Publications

Stramski, D; Joshi, I; Reynolds, RA (2022). Ocean color algorithms to estimate the concentration of particulate organic carbon in surface waters of the global ocean in support of a long-term data record from multiple satellite missions. REMOTE SENSING OF ENVIRONMENT, 269, 112776.

Abstract
As the concentration of particulate organic carbon (POC) in the surface ocean plays a key role in marine biogeochemical cycles and ecosystems, its assessment from satellite observations of the global ocean is of significant interest. To achieve a global multi-decadal data record of POC by merging observations from multiple satellite ocean color missions, we formulated a new suite of empirical POC algorithms for several satellite sensors. For the algorithm development we assembled a field dataset of concurrent POC and remote-sensing reflectance, R-rs(lambda), measurements collected in all major ocean basins encompassing tropical, subtropical, and temperate latitudes as well as the northern and southern polar latitudes. This dataset is characterized by a globally-representative probability distribution of POC with a broad range of values between about 10 and 1000 mg m(-3). This development dataset was created with the use of additional inclusion and exclusion criteria based on well-assured and documented consistency of measurement protocols as well as specific bio-optical and particle characteristics of seawater which are consistent with vast areas of open-ocean pelagic environments. To formulate the algorithms the development dataset was subject to parametric regression analysis. Overall we evaluated over seventy formulas for estimating POC from R-rs(lambda) using seven distinctly different algorithmic categories, each with a fundamentally different definition of independent variable involving R-rs(lambda). Through the goodness-of-fit analysis, we selected the best candidate POC algorithms, referred to as the hybrid algorithms, which are tuned specifically for the spectral bands of SeaWiFS, MODIS, VIIRS, MERIS, and OLCI satellite sensors. These hybrid algorithms consist of two components, the MBR (Maximum Band Ratio)-OC4 cubic polynomial function and BRDI (Band Ratio Difference Index) quintic polynomial function. The MBR-OC4 uses four spectral bands and the BRDI three spectral bands from the blue-green spectral region. The MBR-OC4 algorithm is used for POC > 25 mg m(-3) and the BRDI for POC < 15 mg m(-3). In the transition region the weighting approach is applied to POC derived from the two algorithmic formulas. While the main role of the BRDI is to improve POC estimates in ultraoligotrophic waters where POC is very low, the MBR-OC4 provides improvements, compared with the predecessor algorithms, over a broader range of POC but especially for relatively high POC values. A preliminary analysis of field-satellite matchup datasets based on SeaWiFS and MODIS-Aqua observations shows improved performance of hybrid algorithms compared with current standard algorithms for both SeaWiFS and MODIS. In addition, a reasonable consistency is demonstrated between POC derived from hybrid algorithms applied to example satellite observations with SeaWiFS, MODIS-Aqua, and VIIRS-SNPP. The suite of newly developed algorithms provides the potential next generation version of global algorithms that better represents the spatial and temporal variability within a broader range of POC than the predecessor global algorithms, while also offering a capability to generate a long-term sensor-to-sensor consistent data record of POC that begins with the launch of SeaWiFS mission in 1997.

DOI:
10.1016/j.rse.2021.112776

ISSN:
1879-0704