Publications

Joshi, ID; Stramski, D; Reynolds, RA; Robinson, DH (2023). Performance assessment and validation of ocean color sensor-specific algorithms for estimating the concentration of particulate organic carbon in oceanic surface waters from satellite observations. REMOTE SENSING OF ENVIRONMENT, 286, 113417.

Abstract
Global observations of ocean color from space provide a unique capability to conduct multi-decadal studies of biogeochemical constituents and processes in the upper ocean layers at the regional, basin, and global scales. The concentration of particulate organic carbon (POC) in the surface ocean waters is one of the biogeochemically important data products derivable from satellite ocean color observations. The data record of standard global POC product is available from NASA and has been obtained from the empirical power-function relationship between POC and a single blue-to-green band ratio of ocean remote-sensing reflectance, Rrs(lambda), which is here referred to as the BR-PF algorithm. Recently, we formulated a new suite of satellite ocean-color sensor-specific global POC algorithms that are referred to as hybrid algorithms because they contain two components, one based on the maximum band-ratio (MBR) and the other on the band ratio difference index (BRDI) of reflectance (Stramski et al., 2022). The present study describes the validation analysis and provides performance assessment of hybrid algorithms for SeaWiFS, MODIS, and VIIRS-SNPP satellite ocean color sensors. For comparison, the standard global (BR-PF) algorithms and the color-index (CI) algorithm from the literature were also analyzed. The analyses were made with an in situ validation dataset containing concurrent measurements of POC and Rrs(lambda) as well as three satellite validation datasets containing data matchups of field measurements of POC and satellite-derived Rrs(lambda) from the SeaWIFS, MODIS-Aqua, and VIIRS-SNPP missions. The hybrid algorithms showed an overall improvement of performance compared to other algorithms, in particular in waters with relatively high POC compared to the standard BR-PF algorithms. The aggregate bias of hybrid algorithm-derived POC is small or negligible (<10%) for the validation datasets, and a measure of median percentage difference is about 20% for the in situ validation dataset and ranges between about 20% and 30% for the satellite validation datasets. In addition, comparisons of POC retrievals from the three sensor-specific hybrid algorithms show good inter-sensor consistency. The validation results of this study indicate that hybrid algorithms have high potential to provide improved POC retrievals within a broader range of POC than the predecessor global algorithms and the capability for generating a consistent long-term POC data record from multiple satellite missions.

DOI:
10.1016/j.rse.2022.113417

ISSN:
1879-0704