Algorithm development and validation of CDOM properties for estuarine and continental shelf waters along the northeastern U.S. coast

Abstract

An extensive set of field measurements have been collected throughout the continental margin of the northeastern U.S. from 2004 to 2011 to develop and validate ocean color satellite algorithms for the retrieval of the absorption coefficient of chromophoric dissolved organic matter (aCDOM) and CDOM spectral slopes for the 275:295nm and 300:600nm spectral range (S275:295 and S300:600). Remote sensing reflectance (Rrs) measurements computed from in-water radiometry profiles along with aCDOM(λ) data are applied to develop several types of algorithms for the SeaWiFS and MODIS-Aqua ocean color satellite sensors, which involve least squares linear regression of aCDOM(λ) with (1) Rrs band ratios, (2) quasi-analytical algorithm-based (QAA-based) products of total absorption coefficients, (3) multiple Rrs bands within a multiple linear regression (MLR) analysis, and (4) diffuse attenuation coefficient (Kd). The relative error (mean absolute percent difference; MAPD) for the MLR retrievals of aCDOM(275), aCDOM(355), aCDOM(380), aCDOM(412) and aCDOM(443) for our study region range from 20.4 to 23.9% for MODIS-Aqua and 27.3–30% for SeaWiFS. Because of the narrower range of CDOM spectral slope values, the MAPD for the MLR S275:295 and S300:600 algorithms are much lower ranging from 9.9% and 9.1% for SeaWiFS, respectively, and 8.7% and 9.7% for MODIS, respectively. Multi-year, seasonal and spatial MODIS-Aqua and SeaWiFS distributions of aCDOM, S275:295 and S300:600 processed with these algorithms are consistent with field measurements and the processes that impact CDOM levels along the continental shelf of the northeastern U.S. Several satellite data processing factors correlate with higher uncertainty in satellite retrievals of aCDOM, S275:295 and S300:600 within the coastal ocean, including solar zenith angle, sensor viewing angle, and atmospheric products applied for atmospheric corrections. Algorithms that include ultraviolet Rrs bands provide a better fit to field measurements than algorithms without the ultraviolet Rrs bands. This suggests that satellite sensors with ultraviolet capability could provide better retrievals of CDOM. Because of the strong correlations between CDOM parameters and DOM constituents in the coastal ocean, satellite observations of CDOM parameters can be applied to study the distributions, sources and sinks of DOM, which are relevant for understanding the carbon cycle, modeling the Earth system, and to discern how the Earth is changing.