water quality algorithms are often developed separately for marine and freshwater systems, introducing artificial boundaries to the ocean-inland water continuum. In coastal transitional waters, complex optical conditions and mixing of different water types poses significant challenges in development and selection of algorithms. We present a framework using optical water type (OWT) cluster analysis to address these challenges. Using in-situ hyperspectral data from inland waters together with an objective framework for selection of best performing algorithm, significant improvements in performance across a wide variety of systems is achieved. Building upon insight from two European Space Agency projects (Ocean Colour Climate Change Initiative, OC-CCI; Lakes-CCI), fuzzy clustering methods are used to: 1) generate OWT classes (clusters) representing spectrally distinct waters occurring over the space/time period analysed, 2) assign pixels a membership value for each OWT class, and 3) seamlessly blend optimal algorithms. The improved open-source toolbox is used to assess source-to-sea linkages, allowing more flexible representation of water province transition zones than traditional hard clustering techniques. OWT information can further provide unique insight into optical regime changes within a lagoon or estuary. The toolbox has been developed using MSI and OLCI data, and will be expanded to accommodate hyperspectral inputs through the project. We will show examples from study sites at Plymouth Sound, UK; Tagus Estuary, PT; Elbe Estuary, DE; Venice Lagoon, IT; Curonian Lagoon, LI; Danube Razelm-Sinoe Lagoon System, RO.  Work is part of EC Horizon 2020 CERTO (Copernicus Evolution: Research for harmonized Transitional water Observation), which focuses on closing knowledge gaps in transitional water remote sensing and aims to harmonize different satellite-derived water quality and ocean colour products for the EU Copernicus portfolio.