ROSES Proposals
Maximizing Utility of PACE in Coastal and Major Freshwater Ecosystems: Advancing Science for Societal Benefits
PI: Nima Pahlevan - Science Systems and Applications Inc. (SSAI)Co-Is: Caren Binding (Environment and Climate Change Canada); Emmanuel Boss (University of Maine); Raphael Kudela (University of California Santa Cruz); Richard Stumpf (National Oceanic and Atmospheric Administration); Stephanie S. Uz (NASA Goddard Space Flight Center)
With the increasing pressure imposed by the coupled effects of climate change and human-induced activities, harmful algal blooms (HABs) are becoming more frequent; posing a threat to global coastal and inland waters and the ecosystem services they provide. Ocean color (OC) remote sensing has long been harnessed to operationally or semi-operationally monitor spatio-temporal distribution of HABs across the United States coastal/inland waters (e.g., Chesapeake Bay, Lake Erie) or other regions like the Baltic Sea. Such HAB monitoring efforts, however, have primarily been limited only to "detection" using band-arithmetic indices. Such limitations are partially driven by limited accuracy in the retrievals of in-water optical properties in highly eutrophic/turbid waters as well as imperfect atmospheric corrections. Enhanced capabilities offered by the PACE suite of sensors will open pathways to go beyond generating bloom indices and enable identification (bloom type, dominant pigments) and accurate quantification
(pigment concentration) of HABs empowering resource managers with more consistent, quantitative, and robust products for decision-making.
Considering the optical complexity of coastal/inland waters, this research proposes a novel machine-learning algorithm for enhanced retrievals of hyperspectral inherent optical properties (IOPs) from the Ocean Color Imager (OCI) observations. From the IOPs, algal bloom characteristics including types and pigment compositions are estimated. Further, concentrations of pigments like chlorophyll-a and phycocyanin as well as suspended particulate matter (SPM) are predicted, in order to enable a more complete assessment of water quality. Our proposed algorithm exploits the full spectral capabilities of OCI (345-890 nm) and learns to utilize the most relevant spectral regions for the desired products. We will train and validate our proposed machine-learning approach using high-fidelity data from the SeaWiFS Bio-optical Archive and Storage System (SeaBASS) representing a broad range of optical regimes in coastal waters. We will further enrich this dataset using high-quality data available in international databases. The algorithm will also be implemented on data from Sentinel-3 Ocean Land Color Imager (OLCI) imagery to allow for creating seamless IOP and water quality products for global coastal and inland waters.
Our team of experts with a diverse range of skills and experiences in remote sensing, ocean color/optics, in situ and satellite-based HAB monitoring and identification, oceanography, and applied sciences in partnership with resource managers from regions with a long record of HAB episodes, impaired water quality, and applications to aquaculture and fisheries will work with the PACE Project Science Team to develop this suite of products in coordination with operational agencies (i.e., NOAA, EPA) and water authorities serving as end-users.
Considering the optical complexity of coastal/inland waters, this research proposes a novel machine-learning algorithm for enhanced retrievals of hyperspectral inherent optical properties (IOPs) from the Ocean Color Imager (OCI) observations. From the IOPs, algal bloom characteristics including types and pigment compositions are estimated. Further, concentrations of pigments like chlorophyll-a and phycocyanin as well as suspended particulate matter (SPM) are predicted, in order to enable a more complete assessment of water quality. Our proposed algorithm exploits the full spectral capabilities of OCI (345-890 nm) and learns to utilize the most relevant spectral regions for the desired products. We will train and validate our proposed machine-learning approach using high-fidelity data from the SeaWiFS Bio-optical Archive and Storage System (SeaBASS) representing a broad range of optical regimes in coastal waters. We will further enrich this dataset using high-quality data available in international databases. The algorithm will also be implemented on data from Sentinel-3 Ocean Land Color Imager (OLCI) imagery to allow for creating seamless IOP and water quality products for global coastal and inland waters.
Our team of experts with a diverse range of skills and experiences in remote sensing, ocean color/optics, in situ and satellite-based HAB monitoring and identification, oceanography, and applied sciences in partnership with resource managers from regions with a long record of HAB episodes, impaired water quality, and applications to aquaculture and fisheries will work with the PACE Project Science Team to develop this suite of products in coordination with operational agencies (i.e., NOAA, EPA) and water authorities serving as end-users.