THE FUTURE OF

GLOBAL DISTRIBUTION OF COCCOLITHOPHORE BLOOMS

By Christopher W. Brown

BLOOMS OF THE COCCOL1THOPHORE Emiliania that would allow the blooms to be spectrally The ability to detect huxleyi regionally act as an important source of di- distinguished from the other conditions. An inde- methyl sulfide (DMS) and calcium and pendent data set was also used to establish that the E. huxleyi blooms in alter the optical properties of the surface mixed algorithm was effective in distinguishing coccol- satellite imagery . . . layer (Balch et al., 1991; Holligan and Balch, ithophore blooms from the other water conditions, 1991). These blooms, often covering vast areas, with the exception of whitings, at the spatial reso- provides a method to can be identified in visible satellite imagery be- lution of the global imagery. The classified images cause of the large amount of light backscattered generated from the scheme were then combined assess their biogeo- from the water column. Their presence gives the into monthly, annual, and mission climatologies of chemical importance a milky white to turquoise appearence. The bloom and nonbloom locations. ability to detect E. huxleyi blooms in satellite im- Spectral signatures similar to that of E. huxleyi on basin to global agery, in addition to furnishing biogeographical blooms were found to be most extensive at sub- scales. knowledge of the species at time and space scales polar latitudes, particularly in surface waters of unattainable with shipboard sampling, provides a the North Atlantic, the North Pacific, and the Ar- method to assess their biogeochemical importance gentine shelf and slope (Figure 1). Classified on basin to global scales. blooms covered an average of 1.4 × 10 <' kin-' an- Global composites of Coastal Zone Color nually, with the subpolar latitudes accounting for Scanner (CZCS) imagery (Feldman et al., 1989) 71% of this area. The classified blooms at these were used to map the distribution pattern of E. higher latitudes were inferred to represent the huxleyi blooms and to estimate the magnitude and presence of E. huxleyi blooms because the periodicity of their CaCO~ and DMS production in classification scheme proved efficient in these re- the world's (Brown and Yoder, 1994). Pix- gions and their locations are supported by previ- els of 5-day composite imagery from the entire ous biogeographic investigations. Numerous CZCS mission (November 1978 to June 1986) classified blooms, often quite extensive, were were classified into either bloom or nonbloom also detected in low-latitude marginal , classes based on their mean normalized water-leav- though the conditions responsible for this signal ing radiances using a supervised, multispectral are equivocal. Seasonally, the classified blooms scheme. This empirically based classification tech- in subpolar oceanic regions achieved their great- nique is common in terrestrial remote sensing but est spatial extent in summer to early autumn, has only recently been applied by oceanographers. while those in lower latitudes peaked in midwin- A classification algorithm was developed which ter to early spring. compared the spectral signature of known E. hux- Two important caveats of this approach should leyi blooms (e.g., Holligan et al., 1983) to spectral be noted. First, the results displayed in Figure 1 signatures of nonbloom conditions. Spectral signa- reflect the distribution pattern of coccolithophore tures of E. huxleyi blooms, "clear" blue water, sed- blooms occurring in the surface layer and are bi- iment-laden water, "whiting" (suspended lime ased toward the declining stage (stationary phase) muds), and atmospheric haze were extracted from of the bloom. Detection of blooms is sensitive to CZCS imagery. Decision boundary values for each light backscattered fi'om approximately one attenu- of five spectral feature characters were assigned ation depth and is primarily a function of detached concentrations. Blooms composed pri- mary of cells or occurring at depths deeper than C.W. Brown, Oceans and Ice Branch, Code 971, that sensed by the CZCS would be missed. Sec- NASA/Goddard Space Flight Center, Greenbelt, MD 2077 I, ond, the distribution pattern of blooms and their USA; Ph.D. 1993, University of Rhode lsland (advisor: James spatial extent are biased by both image coverage A. Yoder). and regional atmospheric conditions.

OCEANOGRAPHY'VoI.8, No. 2"1995 59 Fig. 1." Mission climatology of classified coccolithophore blooms in the world's oceans. The maximum spatial extent of blooms de- tected during this period are displayed. Coccolithophore bloom class, white; nonbloom class, blue; land, green," lack of data, black (From Brown and Yoder, 1994).

Acknowledgements • . satellite-detected The amount of -carbon and DMS-sulfur produced by the classified E. huxleyi blooms was I thank Jim Yoder for support and guidance blooms play only a estimated using the mean annual areal extent of throughout my Ph.D. Financial support was pro- minor role in the the blooms and representative values of mixed vided by a NASA Graduate Student Research Pro- layer depth, average cell concentrations found in gram Fellowship (NGT-50605) and NASA HQ annual production of blooms, DMSP concentration per cell, and mass grant NAGW-1891. calcite and DMS on of calcite per coccolith. The blooms detected at subpolar latitudes (40-60 °) are estimated to pro- References Ackleson, S.G., W.M. Balch and P.M. Holligan, 1994: Response a global scale. duce an average of 0.4-1.3 × 106 metric ton of water-leaving radiance to particulate calcite and chloro- CaCO 3 carbon and 1 × 104 tm DMS sulfur annu- phyll a concentrations: a model for Gulf of Maine cocco- ally. These standing stock estimates suggest that lithophore blooms. J. Geophys. Res., 99, 7483-7499. satellite-detected blooms play only a minor role in Balch, W.M., P.M. Holligan, S.G. Ackleson and K.J. Voss, the annual production of calcite and DMS on a 1991: Biological and optical properties of mesoscale coccolithophore blooms in the Gulf of Maine. Limnol. global scale. Oceanogr., 36, 629-643. Although no satellite ocean color sensor has Brown, C.W. and J.A. Yoder, 1994: Coccolithophorid blooms operated since the demise of the CZCS in June in the global ocean. J. Geophys. Res., 99, 7467-7482. 1986, future missions, such as the -viewing Feldman, G., N. Kuring, C. Ng, W. Esaias, C. McClain, J. Elrod, Wide-Field-of-View Sensor (SeaWiFS), will allow N. Maynard, D. Endres, R. Evans, J. Brown, S. Walsh, M. Carle and G. Podesta, 1989: Ocean color. Availability E. huxleyi blooms to be monitored once again. of the Global Data Set. Eos, 70. 634-635, 64~641. These dedicated ocean color missions, in conjunc- Holligan, P.M. and W.M. Balch, 1991: From the ocean to cells: tion with techniques to estimate coccolith and cell coccolithophore optics and biogeochemistry. In: Parti- concentrations in coccolithophore blooms from cle Analysis in Oceanography, S. Demers, ed. Springer- satellite imagery (Ackleson et al., 1994), will im- Verlag, Berlin, 301-324. , M. Viollier, D.S. Harbour, P. Camus and M. Cham- prove our ability to assess the impact of cocco- pagne-Philippe, 1983: Satellite and ship studies of coc- lithophore blooms on the carbon and sulfur cycles colithophore production along a continental shelf edge. in the future. Nature, 304, 339-342. FI

60 OCEANOGRAPHY'VoI.8, No. 2"1995