Sherwin D. Ladner 1, Robert A. Arnone 2, Richard W. Gould, Jr. 2, Alan Weidemann 2, Vladimir I. Haltrin 2, Zhongping Lee 2, Paul M. Martinolich 3, and Trisha Bergmann 4 Planning Systems Incorporated Naval Research Laboratory Neptune Sciences Incorporated Rutgers University Planning Systems Incorporated Naval Research Laboratory Neptune Sciences Incorporated Rutgers University MSAAP Building 9121 Ocean Optics Section Code Highway 190 East Institute of Marine and Coastal Sciences MSAAP Building 9121 Ocean Optics Section Code Highway 190 East Institute of Marine and Coastal Sciences Stennis Space Center, MS Stennis Space Center, MS Slidell, LA New Brunswick, NJ Stennis Space Center, MS Stennis Space Center, MS Slidell, LA New Brunswick, NJ voice: (228) ; PSI VARIABILITY IN THE BACKSCATTERING TO SCATTERING AND F/Q RATIOS IN NATURAL WATERS Ocean Optics XVI November 18-22, 2002 Santa Fe, New Mexico (1) (3)(4) Variability of the backscattering to scattering (b b /b) and the upwelling irradiance/upwelling radiance (f/Q) ratios affect the accurate retrieval of inherent optical properties from ocean color satellite algorithms (SeaWiFS and MODIS). We investigate the variability of b b /b and the f/Q in coastal and open-ocean surface waters of the Northern Gulf of Mexico and U.S. East Coast off New Jersey. In situ measurements of scattering (b) from an ac9 probe were collected with concurrent measurements of b b from a six channel Hydroscat sensor. We compare the measured b b /b values with values derived from the Petzold Volume Scattering Functions (Petzold 1972), examine the spectral variability of the ratio and examine the variability in different water types related to the changes in the Volume Scattering Function (VSF). In addition, we estimate the T*f/Q (Mobley 1999) term from above-water measurements of remote sensing reflectance (Rrs) coupled with direct measurements of absorption (a) and backscattering (b b ) coefficients. We will examine the spectral dependence of the T*f/Q term and its relationship to the b b /b ratio, which we use as a substitute for the changing VSF. Finally, we will show how the estimated T*f/Q values vary from the commonly used value of used for satellite processing. Introduction Remote sensing algorithms for ocean color visible imagery are based on relationships between remote sensing reflectance and such inherent optical properties as absorption and scattering. Previous studies have simplified these relationships and are currently used in present ocean color algorithms such as with SeaWiFS. The limitations of the relationships in different optical water types are under investigation. This effort is directed at examining different water types defined by the backscattering to scattering ratio and the changes in the T*f/Q term (where Q is the ratio of upwelling irradiance to upwelling radiance). Both the ratios affect the accurate retrieval of the inherent optical properties. The remote sensing reflectance (Rrs) is dependent on the angular scattering described by the volume scattering function (VSF). The VSF describes the angular distribution of scattered light in the water column. The integral of the VSF over full solid angle is the scattering coefficient (b); the integral over angles in the backward direction only ( degrees) is the backscattering coefficient (b b ). The VSF is influenced by particle characteristics such as shape, index of refraction, and size distribution. Because the VSF is difficult to measure (new instruments are just now becoming available), we use the b b /b ratio (or probability of backscattering) as a substitute for the VSF to describe water mass differences. This ratio provides a method to describe the differences in the VSF for different water masses and how the angular differences influence the Rrs. Understanding the differences in b b /b for different water masses will help characterize the changes in the VSF, which affect Rrs estimates, and provide better estimates of inherent optical properties in coastal and open-ocean waters (Northern Gulf of Mexico and New Jersey Coast) where different b b /b water types are observed. This is important for researchers to model optical remote sensing algorithms, visibility and laser propagation in seawater. Compare the measured b b /b ratios and linear Petzold relationship. Determine how the b b /b ratio (VSF surrogate to characterize water mass differences) changes both spectrally and in a wide variety of coastal and open-ocean water using in situ measurements. Use the relationship between measured remote sensing reflectance, backscattering and absorption to evaluate the T*(f/Q) term. Examine the spectral nature of the estimated T*(f/Q) term using in situ measurements, variation in relation to the commonly used satellite processing value of 0.05 and compare to past studies GULF OF MEXICO CRUISES Stations Stations High scattering waters with elevated suspended sediment loads and CDOM. NEW JERSEY CRUISES Stations Stations Clearer waters containing less sediment. Above Water Rrs – Analytical Spectral Device (ASD) field spectroradiometer processed using Near-Infrared surface glint removal which accounts for the Rrs at NIR wavelengths in high scattering waters (Gould et al 2001). This instrument measures the spectrum at 1.3 nm resolution from nm and a 12% spectralon (gray) card was used for calibration. Absorption [a] and Beam Attenuation [c] were measured using a WetLabs ac9 at 9 wavelengths (412, 440, 488, 510, 532, 555, 650, 676, 715nm). The instrument was calibrated using milli-Q water and the (Zaneveld et al. 1994) scatter correction was applied. We derived scattering by difference (c-a). Note: Red wavelengths are coincident between the ac9 and Hydroscat instruments. The spectral backscattering coefficient [b b ] was measured with the Hydroscat instrument, which measures the scattering at 140 degrees and extrapolates the b b at 6 wavelengths (442, 488, 532, 589, 620, 676nm). The instrument was calibrated at the factory prior to the deployments. Pure Water Correction : Backscattering and scattering due to pure water were removed using (Smith and Baker 1981) prior to investigating the differences in the spectral b b /b ratio. In Situ Measurements 2. Spectral b b /b Measured b b (532nm) vs. b (532nm) for the two Gulf of Mexico Cruises. Measured b b (532nm) vs. b (532nm) for the two New Jersey Cruises (LEO15). Measured b b /b (532nm) ratio (y-axis) as a function of Wavelength (x-axis) for the two Gulf of Mexico Cruises. Measured b b /b (532nm) ratio (y-axis) as a function of Wavelength (x-axis) for the two New Jersey Cruises. Two distinct water types for both Gomex and NJ regions. b b /b ratios of the Northern Gulf of Mexico waters during 2002 (left) and the New Jersey 2000 (right) are similar to the (Petzold 1972) ratio. b b /b ratio can vary widely from the (Petzold 1972) ratio. Slope 2002 = Slope 2001 = 0.01 Slope 2000 = Slope 2001 = Note: Each point represents the average bb/b value for each wavelength (slope of the line from the plot in section 1). Data seems to depict two distinct water types although there is spread and overlap in the data. First water type (Gomex 2002) is similar to the (Petzold 1972) relationship (white circles) which consisted of 15 VSF measurements (515nm) made in San Diego Harbor over 30 years ago. Second water type consist of Gomex 2001 stations collected in waters containing less scattering, sediment loads, and CDOM. Both Gulf of Mexico cruises were conducted in the same vicinity one year apart. Data seems to depict two distinct water types more clearly. First water type (NJ 2000) is similar to Petzold relationship (white circles) and Gulf of Mexico waters. Second water type (NJ 2001) consist of mostly open ocean waters dominated by biological rather than inorganic particles and produced a bb/b ratio lower than other cruises. Both New Jersey cruises were conducted in the vicinity one year apart. Spectral shapes for both Gomex and NJ regions are nearly flat. New Jersey waters show more distinct difference in magnitude. Changes in b b /b ratio can possibly be used to characterize the VSF in different water types and agrees with the behavior seen in actual VSF measurements (Haltrin et al., OOXVI 2002) 3.Measured Rrs vs. [b b /(a+b b )] : Evaluating the T*(f/Q) Term Rrs = (T*f/Q) * [b b /(a+b b )] ; (Mobley 1999) Theoretically, the T*(f/Q) should range between approximately – (0.051 is commonly used in most satellite processing techniques). Used measured properties of a, b b, and computed the ratio [b b /(a+b b )] for Gulf of Mexico and New Jersey stations. Regions were plotted separately but did not show a regional dependence. Relationships show a spectral variance in the T*(f/Q) term. Using in situ measurements of a, b b, and rrs, we observed the T*(f/Q) term to range from 0.04 (532nm) – 0.07 (440nm). The scatter could be due to higher in-water transparency at 532 nm than the other channels and is the wavelength of minimum absorption. Any small variation in absorption, whether it is due to the error in measurement or stability of the instrument, could possibly cause such scatter. Measured Rrs vs. b b /(a+b b ) ratio for all stations collected in the Gulf of Mexico and New Jersey regions by wavelength. 4. Spectral T*(f/Q) Comparison Mean T*(f/Q) vs. wavelength. Error bars indicate + or – one standard deviation. indicate + or – one standard deviation. Conclusions (Morel and Gentili 1993) results (Case I waters) are stable and linear as a function of wavelength whereas results in this study (non-Case I waters) show a larger variation. The b b /b ratio derived from in situ measurements varies spectrally over different water types and could be related to regional differences in the VSF. This ratio could act as a surrogate for differences in the VSF for different water masses. Understanding the b b /b relationship will help characterize the changes in the VSF, which affect Rrs estimates. The estimated T*(f/Q) in this study derived using above- water measurements of Rrs coupled with direct measurements of absorption (a) and backscattering (b b ) has more spectral variability in non-Case I waters compared to values presented by Morel and Gentili in case I waters. We found the spectral range of the T*(f/Q) term to range from 0.04 – 0.07 (different from the constant commonly used in most satellite processing techniques). From theoretical considerations, the term should range between and A regional dependence due to water type was not observed in this study. The parameterization of the T*(f/Q) term for different water types certainly can improve the accuracy of algorithms applied to remotely sensed ocean color data. Abstract Objectives Methods 1. Spectral b b /b vs. Petzold’s Relationship (2) Petzold Slope Constant : Common Satellite Value Petzold Slope Using b b /b as a surrogate to the VSF to characterize water masses. Slope = 0.04 Slope = Slope = T*(f/Q) = 0.07 Slope = All Stations & Regions – Black lines indicate least squares linear regression. 440 nm 676 nm 488 nm 532 nm