A model for remote estimation of ultraviolet absorption by chromophoric dissolved organic matter based on the global distribution of spectral slope

Abstract Absorption of ultraviolet radiation (UV, 280–400 nm) by chromophoric dissolved organic matter (CDOM) precedes a host of light-sensitized surface ocean processes relevant to global climate. These include photo- and biogeochemical cycling of organic material, release of sulfur and carbon-containing gases to the atmosphere, and the photoprotection of marine microorganisms. Synoptic CDOM absorption data in the UV is highly desired yet difficult to estimate by satellite methods as the atmosphere interferes with direct detection of water-leaving UV radiance. The absorption spectrum of CDOM is typically modeled as an exponential function in which a spectral slope parameter, S, describes the rate of decrease in absorption with increase in wavelength. Significant functional relationships are observed in aquatic environments between S and the CDOM absorption coefficient at 443 nm, aCDOM(443). In this paper, we use a large, systematic dataset of spectroscopic CDOM measurements from the U.S. CO2/CLIVAR Repeat Hydrography Survey to examine the relationship between S and aCDOM(443) as a means to model aCDOM(λ) in the UV from ocean color. Our resultant model predicts aCDOM(λ) at wavelengths from 325 to 412 nm from the absorption coefficient of colored dissolved and detrital materials (CDM) at 443 nm, aCDM(443), retrieved by an existing semi-analytical ocean color algorithm. Expected agreement (near 1:1) with the training dataset was achieved (r2 = 0.71–0.85, p = 0, n = 127). Considering inherent satellite data uncertainties as well as the model's limitations in regions with potential terrestrial influence, good correspondence between modeled and in situ values was observed during independent validation with open ocean CDOM data, such as from BIOSOPE (r2 = 0.77–0.85, p   7500 samples) from diverse Case I waters.