Watercolors in the Coastal Zone: What Can We See?

Hydrological optics has a rich history, play­ ing a significant role in physical, chemical, and biological oceanography. The success over the last 30 years has provided oceanog­ raphers with a non-invasive means to study regional and global scale physical, chemical, and biological processes (Figure 1). The abil­ ity to map the color of the world’s oceans has been used to estimate global ocean pro­ ductivity (Longhurst et al., 1995; Platt and Sathyendranath, 1988; Sathyendranath et al., 1989; Behrenfeld and Falkowski, 1997), aid in understanding radiant heating processes (Ohlman et al., 2000), assist in delineating oceanic biotic provinces (Longhurst, 1998), and document regional shelf break frontal processes (Ryan et al., 1999a, 1999b). The scientific utility of mapping ocean color led to wide community support that has result­ ed in three generations of satellites launched by the United States, complemented by an international constellation of ocean color satellites from Europe, Japan, China, and India. The utility of remote sensing results from algorithms that use satellite-measured re flectance to estimate the concentration of biogeochemically signifi cant constituents. These algorithms were developed for opti­ cally simple waters where the optical prop­ erties of the ocean are largely defi ned by phytoplankton and water molecules (Fig­ ure 2; see article by Mobley et al., this issue). The spectral properties of water (Figure 2B) and phytoplankton are distinct. Increasing the concentration of phytoplankton (Figure 2C) in a volume of water selectively absorbs blue wavelengths of light, effectively green­ ing the water reflectance in a predictable fashion. This greening allows empirical relationships to be derived that estimate chlorophyll a concentrations from the re flectance ratio of blue-to-green wavelengths of light. Many times, however, the optical signature of the ocean reflects the presence of materials other than phytoplankton and water molecules. The resulting complexity can directly influence the interpretation of what you see using satellite refl ectance sig­ nals. A good example is the usually optically simple, high nutrient-low chlorophyll zones (HNLC). It has been proposed that deposi­ tion of atmospheric dust is a signifi cant fac­ tor regulating overall productivity in HNLC zones (Martin, 1990; Prospero and Nees, 1986). Yet, if present in signifi cant concen­ trations, the optical signature of the dust can compromise the empirical satellite algo­ rithms (Moulin et al., 2001). The presence of significant submicron dust particles, which


THE ROLE OF OPTICS IN OCEANOGR APHY
Hydrological optics has a rich history, play ing a significant role in physical, chemical, and biological oceanography.The success over the last 30 years has provided oceanog raphers with a non-invasive means to study regional and global scale physical, chemical, and biological processes (Figure 1).The abil ity to map the color of the world's oceans has been used to estimate global ocean pro ductivity (Longhurst et al., 1995;Platt and Sathyendranath, 1988;Sathyendranath et al., 1989;Behrenfeld and Falkowski, 1997), aid in understanding radiant heating processes (Ohlman et al., 2000), assist in delineating oceanic biotic provinces (Longhurst, 1998), and document regional shelf break frontal processes (Ryan et al., 1999a(Ryan et al., , 1999b)).The scientific utility of mapping ocean color led to wide community support that has result ed in three generations of satellites launched by the United States, complemented by an international constellation of ocean color satellites from Europe, Japan, China, and India.
The utility of remote sensing results from algorithms that use satellite-measured re flectance to estimate the concentration of biogeochemically signifi cant constituents.
These algorithms were developed for opti cally simple waters where the optical prop erties of the ocean are largely defi ned by phytoplankton and water molecules (Fig ure 2; see article by Mobley et al., this issue).
The spectral properties of water (Figure 2B) zones (Martin, 1990;Prospero and Nees, 1986).Yet, if present in signifi cant concen trations, the optical signature of the dust can compromise the empirical satellite algo rithms (Moulin et al., 2001).The presence of significant submicron dust particles, which  (LEO) electro-optic fiber optic cabled seafloor node (Oliver et al., 2004).Rapid changes in CDOM concentration are associated with the passage of storms and a large plume of Hudson River water.
can remain in the water column for months to three using standard ocean color satellite the perceived color change reflects the dust (Claustre et al., 2002), influences the relaalgorithms (Claustre et al., 2002).Therefore, itself.The interpretation of ocean-color im tive reflectance of the blue and green wavewhen interpreting ocean-color imagery, one agery is even more difficult in optically com lengths of light and can result in an overmust ask whether the iron-rich dust leads to plex waters where many different optically estimate of chlorophyll a by a factor of two a true increase in phytoplankton or whether significant constituents infl uence remotesensing refl ectance.(Figure 3).One such effort, the Hyperspec tral Coupled Ocean Dynamics Experiment (HyCODE), has integrated these instru ments into an ocean observatory (Glenn and Schofield, 2003;Schofield et al., 2003), en abling bio-optical adaptive sampling of the Mid-Atlantic Bight (Schofield et al., 2003).

Oceanography
Given the desire to develop coastal remotesensing applications, HyCODE focused on a wide range of optical issues that are high lighted in this issue of Oceanography.As an introduction to those efforts, this manu script reviews some of the major optically active constituents that underlie the spectral variability of remote-sensing refl ectance in coastal waters.
Understanding the spectral variability in ocean-color reflectance is key to using remote-sensing approaches, so our fi rst need is to understand what underlies refl ectance.
Remote-sensing reflectance is the abovewater ocean color and is defined as the ratio where G is a relatively constant param eter dependent on the angular distribution of the light field and the volume scatter ing coefficient (Gordon, 1975;Morel and Prieur, 1977).Given that the magnitude of   systems have utility for ocean applications is the compounds that dominate ocean color.

WHAT GIVES COASTAL WATER
an open area of research).Increased spectral This is often accomplished by collecting dis-

ITS COLOR?
resolution expands the potential to improve crete and in situ measurements.The fi rst ef-CDOM refers to organic matter that can approaches that can invert the measured forts have focused on characterizing the relanominally pass through a 0.2 micron fi lreflectance into its constituent components, tive importance of the dissolved and particter and can be detected optically, as not which have distinct absorption and scatter-ulate constituents, including phytoplankton, all organic compounds absorb light.High ing properties.detritus, and sediments, which dominate the concentrations of CDOM decrease light re-Given the desire to invert bulk refl ectance particulate material; and CDOM, small bacfl ectance dramatically because its spectral into its constituent components, the ocean teria, colloidal material, and viruses, which absorption can be high (Figure 2B).CDOM optics community has focused on defi ning dominate the dissolved phases.is often predominantly composed of humic the scattering and absorption properties of Oceanography June 2004 and fulvic acids, but can also include small colloidal material.Generally, CDOM's larg est impact is on absorption, but some colloi dal material can contribute to the backscat tering of light.CDOM sources include cel lular exudation/lysis/defecation (Kalle, 1966;Bricaud et al., 1981;Guixa-Boixeru et al., 1999), resuspension from sediments (Chen, 1999;Komada et al., 2002) and humic/fulvic acids from rivers and terrestrial watersheds (Blough et al., 1993;Vodacek et al., 1997).
Satellite algorithms for CDOM absorption are being developed and offer in the near fu ture the possibility of mapping CDOM con centration.These algorithms are based on empirical models that relate CDOM absorp tion to the ratio of green to red wavelengths of refl ectance.
Absorption properties can also be used to describe CDOM composition.Spectral absorption of CDOM decreases exponen tially with increasing wavelength (Figure 2B).A low CDOM slope is generally inter preted as freshly produced material, which is then degraded through either photo-oxi dation (Mopper et al., 1991;Kouassi and Zika, 1992;Nelson et al., 1998) or microbial activity (Whitehead, 1996).Spectral slope increases as the material is chemically modi fied (Rashid, 1985), and over monthly time scales, CDOM becomes increasingly refrac tory (Nelson et al., 1998).Mechanistic inter pretation of the age and source of CDOM from the spectral slope is difficult due to the complexity of the degradation process, which is described by at least two rate con stants that span days to weeks (Twardowski and Donaghay, 2002).Despite these uncer tainties, CDOM slope has been effectively associated with specific water masses in the Baltic Sea (Højerslev et al., 1996), Gulf of Mexico (Carder et al., 1989), Caribbean (Blough et al., 1993), Mid-Atlantic Bight (Vodacek et al., 1997), and Sargasso Sea (Nelson et al., 1998).Developing methods to characterize CDOM composition from space does not exist, but as the spectral resolu tion of satellite systems increases, the ocean optics community will undoubtedly explore the potential for new algorithms.
Depending on the filter, some organic particles are often included in the CDOM fraction.The most notable particles are vi ruses and bacteria.Viruses are very small (~100 nm) and are not believed to contrib ute significantly to the overall absorption and scattering of light.Bacteria are also abundant in the water column and con tribute significantly to the overall scattering properties (Morel and Ahn, 1990)  issue; also Schofield et al., 1999;Kirkpatrick et al., 2000;Millie et al., 2002); however, this capability has yet to be robustly achieved  ramaniam et al., 1999) or minerogenic ar mor (Iglesias-Rodriguez et al., 2002), or the physiological state of the cell (Stramski, and Reynolds, 1993;Reynolds et al., 1997).For remote-sensing applications, phytoplankton have a high scattering signal, but generally only a small proportion is backscattered, therefore phytoplankton absorption domi nates their contribution to refl ectance.
Oceanography June 2004  (Glenn and Schofield, 2003) where scientists can sit and drink coffee while all the data are delivered in near real-time.This capability allows for adaptive sampling (Schofield et al., 2003).

Oceanography June 2004
Detrital particles represent nonliving organic matter including fecal pellets, cell fragments, large colloids, and marine snow.
An exponentially decreasing slope describes detrital spectral absorption, but the slope is typically less steep than for CDOM (Figure 2B).The relative backscatter of detritus can be high compared to that of phytoplankton (Stramski and Kiefer, 1991), but of all the optical constituents, detrital particles remain relatively understudied.This lack of infor mation is problematic in coastal-water stud ies where detrital absorption can represent up to 30 percent of the blue light absorption signal (Schofield et al., In press).

WHERE DO WE GO FROM HERE?
Coastal waters are complex, but recent advances in optical instrumentation al low the oceanographer to decipher this complex soup, as highlighted in the other manuscripts in this issue.Optical data allow ocean-observing networks to serve the needs of chemists and biologists by providing data over ecologically relevant scales.It is our hope that the wider community will adopt these measurements and approaches so that one day they become as standard as a Con ductivity-Temperature-Depth (CTD) sensor in the oceanographer's tool box.
S C A R S C H O F I E L D , R O B E R T A .A R N O N E , W . P A U L B I S S E T T , T O M M Y D .D I C K E Y, C U R T I S S O .D A V I S , Z O E F I N K E L , M A T T H E W O L I V E R , A N D M A R K A .M O L I N E and phytoplankton are distinct.Increasing the concentration of phytoplankton (Figure 2C) in a volume of water selectively absorbs blue wavelengths of light, effectively green ing the water reflectance in a predictable fashion.This greening allows empirical relationships to be derived that estimate chlorophyll a concentrations from the re flectance ratio of blue-to-green wavelengths of light.Many times, however, the optical signature of the ocean reflects the presence of materials other than phytoplankton and water molecules.The resulting complexity can directly influence the interpretation of what you see using satellite refl ectance sig nals.A good example is the usually optically simple, high nutrient-low chlorophyll zones (HNLC).It has been proposed that deposi tion of atmospheric dust is a signifi cant fac tor regulating overall productivity in HNLC This article has been published in Oceanography, Volume 17, Number 2, a quarterly journal of The Oceanography Society.Copyright 2003 by The Oceanography Society.All rights reserved.Reproduction of any portion of this arti cle by photocopy machine, reposting, or other means without prior authorization of The Oceanography Society is strictly prohibited.Send all correspondence to: info@tos.orgor 5912 LeMay Road, Rockville, MD 20851-2326, USA.

Figure 1 .
Figure 1.The time and space scale variability in ocean color.(A) An annual global chlorophyll a map measured using SeaWIFS (from http://seawifs.gsfc.nasa.gov/SEAWIFS.html).(B) Backscattering measured in summer 2001 in the Mid-Atlantic Bight using SeaWIFS.(C) An enlarged section in panel B focusing on the backscattering signal derived from the SeaWIFS observations; the satellite's 1-kilometer pixel is clearly visible and illustrates the features in the coastal ocean that are poorly resolved.It should be noted that some of newer ocean-color satellites have spatial resolutions down to 250 km.Ocean-color satellites with 30-meter resolution are proposed.( D) An enlargement from panel C showing backscatter measured by aircraft.Note the features clearly visible in the aircraft imagery that are missed with the standard 1-km pixels in the satellite imagery.(E) The visible image viewed by aircraft, with resolution on the order of tens of meters, showing the dramatic color change associated with crossing an upwelling front in the Mid-Atlantic Bight.The visible "greening" of the water is associated with enhanced blue light absorption.This color shift underlies empirical algorithms for ocean color remote sensing.(F) Time series of CDOM absorption, estimated from inverting bulk absorption measured with an ac-9 mounted on the Long term Ecosystem Observatory very often optically com plex.In nearshore continental shelf waters, organic detritus and colored dissolved organic matter (CDOM) are often present in quantities sufficient to obscure the plant biomass signal because they infl uence the blue-to-green reflectance ratio (Figure2B).Additionally, the presence of highly scat tering inorganic particles and photons re flected off the seafl oor (see Limnology and Oceanography, 48: 323-585, Figure 2D) can complicate the quantitative interpretation of the satellite imagery.Coastal waters are also characterized by numerous distinct frontal boundaries.Large changes in the in situ con centration of optically active constituents are often observed across these frontal boundar ies (Figure 1).The spatial variability of these frontal features is often on scales of kilome ters to meters (Figure 1D) and is diffi cult to resolve with a standard 1-km satellite pixel.How this spatial variability within a pixel influences the observed satellite signal is an open question (see Bissett et al., this issue).Efforts to decipher this complex coastal optical soup often use in situ measurements to characterize observed refl ectance spectra.This has become much easier in recent years due to advances in instrumentation that measure the in situ inherent optical proper ties [absorption (a), backscattered (b b ), and attenuation (c), note c -a = scattering (b)] of the upward flux of light to downward flux of light incident on the ocean surface.Atmospheric effects aside (which represent ~95 percent of the actual satellite signal), reflectance is a function of both the spectral backward scattered light [b b (λ)] and spectral absorption [a(λ)] within the water column.The relationship between spectral refl ec tance [R(λ)] and the inherent optical prop erties can reasonably be described as b the inherent optical properties represents the linear combination of all optically ac tive constituents, the color of the refl ected light integrates the spectral absorption and backscattering properties for all the materi als present.For example, in coastal waters the total absorption might be described by the absorption of phytoplankton, detritus, sediment, water, and CDOM.The net effect is that the first-order factor determining the spectral shape and total amount of refl ec tance is the concentration of absorbing and scattering constituents present; however, it should be emphasized that the absorp tion and scattering efficiencies of different constituents vary dramatically due to their specific molecular properties.For oceancolor remote sensing, backscatter is a source of photons from the ocean to the satellite.Backscattered light is a small proportion of the total scattered light, and the relative amount of backscattered light is dependent on the type and size of the material pres ent in the ocean.In contrast, absorption is a sink for photons.Absorption is very high in aquatic systems due to H 2 O molecules that are extremely effective at absorbing red light and phytoplankton that are effective at absorbing blue light.The net result is that reflectance signal over the ocean is low (Fig ures 2F and 2G offer an opportunity to com pare the scales of reflectance for terrestrial leaves versus that of coastal waters).This presents the remote-sensing engi neer the difficult task of designing sensi tive sensors that will not become saturated if the signal is high, such as is the case on a windy day when white caps and the presence of air bubbles can lead to enhanced light scatter.The available satellites have varying degrees of spectral resolution ranging from five spectral bands to hyperspectral sys tems (often designed for terrestrial remote sensing,<http://eo1.

Figure 2 .
Figure 2. Spectral signatures that dominate reflectance in coastal zones.(A) A visible image taken from an aircraft spanning nearshore coastal waters over an urbanized area into an estuary in New Jersey.The letters illustrate different optical zones in which different optical constituents dominate the reflectance perceived from the aircraft.(B) The relative spectral absorption of water, colored dissolved organic matter (CDOM), and detritus.Absorption of water in the red wavelengths is orders of magnitude higher than CDOM and detrital particles.(C) Relative absorption of diff erent phytoplankton species (Johnsen et al., 1994) illustrating spectral variability due to different phytoplankton pigments.(D) Relative spectral backscattering associated with inorganic particles (re drawn from Babin et al., 2003) for different particle sizes with a constant refractive index.(E) Relative spectral backscattering associated with phytoplankton (redrawn from Babin et al., 2003) for two different particle sizes with a constant refractive index.(F) Reflectance spectra for a healthy and dry leaf.Note scale for the y-axis.(G) Reflectance for nearshore and offshore waters of the Mid-Atlantic Bight.Columns along the x-axis indicate the spectral bands measured by SeaWIFS.
Figure 2A).Phytoplankton represent a major absorb ing constituent in the world's oceans due to the presence of photosynthetic and nonphotosynthetic pigments.The absorption variability associated with the diversity of phytoplankton pigmentation impacts the blue and green light reflectance (Figure 2C).In coastal waters, all major spectral classes of phytoplankton (chlorophyll c-contain ing, phycobilin-containing, and chlorophyll using presently available remote-sensing techniques.This remains an active area of research.Phytoplankton contribute signifi cantly to the total scattering of the light.Total scat tering is, to first order, regulated by biomass; however, the efficiency with which indi vidual phytoplankton cells contribute to scattering can vary with phytoplankton size and refractive index.Some of the changes in refractive indices are due to phenotypic features, such as internal air pockets (Sub

Figure 3 .
Figure 3.Some examples of now-robust off-the-shelf optical technology during the HyCODE effort that are currently available to oceanographers.(A) A Webb glider (http://www.webbresearch.com/slocum.htm)outfitted with an attenuation meter and two backscattering sensors embedded in the belly of the autonomous vehicle.(B) Another Webb glider outfitted with a 2-wavelength backscatter sensor and a fluorometer.(C) A robotic-profiling optical mooring outfitted with absorption-attenuations meters (Wetlabs, Inc.), bioluminescence bathyphotometers, forward-scatter ing sensors (Sequoia Instruments), and backscatter sensors (HOBI labs and Wetlabs, Inc.).(D) A biplane, one of the world's slowest, outfitted with a hyperspectral, remote-sensing reflectance sen sor.The slow speed is ideal for allowing detailed calibration studies.(E) A radiometer (Satlantic) outfitted with a copper shutter.The copper shutters have been demonstrated to be highly eff ective at minimizing biofouling problems.(F) An optical mooring being deployed by the Ocean Physics Laboratory.(G) Living the dream in the COOL room(Glenn and Schofield, 2003) where scientists gsfc.nasa.gov/miscPages/home.html>, the degree with which these Oceanography June 2004