This paper provides an introduction and overview of the discipline known as "ocean optics". Emphasis is on basic concepts, the optical quantities involved, their measurement, and interconnecting theoretical relationships. Specific topics include radiometric quantities, inherent optical properties, apparent optical properties, measuring the spectral absorption coefficient, effect of bandwidth on Gershun's Law, measuring the volume scattering function, effect of the deep chlorophyll layer, removal of cloud effects from depth profiles of irradiance, and future directions.

The distribution of aerosols in the marine boundary layer can be viewed as a dynamic balance of production, transport and removal processes. The balance of these processes can be represented by a simple mixed-layer model. The vertical distribution of aerosols is dominated by turbulent transport. When mixing is dominated by surface shear or cloudtop cooling (as is typical in mid-latitudes), a single "well-mixed" layer is sufficient to describe the aerosol profile. When scattered cumulus clouds are present (called the "trade wind" or "weak cumulus convection" regime), the well-mixed layer is confined to the region below cloud base. In the region above cloudbase and below cloudtops, strong vertical gradients of aerosol concentration may be observed. A simple parameterization of this gradient is presented.

Much of the optical variability in the upper sea is caused by variations in the biogenous microparticles, which include the phytoplankton, the Chroococoid cyanobaccteria, and detritus. It is instructive to consider two sources of variability in these biogenous particles. First, changes in the absorption cross section of individual cells result from responses of phytoplankton to the ambient nutrient and light fields. These responses are relatively rapid (days) and a mathematical description of the response is presented. Second,changes in the concentration of detrital particles and cells are more difficult to predict and occur on time scales of weeks. A crude hypothetical model is presented which attempts to describe the general response of the plantonic community to light and nutrient fields.

By means of a pair of boresighted and synchronized cameras fitted with orthogonally oriented polarizing filters and carried aboard the Space Shuttle, a large number of polarized images of the earth's surface have been obtained from orbital altitude. Selected pairs of images, both in color and in black and white, have been digitized and computer processed to yield analogous images in each of the three Stokes parameters necessary for characterizing the state of linear polarization of the emergent light. Many of the images show surface properties more distinctly in degree and plane of polarization than in simple intensity alone. However, the maximum information content as well as noise suppression and minimization of atmospheric interference, is achieved by proper combinations of the Stokes parameters. It is believed that these are first, and certainly the most extensive, set of polarized images of the earth ever obtained from space.

In many regions of the ocean, the phytoplankton population dominates both the attenuation and scattering of light. In other regions, non-phytoplankton contributions to the absorption and scattering may change the remote sensing reflectance and thus affect our ability to interpret remotely sensed ocean color. Hence, variations in the composition of both the phytoplankton population and of the non-phytoplankton material in the water can affect the optical properties of the sea. The effects of these contributions to the remote sensing reflectance and the submarine light field are modeled using scattering and absorption measurements of phytoplankton cultures obtained at the Friday Harbor Laboratory of the University of Washington. These measurements are used to develop regional chlorophyll algorithms specific to the summer waters of Puget Sound for the Coastal Zone Color Scanner, Thematic Mapper and future Ocean Color Imager, and their accuracies are compared for high chlorophyll waters with little or no Gelbstoff, but with variable detrital and suspended material.

The downward albedo (irradiance reflectance) r_ and the upward albedo r₊ of a random air-water surface, formed by capillary waves, are computed as a function of lighting conditions and wind speed by Monte Carlo means for incident unpolarized radiant flux. The possibility of multiple scattering of light rays and of ray-shielding of waves by other waves is included in the calculations. The Monte Carlo procedure is used to generate reflected and transmitted glitter patterns as functions of wind speed and sun position. Plots of the Monte Carlo r_± are drawn as functions of wind speed and angle of incidence of light rays. The albedos r_±; are also found for various continuous radiance distributions simulating overcast skies and upwelling submarine light fields just below the air-water surface.

The distribution of irradiance on the sea surface due to an isotropically emitting point source in the water is computed using Monte Carlo techniques for a pure Rayleigh scattering ocean and for a more realistic ocean exhibiting strong forward scattering. The resulting irradiance distributions show at least two regimes. The first is for the source relatively close to the detector, in which the irradiance is more strongly influenced by geometry than by the water's optical properties. The second is for large separations between the source and the detector in which case multiple scattering is dominant. In the latter regime, the irradiance is relatively independent of the source depth and exhibits a near-exponential behavior with distance along the surface. The exponential decay coefficient in this diffusion-like regime is very nearly equal to the decay constant of the asymptotic light field for a homogeneous ocean with the same optical properties.

Remote sensing of water properties using passive, optical systems has been limited by the ability to model underwater radiative transfer for anything other than homogeneous waters. An approximate model, the Singly-scattered Irradiance (SSI) model, has been developed which is designed to describe optically shallow, stratified waters and for which there exists a general analytic solution. Experimental evidence is presented which partially substantiates the adequacy of the approximation of single-scattering for irradiance. Although limited in scope, these preliminary experiments suggest that the SSI model will yield results which are adequate for the purpose of remote sensing of water properties.

A method for modelling light pulse propagation through a multilayer ocean is presented which employs metric analysis using fractals. Fractal analysis is a novel method which, in this application, has the follcmin,g features: 1. nonlinear surface and volume scattering, absorption, reflection, and refraction can be described from empirical data, 2. the optical path length of multiple-collision scattering paths can be computed using the fractal dimension, yielding improved processing speed, and 3. radiative transfer between multiple layers can be described in terms of a convolutional method which is not computationally burdensome. Vertical and temporal scalar irradiance distributions are computed and compared with previously published results. The effect of multiple scattering on beamspread and temporal pulse distortion is discussed in terms of the fractal dimension and irradiance field profiles. Although the model can be quite complex, its computational load is shown to be low.

The results are presented for an experiment in which the full angular width of the solar radiance distribution was measured at varying depths using a Silicon Intensified Target (SIT) underwater camera. The data show that for shallow depths, the width of the solar radiance distribution is strongly affected by the optical thickness of the cloud cover. In this experiment, under clear skies, as the depth increased from 27 to 677 ft, the angular width, full width at half maximum (FWHM), went from 35° to 59°. For very cloudy conditions, the solar radiance appears to be diffuse with an angular width of 62° FWHM even before striking the ocean's surface, and little change in angular width is observed from the surface to maximum depth.

The accurate estimation of the diffuse attenuation coefficient, K(λ,z), over a wide range of space/time scales is essential for the solution of a number of scientific and operational problems. In order to obtain shipboard estimates of K over large areas rapidly, it is necessary to make optical measurements quickly in all environmental conditions. Optical estimates of apparent optical properties under adverse environmental conditions require appropriate corrections to obtain suitable accuracies. We have developed techniques to correct spectral irradiance data for shadowing due to the ship's presence. This analysis makes use of contemporaneous above water spectral irradiance measurements, of a rough atmospheric model for the estimation of global transmittance through the air-water interface as a function of solar zenith angle, and of bio-optical models. The presence of a reasonably homogeneous upper mixed layer is assumed. By comparison with bio-optical model calculations, with Monte Carlo computations and with data obtained under optimum environmental conditions, these techniques provide improved spectral K data. Discussions of the photosynthetically active radiation are also presented. Limiting values of optical properties, including K(λ) and K_par, are explored with respect to the maximum penetration of radiant energy through ocean waters.

A method is developed, and tentatively tested using a single example, for using the different wavelength dependencies of variations in incident daylight and the diffuse attenuation coefficient of seawater to estimate fluctuations in incident irradiance at the surface during profile measurements of spectral irradiance, and to correct the profiles accordingly. Smoothed estimates of downwelling irradiance and its vertical derivative are obtained at 10 m intervals by fitting cubic Hermite polynomials; these values for 6 wavelengths are then combined to estimate the surface irradiance, its rate of change, and the diffuse attenuation coefficient.

A series of model calculations illustrate how the spectral characteristics of an irradiance sensor can affect estimates of the vertical structure of the diffuse attenuation coefficient for downwelling irradiance (K(z,λ)). The effect of a finite spectral response function is observed in regions of the spectrum where large dispersive changes exist in the optical properties. For open ocean environments, this occurs in the orange-red region of the spectrum (λ550nm). Four model calculations are made using different spectral response functions (h(λ)) to determine the effects on the inferred K(z,625nm) profiles related to: 1) a finite bandpass, 2) the "leakage" of a finite amount of blue-green light, 3) spectral tails, and 4) a spectral response function measured with a spectroradiometer. The results of these model calculations indicate that the background irradiance (or equivalently- the blocking level) sampled by the spectroradiometer must be less than 10^-6% (or 8 orders of magnitude) of the irradiance of interest for an accurate estimate of K(z,x). Also, the half power bandwidth of h(λ) should he lOnm or less. The model results are compared to observed vertical profiles of K(z,λ) sampled from the R/P FLIP in the North Pacific Ocean during the Optical Dynamics Experiment (ODEX). These calculations illustrate the possible implications of inappropriate sensor design on the interpretation of observed K(z,λ) profiles.

Spectral irradiance and pigment (porphyrin and carotenoid) measurements were made across strong frontal boundaries of the Gulf Stream in the northwest Atlantic. Upwelling and downwelling irradiance were measured at 2-nanometer (nm) increments across the visible spectrum (400-700 nm). Pigment concentrations were determined by high performance liquid chromatography (HPLC). The irradiance spectra suggest that passive fluorescence occurs in frontal waters. The phytoplankton pigment composition showed significant changes across the frontal region. Pigment changes are reflected in the spectral attenuation coefficients, even at low pigment concentrations measured in Gulf Stream waters. Application of the spectral attenuation coefficient suggests an alternate method for accessing the pigment character in surface waters. The surface pigment distribution aids in our understanding the ocean color variability measured with satellite sensors.

The Laser Troller is an instrument for recording the scattered light intensity at a plurality of scattering angles from single marine microparticles in situ. The breadboard system contains 16-detector elements each of which may be placed in a supporting framework at any of a possible 83 angular locations measured with respect to a polarized incident laser beam. Each detector element consists of an optical collimator attached to the photocathode of a remote photomultiplier tube. Also described are provisions for measuring depolarization, using low power short duty cycle laser sources, making passive measurements of single, bioluminescent phytoplankta, and making measurements at great depth.

Many researchers have measured the beam attenuation coefficient of monochromatic light in seawater in conjunction with the concentration (volume or mass) of suspended material. The ratio of the beam attenuation coefficient to suspended load (known as the specific beam attenuation coefficient) has been shown to vary considerably for different oceanic areas and depths. However, the specific beam attenuation coefficient has also been shown to be a reliable identifier of water masses and tracer of particles in many marine systems. The factors which control the value of the specific beam attenuation coefficient are the size distribution, refractive index distribution and shape of the suspended particles. The theoretical response of the beam attenuation coefficient to changes in the size and refractive index of a particle is well known. Consequently, the response of the beam attenuation coefficient to changes in suspended load may be quantified and predicted and the specific beam attenuation coefficient may provide a high resolution measure of the refractive index and particle size distribution of the suspended load. Calibration curves for the specific beam attenuation coefficient as a function of the particulate refractive index and the slope of the hyperbolic particle size distribution are presented. The curves have been derived for a light wavelength of 660 nm; this is the spectral line most commonly used for optical studies of the marine suspended load. The application of the curves to existent data sets supports their utility and demonstrates the resolution of the optical measurements in defining suspended particulate material. For a known particle load (i.e. approximate bulk refractive index and particle size distribution), measurements of the beam attenuation coefficient may be used to accurately estimate suspended volume. Similarly, measurements of the specific beam attenuation coefficient may be used to determine the particulate bulk refractive index and particle size distribution.

A selected number of our microwave extinction measurement results on preferentially/randomly oriented nonspherical particles are presented. These particles span in size near the first major resonance, possess refractive indexes characteristic of natural aerosols, and have axisymmetry in shape, such as spheroids, cylinders and dumbbells. Subtle dependence of light obscuration on particle orientation is shown in our standardized form of P,Q plots. Such a P,Q plot is a cartesian display of the complex scattering amplitude S(0) versus particle orientation (X,ψ) at scattering angle 0 = 0 and contains detailed information on the single-particle obscuration process in perhaps the most comprehensive form. We also routinely deduce C_ext, the extinction cross section ext' averaged over random particle orientations, from such a P,Q plot. The ensuing, vivid overall pictures of obscuration by a number of randomly oriented nonspherical particles are also depicted in graphs of Q _ext,v versus ρv, showing the volume-equivalent extinction - efficiency averaged over random orientations versus the volume-equivalent phase-shift parameter of each particle. Comparisons with theoretical predictions, where applicable, are also shown in these graphs.

Nonspherical particles have projected areas, averaged over all their orientations, greater than those of equal volume spheres. As a result, the refractive index of such particles may be overestimated when it is determined from light scattering and a size distribution obtained with a Coulter counter or a similar volume-sensitive device. When the refractive index is known, the volume of nonspherical particle may be overestimated when calculated using a size distribution obtained from light scattering. This can be avoided if the nonsphericity of the particles (ratio of the projected area of the particle averaged over all its orientations to the projected area of an equal volume sphere) is known. A method for the determination of the nonsphericity of suspended particles is proposed. It is based on a relationship between the size distributions of the particles obtained using a volume-sensitive and projected-area-sensitive particle counter. The nonsphericity of several kinds of particles, including coastal marine particles was estimated using this method. The values obtained agreed with those determined using a SEM and an optical microscope.

Characterization of the polarization properties of light scattered from biologically-derived marine particles can be used to understand the effect of light propagated through seawater. To interpret scattering measurements from oceanic samples, it is important to investigate well-characterized marine organisms under controlled conditions. In this study, light scattering from very small, nearly spherical marine plankton, Chlorella, were investigated using a polarization-modulation technique originally developed by Hunt. This technique permits the determination of the elements of the scattering matrix that completely describe the intensity and polarization effects induced by the scattering system. The total scattered intensity and the value of the normalized matrix elements were measured as a function of angle at wavelengths of 442 and 633 nm. The scattering intensity profile was interpreted with the help of a Rayleigh-Debye theory using a spherical shell model. It was found that the Chlorella exhibited scattering behavior approximating that predicted by the Rayleigh-Debye theory. However, consistent deviations from the factored Rayleigh matrix were observed. These results and their implications for light scattering studies of marine waters are discussed.

An Optical method for instantaneous measurement of suspended particle size distribution shows promise for the study of boundary layer dynamics in the ocean, where suspended particles affect boundary layer flow through stratification. The method is based on observing the an-gular distribution of scattered light from a sample, and inversion to produce the suspended particle size distribution, n(x). In a previous paper, an analytic inversion based on the Fraunhoffer approximation was examined. The function 0³I(0) was found approximately to be related by a Fourier transform to xn(x), with the result that elementary signal processing concepts apply. In this work, the issues of uniqueness and stabilty of the inversion are considered. Instability of inversion for small particles is observed, and has the result that while matrix inversion algorithms show promise, those which manipulate small eigenvalues distort inversions for small sizes. Uniqueness in terms of sampling is revisited, yielding a refined Nyquist sampling criterion. The validity of these results, derived from approximate diffraction theory, is demonstrated for "exact" Mie theory.

The optical absorption coefficient of a sea water sample was measured at 590 nm by the pulsed photothermal deflection technique. The absorption of the sea water sample was 1.22 x 10^-3 cm^-1. Corresponding measurements on samples from a standard source of distilled water give an absorption coefficient of 6.44 x 10^-4 cm^-1. Simple calculations are shown for the thermal gradients and probe beam deflections seen in these types of studies.

Spectral radiometric, scattering, and absorption measurements have classically been accomplished using grating spectrometers or fixed passband optical filters. Recently, holographic optical elements and linear-geometry detectors have gained popularity in spectrometric instrumentation, thereby eliminating the need for movable mechanical components. Other mechanically-fixed methods of spectral filtering rely upon electro- or acousto-optic interaction in solids or liquids. A system useful for ocean-optical measurements is described which employs an acousto-optic filter under microcomputer control to achieve wavelength tunability from 400-700 nm. This approach allows construction of low-power instrumentation having rapid and programmable control. In addition, output beam collimation is maintained to facilitate measurement over long optical paths. The physical principle of operation is described, various trade-offs discussed, and results presented for seawater containing phytoplankton.

Results of an experimental study of ship shadow effects are presented. Spectral upwelling and downwelling irradiance, and upwelling radiance were measured at 2 distances from the ship on an overcast day and sunny day. On the clear sky day it was found that while the apparent properties and derived properties of reflectance showed only negligible affects, the derived diffuse attenuation coefficients varied measurably. On the overcast day the apparent properties varied, with Ed showing the most variation. All of the derived properties on this day exhibited large variations (on the order of 40%) between casts at 0 and 9 meters.

In a homogeneous ocean that both scatters and absorbs the radiance decreases with depth and the angular dependence of the radiance becomes independent of depth and of the incident distribution at the surface. In the diffusion region the asymptotic radiance distribution is only dependent on the inherent properties of the medium including the scattering phase function. Under these conditions an exact integral equation can be derived for the asymptotic radiance. A numerical calculation of the asymptotic radiance was made with Lobatto quadrature resulting in a precise estimate of the diffuse attenuation coefficient for selected values of the single scattering albedo. Calculations were made using estimated single scattering phase functions derived from scattering measurements made for a wide variety of marine and freshwater water types. A two parameter empirical expression was derived from these model calculations relating the diffuse attenuation coefficient and the single scattering albedo. Predictions are made over the entire range of single scattering albedos and are compared to those given by other investigators. The predictability of this relationship and the influence of the scattering phase function are evaluated for each of the scattering phase functions examined. Individual derived relationships are able to predict the diffusion exponent with RMS errors of less than one percent. The overall variation in determining the two parameters is approximately 3 and 18 percent using samples which varied optically from very clear waters of Sargasso Sea to the turbid waters of Lake Erie.

In August of 1985 a programmable multispectral imager (FLI for Fluorescence Line Imager) was flown along with a Inertial Navigation System over the Red Bay region of Lake Huron. The objective of the mission was to collect digital imagery which could be processed into an accurate geometrically correct bathymetric map. The results of the geometric correction processing were found to be accurate to the output pixel level (5 x 5 meters). Sampled profiles of the geometrically corrected image showed systematic shallow bias errors of less than 1.6 meters for depths less than 6 meters and 0.6 meters for depth less than 4 meters.

A hemispherical (27) light sensor is described, which is of simple mechanical construction and manufactured from Teflon. The logarithmic amplifier uses only 2 devices yet gives 6 or 8 orders of magnitude of linearity of light detection when used with common silicon photodiode detectors. The reduction of calibration uncertainties and the long term stability of calibration are discussed. Measurements of photosynthetically active radiation (400-700 nm) and narrow band radiation in the visible spectral region are presented for sensor measurements made over the side of a stationary research vessel and attached to the Undulating Oceanographic Recorder for various deployments in the western English Channel in the summer of 1984.

Given the problems inherent in electrical sensing in the oceans and the advancing technology of fiber-optics, an optical sensing technique has been developed that measures the refractive index of different materials by wavelength modulation and relates these values to oceanographic parameters such as temperature, pressure, salinity, and density. The refractometer herein described is of innovative design and is capable of stable and remote operation.

The Undulating Oceanographic Recorder (UOR) Mark 2 is a self-contained, depth-undulating, oceanographic sampler, which can be towed on an unfaired steel cable at speeds from 3.5 to 13.5 m.s^-1 and is independent of the vessel for any services other than towing facilities. The UOR carries a suite of electronic sensors for temperature, depth, chlorophyll concentration and downwelling and upwelling solar radiant energy (broad band 400-700 nm and narrow band 445, 520, 550 and 670 nm). Sensor measurements are recorded in situ by a digital tape recorder. Data are presented for measurements in the western English Channel, in the summer of 1984. The data provide attenuation coefficients for sea-water and chlorophyll at each wavelength and sub-surface reflectance ratios at each wavelength from which algorithms are developed, relating reflectance ratios and chlorophyll concentrations. The results are in close agreement with values reported by other workers, using conventional sampling techniques of optical oceanography.

Optically sea ice is a complex material with an intricate and highly variable structure which includes brine pockets, air bubbles, brine channels and internal platelet boundaries. Large variations in the optical properties of the surface layer can occur on horizontal scales of only a few meters, complicating efforts to quantify larger scale interactions between shortwave radiation and the ice-ocean system. Radiative transfer in sea ice is dominated at visible wavelengths by scattering rather than absorption. Because scattering in the ice is essentially independent of wavelength, spectral variations in the optical properties are primarily the result of differences in absorption. Observations show that albedos are particularly sensitive to the presence of liquid water in the surface layers, the effect being most pronounced at wavelengths above 600 nm. Albedos and extinction coefficients in the ice vary inversely with brine volume, and thus temperature. Below the eutectic point, precipitation of solid salts causes a sharp increase in scattering and corresponding increases in albedo and absorption. Biological activity in natural sea ice often affects light transmission and absorption, particularly in coastal regions and in the Southern Ocean. Phase function measurements indicate that the scattering distribution in sea ice is only weakly dependent on wavelength and brine volume.

Radiative transfer models of sea ice applied to date range from a simple Bouguer-Lambert representation for net downwelling irradiance through 16 stream models which takes into account detailed variations in ice microstructure. Both sea ice and snow are strongly multiple scattering media with single scattering albedos well above 0.9 through the visible and into the near infrared. Parameter studies indicate that the optical properties of sea ice are controlled by the density of brine and vapor inclusions which in general undergo substantial seasonal changes. Melting and brine drainage are the principal causes of these variations. For ice below -5°C, temperature effects are relatively weak unless the T_ice drops below the eutectic point. The optical properties of snow depend primarily on grain size, the bulk density, and the presence of impurities such as carbon soot. The theoretical models appear to be able to reproduce observations quite well and have revealed that soot or dust contamination of snow appears to be prevalent even in the Arctic.

Recent interest in the Arctic has led to the application of underwater irradiance measurement techniques to the determination of sea ice optical properties. Both a theoretical model of the bulk optical properties and in-situ instrumentation for the measurement of these properties was developed at the University of Washington in support of field measurements in the Marginal Ice Zone (MIZ). In order to extend the existing data base to higher latitudes and evaluate the technique for making optical measurements in this environment, the authors participated in a short exercise in the Arctic during April 1985, using University of Washington developed instruments. During a period of three weeks, numerous measurements of the spectral attenuation coefficient, k_ice, and albedo, r_ice, were made in refrozen leads of first year ice and new sea ice near the coordinates 86 degrees north and 88 degrees west. Ice thicknesses ranged from 30 to 300 cm. In addition to the optical properties, cores were taken to photograph ice crystal structure and measure the temperature profile. A description of the ice optical property measurement techniques and instrumentation is given. In addition to field measurements of the bulk optical properties of young ice at very low temperatures, data was obtained on the optical properties of the near surface water found in the higher Arctic latitudes during April. Recommendations are made for further field measurements and the use of models in data interpretation.

Optical properties of sea ice depend to a greater or lesser extent on its crystalline properties and on the size, shape, and distribution of brine inclusions systematically trapped in the ice crystals. Here we demonstrate the use of polarized light techniques to examine the internal structure of sea ice. Using both naturally occurring and laboratory simulated sea ice we show how the crystalline and salinity components originate including discussion of the mechanisms by which first-year ice desalinates and recrystallizes into multi-year ice exhibiting optical properties significantly different from those of first-year ice.

There exist unique situations in oceanography where the small space and time scale measurements possible with laser velocimetry are required. In this paper, we review some of the aspects of instrument development for applications in these situations. Aside from the obvious field ruggedness requirements, a number of other factors must be considered; for example, battery consumption per signal photoelectron, platforms, optical coatings, pressure windows and related birefringent considerations, velocity profiling, appropriate signal processors and automatic control including protection of photodetectors from accidental overexposure. We include a discussion of the choices made for our two systems and present example data obtained in the field.

The reflection of optical energy from the air/sea interface and by the volume backscattering immediately below involves a number of complex phenomena which have critical effects on the design and performance of airborne laser hydrography systems. Both of these reflections can be considered as "surface" return, because when the interface return is weaker than the volume return, the latter will be detected, although at a somewhat biased location, as found during lidar/sonar depth measurement intercomparisons. The character of the surface return depends on the ratio of the peak volume backscatter power and the peak interface reflection power. Analytic expressions are derived for mean values of these quantities. The functionalities of the volume-to-interface peak power ratio on wind speed and direction, off-nadir beam incidence angle, and water clarity parameters are examined to determine the parameter ranges for which the a priori origin of surface returns is uncertain. Depth measurement error magnitudes are calculated for the case of a volume return being mistaken for an interface return. The error model sucessfully predicts the shoal bias observed in field data. Potential methods for reducing this error are discussed. An expression for the temporal profile of a volume backscatter return is presented, and a potential method of estimating a key water clarity parameter from the airborne data is reported.

Several domains of underwater visibility are illustrated using a computer program to simu-late underwater image formation. The basic limitations of conventional underwater optical imaging are described. In addition, several image processing techniques have been applied to underwater video images. The results of applying grey level enhancement and spatial filtering are illustrated.

A general visibility theory valid in inhomogeneous hydrosols is presented and experimentally verified for some special cases. The field measurements used in the experimental verification were obtained from both oligotrophic and eutrophic regions in the sea and comprise of a fairly large Danish and French data set consisting of spectral downward and upward irradiances, downward quanta irradiance, spectral light transmission, spectral radiance distributions and finally, spectral Secchi disc measurements.

A subset of the Optical Dynamics Experiment (ODEX) data set taken from the R/P FLIP in the eastern North Pacific on November 8, 1982 is used to examine the relatiohips among physical and bio-optical parameters. Rapid profiling was done with a CTD equipped with a beam transmissometer and an in situ fluorometer. Intense thermohaline structures are observed to vary on time scales as short as 15 min. Interestingly, the temporal and vertical spatial variability in the physical parameters are often well-correlated with the corresponding variability in the optical parameters. In addition, the extent and intensity of the fluorescence maximum region appears to be modulated by a low frequency internal wave.

Vertical profiles of diffuse attenuation coefficients K(λ) have been determined from measurements of spectral scalar irradiance profiles in the central North Pacific. K(λ) in the range of 410-488 nm show a pronounced maximum at the depth of deep chlorophyll maximum. This vertical pattern prevails in clear oceanic waters in which particle concentration shows a minimum vertical change in the euphotic zone. Thus, sunlight attenuation in 410-488 nm appears to be dominated by absorption by chlorophyll pigments. The deep fluorescence maximum was observed extensively, but it was due to higher pigment concentration rather than higher cell population. Chlorophyll pigments per cell increase with depth down to the fluorescence maximum, suggesting light limited conditions start from the fluorescence maximum. Above the fluorescence maximum, nutrient limited conditions prevailed according to earlier observations. In this upper layer the biomass and chlorophyll are not correlated; however, changes in the upper layer influence the depth of the fluorescence maximum. Observed depths of the deep chlorophyll maximum in the central North Pacific are deeper than those of winter mixed layer. Thus, temporal variations in bio-optical properties cannot be explained by variations of the mixed layer. An alternative hypothesis, nutrient flux associated with gyre circulation, is suggested.

The accuracy of chlorophyll estimates by ocean color algorithms is affected by the variability of particulate attenuation; the presence of dissolved organic matter and the non-linear inverse relationship between the attenuation coefficient, K, and chlorophyll. Data collected during the Warm Core Rings Program were used to model the downwelling light field and determine the impact of these errors. A possible mechanism for the non-linearity of K and chlorophyll is suggested, namely, that changing substrate from nitrate-nitrogen to ammonium causes enhanced blue absorption by photosynthetic phytoplankton in oligotrophic surface waters.

During a 17 month period (November 1978 - March 1980) phytoplankton pigment concentrations were remotely sensed in the northern Gulf of Mexico using the Coastal Zone Color Scanner. A total of 29 CZCS orbits were processed into pigment (chlorophyll a + phaeopigments) images and then geometrically warped to a mercator projection. A correction factor of 1.67 was applied to the pigment concentrations to correct for the tendency of the standard fluorometric method to underestimate chlorophyll a concentrations. The spatial and temporal distributions of pigment fronts were quite variable during this time series. Constant features observed throughout the pigment imagery were the entrainment of coastal waters offshore. The most extensive entrainments occurred during intrusions of the Loop Current. For the 17 month survey, the mean HPLC-corrected pigment concentration was 3.30 ± 1.45 mg^-3.

The estimation of oceanic primary production on a global scale is the focus of efforts in remote sensing using the Coastal Zone Color Scanner (CZCS). The goal of this research is to provide a measure of the primary production using only satellite data for the estimate. This estimate requires the measurement of surface pigments (chlorophyll a + phaeophytin a) using the CZCS, an estimate of the sea-surface temperature using the AVHRR and determination of the incident solar irradiance using GOES imagery. In this paper, we describe a model of primary production based upon the responses of phytoplankton to differing light and nutrient fields. This model includes the effects on production of variations in surface pigment concentration, the mixed layer depth and the dependence on the incident solar irradiance. The model has been tested using in situ data provided by the Southern California Bight Studies (Eppley, et al., 1979), California Cooperative Fisheries Investigations (CalCOFI), Organization of Persistent Upwelling Structures (J.B. Soolloo in OPUS Data Report) and other data sets. A synoptic measure of the distribution of surface pigments is derived from the West Coast Chlorophyll and Temperature Time Series (West Coast Time Series Advisory Group, 1985). The features and behavior of the model will be presented together with the results of the model verification.

Nimbus-7 Coastal Zone Color Scanner (CZCS) data were used to compute spatial distribution of surface phytoplankton pigments in coastal and offshore waters near Iceland in April 1979. The standard CZCS processing algorithms were applied and a limited evaluation of their accuracy was conducted. This evaluation utilized in situ measurements from a cruise in the area and the clear water radiance concept. Initial results appear to confirm that the CZCS processing algorithms used at these moderately high latitudes exhibit accuracy similar to the validation studies conducted in the mid Atlantic bight.

The technique of remote sensing chlorophyll pigment concentrations by monitoring sunlight induced fluorescence at 685 nm is an alternative method of surveying the health and productivity of phytoplankton in the oceans . This method measures induced fluorescence from chlorophyll pigments whereas the conventional absorption method measures the reflectance ratio of blue/green channels as an indication of chlorophyll abundance in the water. It should be noted that the method is also differentiated from the laser induced fluorescence technique in which a laser is used to induce the fluorescence, whereas in this method a passive spectroradiometer or imager is used to monitor the occurrence of natural fluorescence under sunlight.

A semianalytic Monte Carlo model has been used to simulate laser fluorosensor signals returned from subsurface distributions of chlorophyll. This study assumes the only constituent of the ocean medium is the common coastal zone dinoflagellate Prorocentrum minimum. The concentration is represented by Gaussian distributions in which the location of the distribution maximum and the standard deviation are variable. Most of the qualitative features observed in the fluorescence signal for total chlorophyll concentrations up to 1.0 μg/liter can be accounted for with a simple analytic solution assuming a rectangular chlorophyll distribution function.