Applications of Analytical Chemistry to Oceanic Carbon Cycle Studies (1993)

To focus the activities of the analytical chemists and oceanographers that participated in this study, the committee developed a list of priority analytes for sensor development (Table 1). The list of priority analytes focuses on carbon system components and related biologically important analytes, topics of critical importance to oceanographers, Earth scientists in general, and society at large. The list of analytes was divided into four categories, with emphasis being placed on the relative importance to understanding the oceanic carbon cycle:

Priority 3 analytes were suggested by chemical oceanographers who responded to a questionnaire. These parameters are of interest to many oceanographers and could benefit from attention by the analytical chemistry community, even though they are not necessarily related to the carbon cycle directly. The position of analytes within a priority group is not significant. The committee recognizes that virtually any chemical sensor that can operate in seawater could be used to produce scientifically valuable data. For example, the list does not include any of the major ions found in seawater (Na⁺, K⁺, Mg²⁺, Ca²⁺, Cl^-, SO₄^2-) because the concentration of each of these

can usually be calculated more accurately from conductivity measurements than it can be measured directly. However, in some environments, such as those existing in the deep ocean, around deep-sea hydrothermal vents, in evaporite basins, and in river plumes, the concentrations of these elements can be altered significantly. For example, the concentration of calcium changes as a result of dissolution of calcium carbonate shells in the deep ocean, by as much as 75 micromoles per kilogram seawater, or 0.75%. Hence, calcium is not conservative in deep ocean environments. In evaporative environments, the ratios of many major ions change as a result of successive precipitation of calcium sulfate, sodium chloride, potassium chloride, and other evaporite compounds. Sensors of major ion concentrations could be extremely valuable for tracing the sources and sinks of these ions. Table 2 summarizes the technologies considered in this report and how they relate to the priority analytes.

Sensors or analyzers exist for some of the priority analytes, such as O₂, pH, and NO₃^-. The challenge in these cases is to improve sensor stability, response rates, or lifetime. However, for most of the priority analytes, there is no existing sensor or analyzer system that will operate for long time periods without operator intervention. The development of sensors for most of these analytes, such as chlorofluorocarbons or dissolved iron, must circumvent the difficulties posed by low analyte concentrations or interference from other dissolved material. Development of specific sensing chemistry is the ultimate means of circumventing these problems.

The carbon dioxide reservoir in seawater is presumably increasing in size in response to the burning of fossil fuels at a rate of approximately 1 micromole per kilogram per year in surface waters; increases are less in the deep ocean. The magnitude of this change could be estimated by measurements of the carbon dioxide flux into the ocean or measurements of the concentration of carbon dioxide. Measurements of the partial pressure of carbon dioxide (pCO₂) allow us to define the flux of carbon across the air-sea interface, whereas measurements of total carbon dioxide dissolved in seawater (TCO₂)' the sum of CO₂ species present in seawater, allow us to assess the size and rate of change of the TCO₂ pool due to all processes. Because of the large size of the carbon reservoir and relatively small rate of increase, however, these measurements must be made with high precision and accuracy. If 50% of the fossil fuel CO₂ emission is taken up by the ocean, the globally averaged air-sea pCO₂ difference required to drive the present oceanic uptake of fossil fuel carbon dioxide should be approximately 8 microatmospheres (µatm). pCO₂ should be measured to a precision significantly better than this, for example 1 µatm, to produce useful

data. pCO₂ can be estimated from pH measurements in combination with TCO₂ measurements, requiring an accuracy of 0.001 pH unit to match a PCO₂ precision of about 1 µatm. Sensors for pCO₂ and pH would be useful for long-term (>1 month) deployments to provide data about temporal variability. Stability would be of prime importance, but sampling could be less frequent (one per hour) than for profiling sensors, requiring a slower response time. If both the CO₂ difference across the air-sea interface and the gas transfer rate (which may be inferred from wind speed) were measured simultaneously, the net CO₂ flux across the sea surface could be determined. To determine a valid parameterization of the air-sea gas transfer constant, the actual rate of gas transfer should be measured simultaneously with relevant parameters, such as ΔpCO₂, wind speed, temperature difference between water and air, sea surface roughness, and possibly other factors. Measurements of TCO₂ require a precision and accuracy of better than 1 micromole per kilogram per year to be useful in detecting, within a few years, the signal of anthropogenic carbon input to the ocean from combustion of fossil fuels.

In seawater, the cationic charges of strong electrolyte species (such as Na⁺, Mg²⁺, and Ca²⁺) always exceed the anionic charges (such as Cl^- and SO₄^2-) . The total ionic balance in seawater is maintained by the anions [OH^-, HCO₃^-, CO₃^2-, and B(OH)₄^-] generated by dissociation of weak acids. Alkalinity is a measure of the ionic charge imbalance of the strong electrolyte species, and hence is independent of temperature and pressure. It is changed by dilution and biological processes such as dissolution or precipitation of calcium carbonate and by utilization or respiration of NO₃^-. Its value ranges from 2200 to 2350 micromoles per kilogram in the surface ocean and ranges from 2400 to 2450 micromoles per kilogram in the deep ocean. Because the pCO₂ in surface waters and hence the air-sea partition of CO₂ is sensitive to alkalinity changes, it is one of the important quantities that needs to be monitored in global seawater.

The stable carbon isotopes, ¹²C and ¹³C, are fractionated in various natural processes. δ¹³C is a measure of the per mil deviation of the ¹³C/¹²C ratio from that of a Pee Dee Belemnite (PDB) limestone standard. The photosynthetic pathways (C-3 and C-4) fix the lighter carbon isotope (¹²C) preferentially, so that δ¹³C values of plant samples are 6 to 23% less than those of the original CO₂ Pools, ¹² CO₂ preferentially passes across the air-sea interface into the atmosphere, so that TCO₂ in the surface ocean has δ13C values about +1 to 2%. Because of the release into the atmosphere of fossil fuel CO₂ (ancient plant material) with a δ¹³C of about-27%, δ¹³C values for atmospheric CO₂ decreased from-7.5% in 1978 to-7.8% in 1988. Likewise, average δ¹³C values for CO₂ dissolved in surface ocean waters decreased from about +2% to +1.6% during the past 20 years. This information has been used to put limits on global CO₂ budget calculations

by Quay et al. (1992). To be useful to estimates of the anthropogenic CO₂ uptake by the ocean, the ¹³C measurements must be extremely carefully standardized. The isotope ratio is measured using isotope ratio mass spectrometry with a precision of about 0.04% in the δ¹³C unit. Control of isotopic standards must be better than 0.05%.

Measurement of the levels of naturally produced ¹⁴C (half-life of 5730 years) in both dissolved CO₂ and organic carbon allows the estimation of long-term turnover in the ocean. ¹⁴C produced through atmospheric testing of nuclear weapons nearly doubled the existing natural levels of ¹⁴C in the atmosphere in the late 1950s and early 1960s. Bomb-produced ¹⁴C provided a spike tracer to observe short-term turnover of carbon within the CO₂ and organic pools. At present, ¹⁴C in DOC is measured on carbon dioxide produced by oxidation of seawater DOC using ultraviolet radiation or high-temperature catalytic oxidation. ¹⁴C in POC is measured on carbon dioxide produced from closed-tube combustion techniques. The ¹⁴CO₂ evolved from these small samples is measured using accelerator mass spectrometric techniques with a precision of 0.4 to 2%. Better precision is needed, especially for deep-ocean samples in which the bomb ¹⁴C signal is smaller.

The combined activity of planktonic organisms in the ocean (phytoplankton, protozoans, zooplankton, viruses, and bacteria) results in the fixation of dissolved carbon dioxide and nutrients into organic matter in the ocean surface layer and the export of some fraction of the fixed carbon downward. This process, the biological pump, is responsible for maintaining the gradient of increasing TCO₂ with depth in the oceanic water column. It is necessary to improve our capability for measuring the spatial and temporal distributions of a variety of chemical compounds, to gain a better understanding of how the biological pump operates, and to be able to predict future changes in the ocean carbon cycle.

Biological activity in the upper and deep ocean results in the formation and breakdown of particulate and dissolved organic carbon and nitrogen (POC, PON, DOC and DON). Particulate organic carbon levels in surface waters of the open sea range from about 0.5 to 50 micrograms carbon per kilogram seawater. The degree of cycling of organic material in the surface ocean and the downward flux is still only poorly known. There is a slow fallout of material to the bottom year-round, with seasonal pulses in areas underlying surface zones of high primary productivity. Presently, organic material can be measured only in discrete samples. New approaches for measuring these compounds at greater frequencies are needed.

Concentrations of dissolved organic carbon in open ocean waters are low (35 to 200 micromoles carbon per kilogram seawater). Despite the low

concentration, the mass of DOC in the ocean is about half as large as the pool of carbon in terrestrial biomass and comparable in size to all the carbon dioxide in the atmosphere. Transformations in the size and composition of this DOC pool have the potential to alter the balance of the global carbon cycle. However, we do not yet have an accurate estimate of the mass of DOC or a detailed understanding of what controls its rate of turnover, much less its composition or function in microbial food webs. There is not yet a generally accepted analytical method to characterize DOC, although there is much interest in improving high-temperature combustion methods using discrete samples. For precise determination of the oceanic carbon pool, DOC sensors and analyzers must be capable of producing data of high accuracy and precision (1%) in the oceanic concentration range. Moored sensors capable of operating for periods of weeks to months are desirable because DOC concentrations may change dramatically during algal blooms, for example. Identification and measurement of specific components of DOC are as important as measuring total DOC. This approach will require basic research into DOC composition, which may also be aided by new measurement technologies.

There is also a great need to measure DON with accuracy. DON is the primary form of nitrogen in much of the surface ocean and may limit phytoplankton production in many areas. The difficulty in measuring DON is largely analytical, and accurate values for deep water will only be obtained after new direct DON methods are developed. Measurement of certain forms of DON is probably more important for understanding the global carbon cycle than is knowledge of concentrations for most trace metals.

The major plant nutrients (nitrate, phosphate, silicate, and ammonium) that support primary production should be measured from once to several times daily over periods of months to a year, or continuously from ships, at a precision of 0.01 micromole per kilogram. The use of these nutrients varies by phytoplankton species. Nutrients can be supplied by mixing and flow from underlying water masses and by regeneration of nutrients through the action of bacteria and herbivores. The ability to predict the timing and strength of phytoplankton blooms depends on an understanding of the effects of nutrient supply and physical factors (such as water density gradients and turbulence) on primary production.

Iron, an important plant micronutrient, is adsorbed rapidly onto particles and becomes unavailable for uptake by phytoplankton. The major source of iron to the ocean is in the form of atmospheric dust. Relatively little is known about the temporal variability of total iron in surface waters, but available data suggest that measurements with a precision of 20 picomoles per kilogram or better might be required. Almost nothing is known about iron speciation [Fe(II), Fe(III)], although the hydrolyzed product of Fe(III) is the stable form. Because algal productivity varies daily, season

ally, and annually, understanding primary production requires that measurements be made with appropriate frequency to describe the variability at each of these time scales.

The fractionation of nitrogen-15 versus nitrogen-14, as designated by δ¹⁵N, can provide information about biological processes involving nitrogenous compounds because such processes fractionate these two stable forms of nitrogen. Techniques for measuring δ¹⁵N natural abundance in nitrate, ammonium, and particulate matter with a precision of 0.1 δ unit are needed.

Phytoplankton contain a wide array of pigments in addition to chlorophyll, and pigment concentrations, types, and ratios are often specific to certain taxonomic groups. Thus, sensitive detection of individual pigment types with a precision of at least 0.1 nanogram per liter can provide information on phytoplankton community composition. More accurate, widespread, and frequent measurements of phytoplankton pigments are also needed to validate satellite estimates of plant biomass. In recent years, flow cytometry has been adapted from medical use to sort and count phytoplankton cells containing different pigment types, increasing the level of information about phytoplankton communities. Knowledge of pigment concentrations is necessary to study the coupling of primary productivity with higher levels of the oceanic food web, such as the productivity of marine fisheries.

Oxygen accumulation in the sunlit upper water column is a reflection of net primary production. Oxygen is produced as part of the photosynthetic process and consumed by respiring organisms. To be a useful indicator of primary production and respiration, oxygen must be measured with a precision of 0.5% or better against a background concentration of 350 micromoles per kilogram in surface waters, at a frequency of about 10 times per day.

Chlorofluorocarbon (CFC) compounds (Freons^®) are important tracers of ocean circulation. Since CFC-11 and CFC-12 were released at different rates, their concentrations as well as their ratios can be used to determine when a water mass left contact with the sea surface over the 50 years since Freons^® were released into the atmosphere and taken up by the ocean. CFC-113 and carbon tetrachloride are being investigated as additional tracers. Picomole levels of CFCs dissolved in seawater are determined using a gas chromatograph equipped with an electron capture detector. The World Ocean Circulation Experiment (WOCE) Hydrographic Program calls for measurements with a precision and accuracy of 1% and a detection limit of 0.005 picomole per kilogram (Sarmiento, 1988).

Argon-39 is a cosmogenic isotope that is chemically inert and has a radioactive decay half-life of 269 years. Because it is excluded from bio-

logical cycles and its decay half-life is roughly comparable to the time scale for global ocean circulation, it is suited for studies of ocean circulation rates. However, because of its low concentrations in seawater (0.03 to 0.1 dpm per liter of argon gas at standard temperature and pressure), a seawater sample of about 1000 liters is needed to provide enough argon for a determination with a 10% precision, counted over several months. Thus, only a few measurements are made per year. A more rapid analytical method is needed.

Helium-3 is a decay product of radioactive tritium (³H, half-life = 12.44 years) that has been produced by nuclear bombs as well as naturally by cosmic rays in the upper atmosphere. Because virtually all ³He atoms escape from the surface ocean to the atmosphere, the ³He/tritium ratio in subsurface seawater samples indicates the time since the water's last exposure to the atmosphere. Both ³He and tritium are measured by gas mass spectrometry. Alternatively, tritium may be measured by gas counting with a detection limit of 0.05 to 0.08 tritium unit, where 1 tritium unit represents a ³H/H ratio of 1 × 10^-18. A degassed water sample is sealed and stored for several months to allow the decay product ³He to accumulate in the container. The amount of ³He is then measured by mass spectrometry, yielding a detection limit of 0.001 to 0.003 tritium unit when 400-gram water samples are used. With this technique, the time since a water mass left the surface can be determined within a range from several months to 30 years.

Analytes in this category are not necessarily related to the carbon cycle, but are of interest to chemical oceanographers. Improvement in technologies to measure these analytes could advance the ability of chemical oceanographers to study how Earth systems function.

Several elements, particularly zinc and copper, could play a role as trace nutrients for phytoplankton. They are known to be important for growth of terrestrial plants, but neither the requirement for these nutrients nor the elemental distributions in seawater are well known. The biological availability of both zinc and copper is controlled by their complexation with organic material. Analytical methods that have the distinction of being able to discriminate chemical forms of the metal are needed. These measurements reflect the chemical reactivity and biological availability or toxicity of the metal more accurately.

The transfer of particles and gases between the ocean and the atmosphere is an active area of research. Information from studies of air-sea transfers is crucial for putting boundaries on models of global cycles of carbon and other biologically important elements. Aluminum and lead are delivered to the surface ocean, primarily on windborne particles, so that

concentrations of these elements can be a measure of windborne particle transport. Dimethyl sulfide is produced in the surface ocean and released into the atmosphere. In order to investigate the circulation rate of oceans and air-sea gas transfer rates in localized areas, inert compounds (such as SF₆) that can be detected in minute quantities have been injected into surface as well as deep oceans. SF₆ is an ideal tracer because of its low solubility in water (hence it does not alter the density of seawater significantly), lack of a natural source, and high detection limit (10^-12 mole by gas chromatograph equipped with an electron capture detector). SF₆ has been injected into water masses of a desired density to investigate their lateral and vertical mixing rates. Use of tracers purposely injected into the ocean will become more common in the future. Because noble gases present in seawater (helium, neon, and argon) are excluded from biological cycles, they often serve as useful tools in oceanography for tracing physical mechanisms and processes. The difference in the ratios of these gases, as well as nitrogen, in air and in surface ocean water has been used to study air-sea gas transfer mechanisms, such as air injection via gas bubbles. Other noble gases, such as radon-222 (having a half-life of 3.8 days) produced in water or in sediments from the uranium decay series have been used to estimate the material transfer rate across the air-sea or sediment-water interfaces. Helium has been injected into surface waters with SF₆ in order to investigate the gas transfer rate in open oceans. Krypton-85, having a half-life of 10.67 years, is released into the atmosphere from nuclear fuel refineries and nuclear weapons factories. It has been used as a tracer of large-scale atmospheric circulation, as well as ocean water circulation.

Tracers of photochemical reactions include low-molecular-weight compounds, such as formaldehyde, pyruvate, and acetylaldehyde. The rates of these photochemical reactions are important to measure so that natural degradation of DOM can be quantified. Also, their variability due to increased ultraviolet radiation (from decreases in tropospheric ozone levels) should be studied. The ChemRawn IV conference had a major focus on photochemical reactions (Goldberg, 1988).

Since the discovery of widespread hydrothermal activity on the seafloor and plumes of altered water in the water column, it has become obvious that circulation of seawater through both high-temperature and low-temperature rocks can add or remove elements, potentially affecting global balances of some elements. For example, dissolved manganese concentrations in open-ocean seawater are as low as 0.2 nanomole per kilogram, but concentrations can be as high as 10 millimoles per kilogram in hydrothermal environments. ³He is enriched in hydrothermal plumes and can be used as a tracer of the volume of hydrothermal fluids and their dispersion away from their source. Sensors with high response rates (>10 measurements per minute) would be useful for determining the spatial distribution of manganese.

Many trace metals are removed from the oceanic water column through adsorption to particle surfaces or incorporation in biological particles. Adsorbed radioisotopes allow oceanographers to study particle transport phenomena. The naturally occurring isotopes of thorium, as well as lead-210, are particularly useful for these studies.

Most analyses consist of four operations—sampling, pretreatment, calibration, and detection. These operations are combined in chemical sensors. For in situ measurements, the transducing device or sensor is brought to the sample, rather than a sample being transported to the sensor. In general, most chemical analyses benefit from minimal sample manipulation, making in situ measurements preferable to ship-based measurements, which in turn are usually better than those in which the sample is brought back to a land-based laboratory. Conversely, the difficulties of performing chemical analyses multiply as the analytical method and instrumentation are brought from land-based laboratories to ships, and perhaps ultimately incorporated into an immersible device. In performing chemical analysis of seawater, several factors determine the desirability and feasibility of performing the analysis in situ, on board ship, in land-based laboratories, or remotely from a satellite or aircraft. These issues are summarized below.

Of course, all chemical analyses may be performed in a land-based laboratory, but there are several reasons for not taking this approach. Ships can only collect a limited number of samples because of time, expense, sea state, and space limitations. Samples from field sites are necessarily brought back in some form that may be altered (e.g., adsorb to or desorb from walls of container). Microorganisms present in samples may, by their metabolism, affect the concentration of the analyte of interest for long periods after the sample is collected, unless metabolic processes can be stopped by sterilization or poisoning of the samples. The sterilization procedure must not affect the analyte of interest. In addition, the composition of a seawater sample can be changed by photochemistry if it is exposed to light. As water samples collected from great depths are depressurized, the equilibria of dissolved gases and speciation of ions change. Procedures have been developed to minimize these negative effects. A possible alternative for some analytes is to deploy a sample collector, rather than a sensor, remotely. If samples could be stabilized, the sampling platforms could be deployed for months, perhaps, and recovered later for sample processing. In addition to these purely chemical issues, it is nearly always useful to achieve rapid analyses of samples to guide further sampling. If samples are not analyzed until later, flaws or bias in the sampling technique will not be evident until

it is too late to resample or to alter the sampling technique for the remaining samples.

Despite these caveats, many analyses are done on land either because the sample can be stored without changing the concentrations of the analytes it contains or because the apparatus required for the analysis cannot be operated on board a ship. For instance, some radionuclides are measured on land for both reasons. Samples analyzed on land can be spiked immediately with another (artificial) isotope of the same element to fill the adsorption sites on container walls and to serve as an internal standard. The mass spectrometers required for isotope ratio analyses are often too sensitive to vibration and motion for shipboard use. Analytes present at greater than trace levels, or which can be stabilized with some pretreatment, may also be analyzed successfully on shore.

The barriers to introducing some analytical techniques for shipboard use can be daunting, particularly those requiring complicated equipment; even simple tasks like weighing chemicals on a balance or reading a buret can be challenging when the seas are rough. Presently, most shipboard chemical analyses are carried out with instruments not designed for shipboard use. Instruments designed for use on ships must endure a humid, corrosive atmosphere; withstand shock and violent motions of the ship; survive greater extremes of temperature than in the usual laboratory; and function with limited and poor quality power and water supplies. Approaches to develop new instruments will need to include not only chemistry but also physical engineering and design to make the shipboard environment less daunting for chemical instrumentation. Better ships and power systems are needed. These problems affect not only the instrument itself but also any computers that control the instrument or collect the data it generates. This is increasingly important, as many oceanographers take part in research cruises that originate in ports distant from their home institutions and may have to take their own specialized equipment with them. At times, it is most convenient to package an entire laboratory in a standard shipping container, which may be transported to the ship and carried as deck cargo. Some common ''utilities'' found in chemical laboratories, such as fume hoods and gas lines, are not routinely used on ships. In addition, it should be noted that shipboard conditions are often suboptimal for the human operators of the instrumentation as well; seasickness affects the efficiency of many scientists at sea.

The above information demonstrates why some analytical techniques are seldom or never used at sea. For instance, the large gas lasers typically used in ultrafast and Raman spectroscopy are fragile, require 500-volt three-phase power, and must be cooled by a significant flow of fresh water. Yet despite these limitations, nearly all types of analytical techniques have been used at sea in some form, including electrochemical techniques, absorption

and emission spectroscopies, optical and electron microscopies, as well as classical titrations and manometry. In fact, most types of analytical chemical determinations on seawater samples have been performed on board ships. However, the majority of shipboard analyses are "routine" analyses, such as measuring oxygen by Winkler titration and nutrients by colorimetry. Almost all metal, isotope, and organic carbon analyses are carried out on land. Shipboard techniques are compromises between the need to process many samples rapidly and the cost and difficulty of performing analyses at sea, as compared with land-based laboratory measurements.

For many of the same reasons that measuring analytes at sea is preferable to returning samples to land, it is desirable to measure analytes in situ. An additional benefit of using in situ sensors is the potential savings in labor cost per sample. In general, adapting chemical analyses for use in the ocean is much harder than adapting methods and instrumentation for shipboard use. Even simple devices must be completely redesigned. Power and data transfer are often problems when the apparatus is at the end of miles of cable. High-density data storage makes real-time data transfer technically unnecessary, although transfer may be required for scientific purposes, as for real-time modeling and prediction. Adequate power remains a problem, however. Pressure-proof housings are available, and electrodes and windows can be engineered, but other limitations exist. In particular, even the simplest mechanical manipulations of solutions, such as dispensing, mixing, and titration, are difficult to automate. Various approaches have been used to circumvent these problems. Laboratory robots have been used in land-based laboratories for multistep analytical procedures, but adaptation for shipboard, and especially in situ, applications will not be possible in the near future.

Perhaps the most important new approach to chemical measurements has been the use of sensors for oceanic chemistry. Sensors comprise a transducer and its supporting electronic instrumentation. The key feature of sensors is their ability to monitor the concentration of a particular analyte continuously, so that the dimension of time can be added to the traditional three dimensions of spatial measurements. An example of a sensor is a pH electrode, coupled with a high-impedance voltmeter and a means of standardization and temperature compensation in situ. In principle, such a sensor can monitor pH continuously for days at a time while transferring the data to a recorder or memory device. One can contemplate towing an array of sensors at various depths simultaneously, obtaining three-dimensional continuous data sets, improving on the two-dimensional data available from vertical profiles. Most sensors must be calibrated regularly.

For in situ measurement, the foregoing discussion has implied the use of profiling devices lowered from stationary surface vessels or towed behind moving vessels. Yet the problems created by currents, weather, and

high-sea states in deploying instruments from surface vessels (and successfully retrieving them) has prompted development of other platforms. In particular, the mechanical problems in deploying a device 1 kilometer deep in the ocean are caused more by the drag of the 1000-meter cable than by the weight of the instrument itself. Thus, other strategies have been used to bring the analysis to the sample, including unmanned underwater vehicles, buoys, and manned submersibles. At present, for unmanned submersible vehicles, the problems of autonomous operation, navigation, small size, power requirements, and reliability overshadow the advantages of cable-free operation, but solution of these problems will eventually make wider use of these devices likely. There have been significant advances over the past decade in the development of optics and electronics for underwater vehicles, including remotely operated vehicles (ROVs). These systems provide the opportunity for optical observations, as well as acoustical and chemical measurements by appropriate instruments attached to the ROV.

Buoys, on the other hand, are already widely used for many sorts of physical oceanographic measurements, and the development of new sensors would allow buoys to be important platforms for chemical and biological oceanographic measurements as well. Their capabilities for long-term unattended operation, even in environments as hostile as the Arctic, are well demonstrated. Issues in the use of buoys are the availability and reliability of sensors, and electrical power sources. However, modest sampling frequencies (e.g., 1 to 10 samples per day) would enable them to operate from solar energy and batteries. Newly developed methods for data storage and on-board processing will allow greater data storage before retrieval. Data transmission from sensors to satellites is still a limiting factor, however.

Although not well suited for chemical oceanography per se, satellite-and aircraft-based remote sensing is a growing adjunct to direct chemical measurements. Despite their limited sensor suites, satellites offer unsurpassed volumes of data covering vast areas of the ocean, virtually in real time. Moreover, the nature of orbits is such that the same measurement may be repeated over the same section of ocean for years at a time, which is very useful for evaluating long-term variability. Satellites primarily conduct passive sensing of electromagnetic radiation in the visible, near-infrared, and microwave frequencies. In addition, specialized satellite radars actively measure mean surface height and sea state. Sea surface temperature may be determined very precisely from satellites, providing additional physical oceanographic information. Physical oceanographic information is an important adjunct to chemical and biological measurements because temperature affects chemical reaction rates, gas solubilities, and organism growth.

Sea state affects gas transfer by changing the sea surface area due to waves and entrained bubbles.

Passive satellite measurements of reflected intensity at visible wavelengths allow mapping of ocean color. Ocean color is dominated by the amount of chlorophyll and phaeopigments present, which in turn can be related to primary productivity. Ocean color is also an important indicator of the presence or absence of nutrients and various physical oceanographic phenomena.

The major constraint to using satellite remote sensing is the relative opacity of the ocean and clouds to electromagnetic radiation. Thus, satellite measurements are limited to the very top layer of the ocean and usually provide little evidence about processes occurring below the ocean surface.

Their high cost has limited the number of satellites launched for ocean observations. Among them have been the U.S. Navy's Seasat and Geosat. In addition, the Landsat series, the Earth Resources Satellite-1, TOPEX/Poseidon, INMARSAT maritime communications satellites, various weather satellites, the Global Positioning System satellites, and the Argos transponder have all greatly aided ocean science in general, if not chemical oceanography in particular. A number of other satellites will be launched in the next 5 years to monitor physical and biological oceanographic conditions. Despite this limited variety of sensors, satellites are an indispensable tool for oceanography.

Aircraft can perform almost all the measurements that satellites can, but on a more limited spatial scale. Some additional measurements can be made from aircraft that cannot be made from satellites. For example, several aircraft have been used as platforms for remote laser-based spectroscopies (lidar), using wavelengths that are strongly attenuated by the atmosphere (e.g., ultraviolet) and therefore not practical for satellites. Also, an aircraft can determine whether ships are in the way before turning on a potentially hazardous laser beam. Although, in principle, aircraft can scan very wide areas of the ocean (e.g., 80,000 square miles per hour in an SR-71 aircraft flying at greater than 70,000 feet altitude), in practice for optical studies they often operate at lower altitudes to minimize attenuation and scattering, and consequently have a much reduced "footprint." In comparison with ships, however, their enhanced capabilities are substantial. One shortfall in existing aircraft is their relatively limited range and the limited time available to make measurements. Aircraft built for maritime surveillance, such as the NASA P-3 aircraft, can spend only a few hours over mid-ocean areas; even large jet transports have less than 15 hours endurance. As a result, aircraft are widely used primarily for coastal monitoring rather than for the open ocean. A notable exception was the use of aircraft passes concurrent with ship-based sampling in the 1989 Joint Global Ocean Flux Study North Atlantic Bloom Experiment. Whereas a research vessel might cost $10,000

to $20,000 per day to operate, aircraft costs approach $48,000 per day for a modest-sized jet transport.

As noted above, aircraft can perform active optical spectroscopic measurements remotely. Potential spectroscopic techniques include fluorescence, Raman, and Rayleigh scattering measurements. Pulsed lasers can, in principle, provide depth resolution in the same manner as the range gate in a monopulse radar. Adequate power and cooling can be provided on board aircraft, but alignment and stability of optical devices remain problems. The cost and difficulty of developing instrumentation for airborne use is thus at least equal to that of developing shipboard devices, because airborne systems are also constrained by weight and size limits. Moreover, the available optical techniques remain limited in the chemical analytes that may be sensed because they are remote methods.

Sensors can be based on a variety of transduction mechanisms, including electrochemical, optical, mass, and thermal. Different types of sensors, along with their transduction mechanisms, will be discussed later in this report. All sensors possess a transduction element connected to supporting instrumentation. Selectivity is achieved via the transducer. The qualities of an ideal sensor obviously depend on the application, with different qualities necessary for sensors used in shipboard laboratories, towed sensors, or sensors deployed on long-term remote moorings. Depending upon the specific application, sensors should possess the following qualities.

• Manufacturable in large quantities. Handcrafted, one-of-a-kind sensors cannot hope to gain the acceptance necessary for routine implementation by ocean scientists. Sensors must be reasonably consistent in their performance characteristics. Moderate-to large-scale manufacturing processes applied to sensor fabrication will assure this type of consistency.

• Cost effective. Many instruments are expensive, yet the scope of the ocean measurement problem ideally is addressed by making many continuous measurements with a large number of instruments. The likelihood of deploying many instruments increases with decreasing cost.

• Fully automated. In order to be deployed for unattended operation, sensors must be fully automated. This automation pertains principally to the supporting instrumentation necessary for operation of the sensor and for signal processing. Power requirements, on-board data handling, and data storage must all be configured appropriately for the particular application. It is desirable, but not required, that sensors and supporting instrumentation be able to transmit data via telemetry to receivers on satellites, ships, or land.

• Stable and long lasting. Most sensors will depend upon a chemically selective layer attached to an appropriate transducer element. The system of chemical reactions contained within the selective layer must be stable or reversible or, if irreversible, be based on a renewed or continuously delivered reagent that maintains the ability of the sensor to detect the desired analyte.

• Calibrated. In order for sensors to gain acceptance as deployable measurement tools, they must remain calibrated for extended periods of time (weeks to months) or must be self-calibrating.

• Resistant to environmental conditions. After deployment, sensors must be resistant to mechanical shocks from waves and be insensitive to, or compensate for, changes in temperature, pressure, salinity, and biofouling that they will invariably encounter in the ocean environment. Biofouling and corrosion are major problems for instruments that are deployed for long periods of time. Appropriate sensor and instrument design, as well as selection of appropriate materials compatible with such a harsh environment, must be taken into account.

• Selective. Sensors should have minimum interference from nonanalyte parameters such as temperature, ionic strength, pressure, and other chemical species not being measured. One route to successful sensor design will rely on the identification of appropriate and selective recognition chemistry.