APRIL 2003 LEE AND LEWIS 563 A New Method for the Measurement of the Optical Volume Scattering Function in the Upper Ocean MICHAEL E. LEE Optical Oceanography Laboratory, Marine Hydrophysical Institute, National Ukrainian Academy of Science, Sevastopol, Crimea, Ukraine MARLON R. LEWIS Department of Oceanography, Dalhousie University, Halifax, Nova Scotia, Canada (Manuscript received 23 January 2002, in ®nal form 29 August 2002) ABSTRACT A new method to measure the optical volume scattering function (VSF) of seawater is presented. The VSF is a fundamental property used in the calculation of radiative transfer for applications as diverse as upper-ocean heating by solar radiation to laser ranging of the sea bottom. The approach differs from traditional ones and involves use of a special periscope prism that allows the direct determination of the VSF over a wide range of angles (0.68±177.38) with an angular resolution of 0.38. Measurements taken in the laboratory using Barnstead International, Inc., Nanopure water, cleaned seawater, and known additions of de®ned scatterers indicate close correspondence between experimental data and theoretical simulations based on Mie theory over the angle range from 128 to 1708. Field deployments in the Atlantic Ocean continental shelf water are shown to produce high quality measurements of the VSF in the angle range from 0.68 to 177.38; such observations have not been available before. The new data show that there is signi®cant environmental variance in the VSF for small angles (from less than 18 to several degrees) in the forward direction and from 1708 to 177.38 in the backward direction. The resulting observations are of profound and fundamental importance to the accurate modeling of the prop- agation of radiation in the ocean. 1. Introduction the range of variability in the VSF in the ocean. This is largely due to the extreme dif®culty in carrying out The volume scattering function (VSF), which de- the direct measurement of the property. Current radia- scribes the angular distribution of light scattered from tive transfer models rely on a very limited set of coarsely an incident beam, is a fundamental inherent optical property of the ocean. The VSF and the absorption co- resolved measurements of the angular distribution of ef®cient completely determine the inherent optical prop- scattering made more than 20 years ago (Tyler and Rich- erties of a medium in the absence of inelastic scattering; ardson 1958; Petzold 1972; Kullenberg 1974; see review coupled with the angular and spectral distribution of the in Morel 1973), despite widespread agreement on the incident radiance ®eld just below the surface, the full importance of the variability in the phase function in radiative ¯ux balance of the ocean can be simulated forecasting the underwater radiance distribution (Morel based on the radiative transfer equation. Such compu- 1973; Plass et al. 1985; Mobley et al. 1993, 2002; Gor- tations are of central importance in areas as diverse as don 1994). As well, the VSF can be inverted to obtain the upper-ocean heat balance, the photosynthetic pro- information on the nature of the particulate matter in ductivity of the ocean, and the chemical transformation the oceans, which is of wide interest (e.g., Brown and of photoreactive compounds. These analyses are also Gordon 1973; Zaneveld and Pak 1973; Zhang et al. key to current applications concerning the diagnosis of 2002) upper-ocean constituents from inversion of the spectral Instruments historically used for measuring the VSF distribution of upwelling radiance measured from re- across a range of angles (e.g., polar nephelometers; Ty- mote airborne and satellite platforms and for the esti- ler and Richardson 1958; Kullenberg 1974) are com- mation of the ef®cacy of laser ranging of the sea bottom. plicated and bulky and have no capability to measure Despite its fundamental nature, little is known about the VSF over the full angular range necessary to solve the general radiative transfer equation. Different devices are available to measure scattering at a few ®xed angles, Corresponding author address: Marlon R. Lewis, Dalhousie Uni- versity, Department of Oceanography, Halifax, NS B3H 4J1, Canada. over a narrow range in the forward direction [e.g., Se- E-mail: [email protected] quoia Scienti®c, Inc., Laser In Situ Scattering and Trans- q 2003 American Meteorological Society Unauthenticated | Downloaded 09/26/21 06:41 PM UTC 564 JOURNAL OF ATMOSPHERIC AND OCEANIC TECHNOLOGY VOLUME 20 missometer (LISST); Agrawal and Pottsmith 2000], or In accordance with these differences, different ap- the total scattering b (integral of the VSF over all solid proaches have been used for different angular ranges. angles), from which a portion of the VSF can be inferred using models that relate scattering at a given angle to elements of the full angular range in a theoretical sense a. Small-angle method of measurements (e.g., Man'kovsky 1971; Morel 1973; Oishi 1990; Maf- The most dif®cult problem of small-angle measure- ®one and Dana 1997; Boss and Pegau 2001). Because ments is the contamination of the scattered signal by of the lack of direct measurement, this link is uncertain light re¯ected and scattered from parts of the optical (but see Boss and Pegau 2001), in large part because unit. The so-called small-angle technique has been de- the nature of the particles responsible for scattering in veloped to avoid this problem. The small-angle tech- the ocean (particularly backscattering) is unknown nique is based on the illumination of a strictly de®ned (Zhang et al. 1998, 2002). volume of scattering media by a narrow parallel beam Here, we describe a new approach to the measurement and measurement of the scattered light in the focal plane of the VSF in the ocean. We present the theoretical of a receiving objective. The problem faced is the de- background, the instrument design, and the ®rst volume tection of a very weak signal at small angles near the scattering observations made in the ocean in the last 20 beam center, which interferes with the focal plane be- years. cause of the imperfect character of the direct beam; for example, the beam can exceed the scattered signal by 5 2. Background up to 10 in clean ocean waters (Bauer and Morel 1967; Petzold 1972; Agrawal and Pottsmith 2000). The volume scattering function b(u) is radiometri- cally de®ned as the radiant intensity I deriving from a volume element in a given direction u per unit of in- b. Measurements of general-angle scattering cident irradiance E and per unit volume V; that is, The typical instruments used for measurement of the dI(u) general-angle VSF have complex mechanical designs b(u) 5 (m21 sr21 ). (1) EdV because their angular deviation is provided by rotating a bulky light source or photodetector unit around the The VSF is often normalized by its angular integral to axis of the scattering volume (e.g., Petzold 1972). Be- yield the phase function, cause of interference by stray light, the minimal forward b(u) b(u) angle is limited to ;108, and for backscattered light b(u) 55(sr21 ), (2) measurements are limited to angles ,1708, owing to p b 2pb(u) sin(u) du physical restrictions based on the dimensions of the light E source and photodetector unit. For such instruments, it u50 is also dif®cult to provide baf¯ing against ambient light. which provides information about the relative angular Since modern photodetectors measure radiant ¯uxes, distribution of the scattering and where the VSF is as- it is more convenient to write Eq. (1) in terms of the sumed to be azimuthally symmetric (e.g., Mobley 1994). exiting and scattered radiant ¯uxes. The scattered ¯ux The denominator of Eq. (2) is the total volume scattering F(u) can be expressed as a function of the optical as- 21 coef®cient b (m ). sembly parameters of the scattering meter: 21 The backscattering coef®cient bb (m ) is computed p 2cr as 2pb#u5p/2 (u) sin(u) du, and the backscattering ratio F(u) 5 I(u)Ve , (3) is de®ned as b /b. A similar computation can be done b where V is the viewing solid angle of the photodetector for the forward hemisphere (e.g., Mobley 1994). (sr), c is the beam attenuation coef®cient (m21), and r There have been three instrumentation approaches to is the distance between the center of the scattering vol- the measurement of the b(u). Differences in the ap- ume and the photodetector (m). The irradiance E in Eq. proaches relate directly to different problems that arise (1) is determined by the ¯ux F from the radiant beam, for VSF measurements in small (e.g., forward), general, 0 which penetrates into the seawater and is attenuated and backward angles. The small-angle problem relates along the water path r from the light source to the center to the high level of background light generated by the 1 of the scattering volume, direct beam, the magnitude of which is several orders greater than that of the measured scattered light. The F0 E 5 e2cr1, (4) leading problems of the general-angle scattering (e.g., S 108,u , 1708) are the large dynamic range of mea- sured scattered radiance (.107) and a very low signal where S is the normal cross-sectional area of the exiting of scattered light at angles in the vicinity of 908. The light ¯ux. Combining Eqs. (1), (3), and (4), problem of insuf®cient (and dif®cult to determine) scat- F(u)Sec(r1r1) tering volume is added to dif®culties of measurements b(u) 5 . (5) of the low light level for angles in the range near 1808.