OC-SMART: A Machine Learning Platform for Satellite Ocean Color Data Analysis

We present OC-SMART: Ocean Color -- Simultaneous Marine and Aerosol Retrieval Tool, a powerful multi-sensor data analysis platform. OC-SMART supports heritage, current, and possible future multi-spectral and hyper-spectral sensors from US, EU, Korea, Japan, and China, including SeaWiFS, Aqua/MODIS, SNPP/VIIRS, ISS/HICO, Landsat8/OLI, DSCOVR/EPIC, Sentinel-2/MSI, Sentinel-3/OLCI, COMS/GOCI, GCOM-C/SGLI, and FengYun-3D/MERSI2. OC-SMART products include spectral remote sensing reflectances, chlorophyll-a concentrations, spectral water inherent optical properties (IOPs), aerosol optical depths, cloud mask results, and uncertainty estimates. OC-SMART retrieves high-quality global ocean color products, especially under complex environmental conditions, such as coastal/inland turbid water areas and heavy aerosol loadings. The atmospheric correction (AC) and ocean IOP algorithms in OC-SMART are driven by extensive coupled atmosphere-water radiative transfer simulations in conjunction with powerful machine learning techniques. For each sensor, huge and comprehensive training datasets have been created to support the development of machine learning AC and ocean IOP algorithms. OC-SMART completely resolves the negative water-leaving radiance problem that has plagued heritage AC algorithms. The comprehensive training datasets created using multiple atmosphere and ocean IOP models ensure global applicability of OC-SMART. The use of machine learning algorithms makes OC-SMART roughly 10 times faster than NASAs SeaDAS platform. OC-SMART also includes an advanced cloud screening algorithm and is resilient to contamination by sunglint and cloud edges. It is therefore capable of recovering large amounts of data that are discarded by heritage algorithms, especially in coastal areas. OC-SMART is currently available as a standalone Python package and as a plugin that can be installed in ESAs Sentinel Application Platform (SNAP).