Limnol. Oceanogr., 44(3), 1999, 618–627

The spatial and temporal distribution of Trichodesmium in the world’s oceans is highly variable and can potentially be assessed using satellite imagery. Distinguishing these organisms from other phytoplankton in the upper ocean using remotely sensed information, however, requires an optical model that uniquely characterizes Trichodesmium. Here, we parameterize a standard remote-sensing reflectance model using measured values of Trichodesmium’s inherent optical properties, namely the spectral dependence of the chlorophyll-specific optical absorption crosssections and the spectral dependence of the chlorophyll-specific backscatter cross-sections. Values for the chlorophyll-specific absorption cross-sections are described in the previous paper. We calculated the spectral chlorophyllspecific backscattering cross-section (b ) from measurements of the chlorophyll-specific volume-scattering function *b and the spectral backscatter coefficients. b was 0.0027 m2 (mg chlorophyll a [Chl a])21 at 436 nm and 0.002 m2 *b (mg Chl a)21 at 546 nm; these cross-sections are approximately one order of magnitude higher than those for ‘‘typical’’ phytoplankton. The optical model revealed that the combination of high backscatter, absorption, and fluorescence could be used to distinguish moderate to high concentrations (.1 mg Chl m23) of Trichodesmium from other phytoplankton. The model also predicted that surface scum blooms of Trichodesmium would have high reflectance in the near infrared. The high reflectance feature of surface Trichodesmium blooms was used in conjunction with sea truth and data from the advanced very high resolution radiometer (AVHRR) to map a 300,000km2 Trichodesmium bloom off the Somali Coast in May 1995. The nitrogen fixed by this bloom was estimated to be 9.4 3 108 gN d21. These results demonstrate the potential of using remote-sensing techniques in the estimation of nitrogen fixation and the contribution of nitrogen fixation to global biogeochemical processes. One of the most attractive aspects of remote-sensing information is the potential to derive fluxes from estimates of standing stocks. In biological oceanography, changes in optical properties have been used to infer upper ocean chlorophyll concentrations, which can, in turn, be related to primary productivity (Longhurst et al. 1995; Antoine et al. 1 Present address: Chesapeake Biological Laboratory, P.O. Box 38, Solomons, Maryland 20688. Acknowledgments This work was funded by the NASA Global Change Graduate Fellowship to A.S., and the support received from NASA is gratefully acknowledged. A.S. would like to thank Gary Borstad for introducing him to remote sensing. A.S. is especially grateful to Shubha Sathyendranath for her kindness, patience, and useful discussions. We thank D. G. Capone and J. Zehr for ship time and acknowledge the help rendered by the masters and crew of the RVs Columbus Iselin and Seward Johnson, especially the electronic technicians and engineers on board. We thank Howard Gordon and Ken Voss for making the Bryce Phoenix instrument available and facilitating the chlorophyll-specific backscatter measurements, and we especially thank Kay Kilpatrick for providing the software and helping with the calculation of b . Bob Evans and Joanie Splain pro*b vided the AVHRR data. We thank the anonymous reviewers for their comments. P.G.F. was supported by the U.S. Department of Energy under contract DE-AC02-76CH00016 and NASA under grant UPN16135-05-08. E.J.C. was supported by NSF grants OCE9317738 and DEB9633744. This is MSRC contribution number 1142. 1996; Behrenfeld and Falkowski 1997). It has heretofore been very difficult, however, to quantify the temporal and spatial extent of new production in the oceans, let alone the contribution of N2 fixation to that flux. The latter plays a critical role in regulating carbon fixation (Capone et al. 1997; Falkowski 1997), and uncertainty in N2 fixation has prevented a quantitative analysis of new production and the factors that lead to its change. The nonheterocystous, colonial cyanobacterium, Trichodesmium spp., is responsible for most of the N2 fixation in the open oceans (Capone et al. 1997). Hence, a remote-sensing algorithm capable of distinguishing these organisms from all other phytoplankton would be of enormous value in constraining estimates of N2 fixation in the world’s oceans. Together, the optical properties and physiological behavior of Trichodesmium potentially provide a basis for developing algorithms capable of uniquely identifying and quantifying their distributions based on remotely sensed information (Subramaniam and Carpenter 1994). For example, Trichodesmium forms extensive surface blooms that discolor vast regions of tropical and subtropical seas. Gas vacuoles within the cells produce large changes in refractive index relative to the surrounding cytoplasm and water, increasing Trichodesmium’s contribution to the backscattered light (Borstad et al. 1992). Additionally, the unique absorption characteristics (Subramaniam et al. 1999), in conjunction with high reflectance, potentially permit the application of red/near-in619 Trichodesmium optical model frared radiance reflectance-based indices for the retrievals of the spatial and temporal distributions of these cyanobacteria. Here, we analyze the backscatter properties of Trichodesmium and, in conjunction with absorption properties (Subramaniam et al. 1999), parameterize a remote-sensing reflectance model that can be used to derive algorithms from satellite observations of ocean color in the visible and near infrared. From numerical sensitivity analysis of simulated data, we demonstrate the applicability of the model in uniquely identifying Trichodesmium under a variety of oceanographic regimes, and we apply the model to map blooms of the organism in the Arabian Sea. Development of the remote-sensing reflectance model After correcting for atmospheric scattering and absorption, satellite ocean color sensors measure water-leaving radiances, which are apparent optical properties. Optical models are used to relate water-leaving radiance (or remote-sensing reflectance) to inherent optical properties of seawater and its constituents through empirical radiative transfer equations. Gordon et al. (1975) derived an expression from Monte Carlo simulations for approximate numeric solutions to the radiative transfer equations, that related the apparent optical property, R, to the inherent optical properties a and bb: