Ocean Color and Atmospheric Dimethyl Sulfide' On Their Mesoscale Variability

The mesoscale variability of dimethyl sulfide (DMS) and ocean color is explored to determine the feasibility of a predictive relationship. During NASA's Global Tropospheric Experiment/Chemical hlstmmentation Test and Evaluation (GTE/CITE 3), simultaneous shipboard and aircraft studies were carried out in the North Atlantic, Ibllowed by aircraft studies in the South Atlantic. Surface concentrations of chlorophyll a were measured with an airborne spectroradiometer, the Ocean Data Acquisition System (ODAS), with simultaneous detemfinations of tropospheric DMS. Shipboard measurements of DMS in air ;tnd water as well as in situ chlorophyll a were taken in the North Atlantic. No relation was observed between shipboard aquatic DMS and chlorophyll a or primary productivity. Higher levels of aqueous DMS were not alxvays reflected by atmospheric DMS, although shipboard and aircraft measurements of atmospheric DMS agreed very. well. A significant relationship between atmospheric DMS and ocean color was seen once at low altitudes in both the North and South Atlantic onl': under clean air conditions. Atmospheric DMS levels during the North Atlantic experiment were probably lowered by the presence of mostly polluted air masses in the study area and were, overall, probably not representative of the in situ sea-to-air flux of DMS. Changes in concentration of aircraft-sensed chlorophyllous pigments were not reflected by atmospheric DMS. If a predictive algoritlun is to be found, phytoplamkton blooms should probably be the first place to study an ocean color-DMS relationship.


INTRODUCTION
The first nteasurements of dimethyl sulfide (DMS) in the surface oceans and quantitative estimates of its sea to air flux were made more than a decade ago. Numerous subsequent shipboard measurements have cmffirmed the ubiqui .ty of DMS in the surface oceans [Andreae, 1990 for review]. and suggest that the flux of organosulfur into the troposphere provides a major source for sulfur aerosols in remote marine air masses [Nguyen et al., 1983. It has been proposed [Charlxon et al., 1987] that the distribution of sulfate aerosols in the marine atmosphere exerts significant control (either directly or indirectly) over the Earth's albedo. If true, then the response of the sulfur cycle to climate change could constitute an important feedback in the Earth climate a (as an indicator of phytoplankton biomass), however, has been reported as highly variable both spatially and temporally. A lack of correlation between seawater DMS and chlorophyll a is most likely because only certain phytoplankton species are known to produce significant amounts of DMS. The biological production of DMS and its precursor dimethylsulfoniopropionate (DMSP) seems to be confined to dinoflagellates and pu'mncsiophytes (including coccolithophorcs) both in the field and in cultures lAckman et al., 1966: Barnard et al., 1984;Turner et al., 1988: Keller et al., 1989]. There are numerous records of massive blooms of DMS-producing coccolithophores [Keller eta/., 1989: :llatrai and Keller, 1993]• other pr3'mnesiophytcs. and dinoflagellates [Holligan et al., 1987: Turner et al., 1988: Gibson et al., 1990. Based on the distribution of such bloom-forming species. it is evident that DMS and DMSP vary. temporarily and spatially. dependent on the species compositio• of the flora and the environmental factors controlling their abundance.
Phytoplankton. by means of their pigment content, are easily visible from space. Remote sensing techniques might bc used to perform a first-order estimate of DMS concentrations, given a good algorithm to convert waterleaving radiance to pigment concentration [Gordon et al., 1993] or primaD • production [Balch et al., 1992], and relations to predict DMS from these quantities. In general. the atmospheric concentrations of DMS appear to 23,469 var3. • in concert with DMS in oceanic waters [,4ndreae et a!., 1985: •altzman and Cooper', 1988Cooper', ' •(atrai eta!., 1992. To the extent that the DMS ill air and water are covariant. one can relate either concentration to pigment concentration in sea surface. Nonetheless, because of the temporal a•d spatial scales involved in oceanic systems, the prediction of the global sea-to-air DMS flux will ultimately involve remote sensing techniques. In this paper we ilwestigate the feasibility of obtaining a predictive relationship between DMS emissions and ocean color as observed froin an aircraft, on a regional basis. The Global Tropospheric Experiment (GTE) calibration comparison for airborne chelnical sensors of sulfur gases (Chemical Instrumentation Test and Evaluation (CITE 3)) provided the opportunity to conduct coordinated shipboard and airborne measurements.
Field seasons occurred over the polluted North Atlantic and the relatively unpolluted tropical South Atlantic, in an effort to contrast these two environments.

METHOD
The project was carried out in two phases, consisting of sinmltaneous shipboard (R/V .,ttlantis II) and aircr,-fft studies in the North Atlantic ocean during July and Every day at approxilnately local apparent noon we measured profiles of scalar irradiance using a PNF-300 optical profiling system (Biospherical Instrulnents) and collected salnples for DMS. chlorophyll. prilnary productivity, nutrients. and cell counts with 10-L Niskin bottles attached to the wire. Pigment concentrations were determined in acetone extracts using a Turner-Designs fluorolnetcr [}'entsch and •,llenzel, 1963]. 14C-based productMty measurements were performed at each depth [Strickland and Parsonx, 1972]. Spiked salnples in 250-mL acid-cleaned polycarbonate bottles were incubated for 24 hours on deck under natural sunlight in seawatercooled tubular incubators wrapped in blue acetate (Madico Film TS-51) to reduce the light intensi•' to the appropriatc relative irradiance. Following the incubation, the samples were filtered onto Whatman GF/F filters, rinsed, and counted in Ecolulne. Nutrient samples were filtered and frozen for analysis ashore of NO-3, NO-2, NH+4 . and pO3-4 [Strickland and Parsons, 1972; k2•ro!•//: lV83]. Water salnples were preserved in lugols iodine solution and settled once ashore for cell counts [Sournia, 1978]• which were done with an Olympus BH2 lnicroscope.

Ocean Color •leasurements
Aircrat't ocean color measurements were performed from the NASA Wallops Flight Facility Lockheed Electra leaving froin Wallops Island, Virginia, for the North Atlantic Ocean flights and from Natal, Brazil, for the South Atlantic flights. In the North Atlantic missions, CITE 3 flight paths were planned to accommodate three sea-truth passes over the vessel. At those times, shipboard water and air samples were collected for DMS, and chlorophyll was measured in surface waters. The ODAS measurements consisted of two primary instruments: an infrared radiometer (PRT-5) to measure sea surface temperature and a three-channel visible spectroradiometer for 460, 490, and 520 nm wavelengths.

Goddard Space Flight Center in Greenbelt, Maryland, as well as after each flight. A compilation of calibrations run during 1988 and 1989 was finally used for each channel and at each gain setting (E. Itsweire, personal communication, 1992) [Harding eta!., 1992]. No atmospheric. correction is necessary in the algorithm for aircraft altitudes of less than 150 m (500 ft) [Campbell and Esaias, 1983].
The upwelled radiances were simultaneously sampled every 0.1 s, then reduced to 2-s averages corrected for sunglint, reflection from other sources, and aircraft changes in elevation, pitch and yaw. Pigment concentrations were determilled using a curvature algorithln [Ilarcling eta!., 1992] that first calculates a ratio among the three radiances measured, then incorporates it into a linear regression with factors determined empirically from our data (in situ chlorophyll and ODAS measurements) collected during the ship overflights in the North Atlantic. Finally, the data for each flight were reduced to blocks of time matching those during which aircraft atmospheric samples of DMS were collected, usually 600 s long. Navigation data were obtained from ODAS or the aircraft system. sampling frequency of one sample every 600 s. The generally shallower than the chlorophyll or the primary detection limit of this FPD was approximately 3 ppt of production maximum, with the latter usually located at DMS in a 10-L air sample. Complete details are described in Cooper and Saltzman [this issue].  . Despite uncertainties such as winds and transfer rates and variations in regional DMS production and release, when we averaged all of our seawater DMS and productivity data. the mean would fit perfectly the spatially macroscale relationship between primary production and seawater DMS concentrations described by Andreae and Barnard [1984]  If our relationships between DMS in ocean waters and chlorophyll as well as between aqueous and atmospheric DMS had been significant. as suggested in the literature, and given that ocean color is mostly a measure of chlorophyllous pigment concentration, a certain degree of covariabilit¾ between atmospheric DMS and surface chlorophyll might have been expected. The presence of mostly polluted air masses during the North Atlantic section of the experiment as lnentioncd above may have masked the natural signal present. While up to 30% of the variabili .ty of DMS in seawater has been explained by its relationship to chlorophyll in different oceans Bates, 1983' Andreae, 1990]. the lack of relationship between DMS and chlorophyll as we saw has also been previously reported [.4ndreae and . A correlation between piglncnt and DMS concentrations could also be dralnatically affected by the physiological history of the cells' a healthy. bloolning population of phytoplanktol• lnight produce much different quantities of DMS than a decaying bloom. yet the piglnent values are the same 1• [atrai and Keller, 19931. This is also a major factor preventing a good correlation between piglncnt and prilnar). ß production [Balch et al., 1992]. In a similar maimer. a "tilne-lag" approach might account for the effect of "stage of the blooln" in these relationships.

Depth distributions of DMS in the North
Spring pl•ytoplankton blooms are occasionally. and often predictably, dominated by DMS-producing species for prolonged periods of' time, covering mesoscale [Barnard e! al., 1984; Holligan el al., 1987; •latrai and Keller, 1993] to megascalc areas (500.000 kin 2, south of Iceland) (P. Holligan and W. Balch, personal communicaiion. 1992). During such cases. significant relationships have been observed between aqueous DMS, chlorophyll a, and/or cell nulnbers [Barnard el al., 1984' Tttrner et ai., 1988' AIatrai and Keller, 1993]. Whether such "hot-spots" of biological production of DMS, now known to be spatially significant, can still be considered to contribute the bulk of DMS to the atmosphere seems to be debatable [Andreae, 1990]. Nonetheless, these blooms are still the first place to study an ocean color-DMS relationship if a predictive algorithm is to be found.