Research Papers

Ocean color patterns help to predict depth of optical layers in stratified coastal waters

[+] Author Affiliations
Martín A. Montes-Hugo

Mississippi State University, Geosystems Research Institute, Starkville, Mississippi 39529

Alan Weidemann, Richard Gould, Robert Arnone, Ewa Jaroz

NASA, Stennis Space Center, Naval Research Lab, Hancock, Mississippi 39529 alan.weidemann@nrlssc.navy.mil; Rick.Gould@nrlssc.navy.mil; Bob.Arnone@nrlssc.navy.mil; ewa.jarosz@nrlssc.navy.mil

James H. Churnside

NOAA, Earth System Research Laboratory, Boulder, Colorado 80305 James.H.Churnside@noaa.gov

J. Appl. Remote Sens. 5(1), 053548 (September 02, 2011). doi:10.1117/1.3634055
History: Received December 20, 2010; Revised August 10, 2011; Accepted August 18, 2011; Published September 02, 2011; Online September 02, 2011
Text Size: A A A

Open Access Open Access

Subsurface optical layers distributed at two different depths were investigated in Monterrey Bay, East Sound, and the Black Sea based on spatial statistics of remote sensing reflectance (Rrs). The main objective of this study was to evaluate the use of Rrs(443)/Rrs(490) (hereafter R1) skewness (ψ) as an indicator of vertical optical structure in different marine regions. Measurements of inherent optical properties were obtained using a remotely operated towed vehicle and R1 was theoretically derived from optical profiles. Although the broad range of trophic status and water stratification, a common statistical pattern consisting of lower ψR1—a deeper optical layer was found in all study cases. This variation was attributed to optical changes above the opticline and related to horizontal variability of particulates and spectral variations with depth. We recommend more comparisons in stratified coastal waters with different phytoplankton communities before the use of ψR1 can be generalized as a noninvasive optical proxy for screening depth changes on subsurface optical layers.

Figures in this Article

A common assumption of remote sensing algorithms based on ocean color sensors is the vertical homogeneity of the water column in terms of optical properties. This approximation is very often not met in coastal and oceanic stratified waters due to the presence of laminar features altering the underwater light field. These submerged layers commonly correspond with thin layers (i.e., <3 m thick), have typically high concentration of dissolved and particulate material with respect to the surrounding medium, and are preferentially developed along the horizontal component.1 Technologies used to detect vertical location of subsurface optical layers are generally based on lidar (light and range detection) systems.13 Unlike these investigations, in a recent contribution we showed a new approach to discriminate waters with shallow versus deep optical layers based on passive optical measurements.4 Briefly, the relative distance of the optical layer to the sea surface is estimated by calculating the third moment around the mean (i.e., skewness or ψ) of a specific remote sensing reflectance ratio [R1 = Rrs(443)/Rrs(490)]. As ψR1 decreases the subsurface optical layer becomes deeper.

The present study has four objectives. First, it will test optical relationships previously found by Ref. 4 by using an alternative approach consisting of independent calculations of ψR1 based on in situ measurements of inherent optical properties (IOPs) obtained with an undulating remotely operated towed vehicle (ROTV). Second, it will examine whether the aforementioned approach can be generalized across different marine coastal domains located at different latitudes and characterized by different trophic status. Third, it will investigate the mechanisms explaining ψR1 changes in terms of IOP modifications. Finally, it will quantify the influence of water stratification on biological “layering” and its impact on ψR1 patterns.

ROTV Surveys

High vertical resolution profiles (0.02 to 0.56 m) of in situ optical (i.e., total absorption coefficient of dissolved + particulate matter, a, and beam attenuation coefficient, c) and conductivity-temperature-depth (CTD) measurements were obtained using a ROTV (Scanfish MK II, intelligent undulating vehicle, EIVA, Denmark) in coastal oligotrophic [Black Sea (BS), 41.22°N to 43.70°N, 28.90°E to 31.26°E], mesotrophic [Monterrey Bay (MB), 36.33°N to 36.83°N, 121.05°W to 122.85°W], and eutrophic [East Sound (ES), 48.62°N to 48.67°N, 122.86°W to 122.89°W) waters during August to September 2008, June 2008, and May 2009, respectively. The ROTV can provide higher temporal (up to 10-fold) and spatial (up to 25-fold vertical, up to 10-fold horizontal) resolved cross sections than oceanographic gliders (e.g., electric Slocum). Thus, and given our future interest in implementing a satellite-derived approach for detecting depth of optical marine layers, we chose Scanfish as the sampling tool in this investigation.

The ROTV (0.9 × 0.3 × 1.8 m, length, height, width) was operated using an undulating mode complemented with a winch to expand the vertical sampling range up to 400 m. The ROTV has a weight of 75 kg and a payload of approximately 50 Kg. ROTV data were collected between 14 and 16 p.m. (local time) and two case studies were selected per study site: “shallow” (NS) and “deep” (FS) subsurface optical layer.

Analysis of Optical and CTD Profiles

Raw a and c coefficients were derived from an ac-9 (WetLabs) at three wavelengths (440, 488, and 675 nm) and corrected by salinity and temperature effects56 using CTD (SeaBird 911) measurements. Scattering residuals were removed.7 Only descending dives including ROTV-derived IOPs, temperature, and conductivity were processed and smoothed every 1 m along the vertical. Salinity and seawater density at each depth was derived from CTD variables and using the standard UNESCO polynomial equation of state.89 For each variable, data gaps were removed by linear interpolation. Missing determinations were more common at finer spatial sampling frequencies (e.g., in ES). The pycnocline depth was defined as the depth at which vertical changes of water density were above 0.015 Kg m−4.

Modeling of Ocean Color

Spectral remote sensing reflectance [Rrs(λ)] was estimated for each descending profile based on vertical distribution of a and c values at specific wavelengths. A light propagation model (Hydrolight, Sequoia Inc.) and ancillary data provided from airport-based meteorological stations (Black Sea, www.infospace.ru, Monterrey and Eastsound, www.wunderground.com) were used to simulate Rrs(440) and Rrs(488) above each Scanfish profile. For each case study, we computed the skewness of Rrs(440)/Rrs(488) ratios4 and its uncertainty by creating synthetic Scanfish transects with the same horizontal length (i.e., n = 10), but using different combinations of dives (ups and downs) obtained randomly from the original datasets. For each experiment, we examined how the skewness of Rrs(440)/Rrs(488) ratios and optical properties measured above the subsurface layer [e.g., a(675)] were interrelated. Also, we explored mechanisms behind these statistical relationships by analyzing the response of ψR1 to magnitude changes on a and total scattering (i.e., b = ca) coefficients at different wavelengths.

Finally, we investigated the depth of the signal source based on first optical depth (FOD) estimates. FOD was calculated as the inverse of Kd(488) or the diffuse attenuation coefficient of downwelling irradiance, with Kd(488) ∼ a(488)/

μ̃
(Ref. 10), where μ̃ is the average cosine of the zenith angle of refracted solar photons (direct beam) just beneath the sea surface.

The ROTV diving “saw tooth” pattern varied between studied areas (Fig. 1). Depth range, averaged vertical resolution, and horizontal spacing between profiles were 2.8 to 65, 0.41, and 370.9 m for BS, 1.6 to 48, 0.36, and 268.5 m for MB, and 0.7 to 22, 0.02, and 87.4 m for ES, respectively. Cross-sections of a(675) suggested that phytoplankton was an important optical component in all subsurface layers under investigation (Fig. 2). Consistent with an increase of solar radiation and freshwater river discharge as the spring–summer season progresses, the pycnocline and the optical subsurface layer characterized by relatively high a(675) with respect to background (up to 3-fold) were always deeper during the second experiment in chronological order. This variation was remarkable in BS, where the vertical position of maximum a(675) and pycnocline was 14 and 8.8 m in NS, and 28 and 23.2 m in FS, respectively. In general, drastic vertical changes of a(675) were observed in the vicinity of the pycnocline but during the FS experiment in ES [Fig. 2]. In this case, it is suggested that phytoplankton communities were not actively growing and were probably sinking as part of a post-bloom stage mainly composed of senescent cells (personal communication Dr. Percy Donaghay). Despite these differences, spatial patterns of ocean color above the sea surface always reflected similar modifications when optical submarine layering became deeper. In general, we found in every experiment that skewness of R1 switched from positive to negative, as the optical submarine layer was placed farther away from the sea surface (Fig. 2). Based on two standard errors [i.e., ∼95 probability of rejecting null hypothesis μ1 = μ2, where μ is the arithmetic average, and subscript 1 and 2 refer to ψR1(NS) and ψR1(FS), respectively], we estimated the uncertainty of ψR1 for each case study and found that a minimum of 15, 5, and 103 profiles are needed to differentiate shallow from “deeper” layers in Black Sea, Monterrey Bay, and East Sound, respectively. The size of these ideal datasets varied between 1.3 km (e.g., Monterrey Bay) and 9 km (e.g., East Sound).

Graphic Jump LocationF1 :

An example of Scanfish MK II dives (ups and downs) in the BS, MB, and ES. BS, MB and ES sampling locations had an averaged bottom depth of 125, 120, and 27 m, respectively.

Graphic Jump LocationF2 :

Vertical position of subsurface optical layers and spatial patterns of Rrs. Depth variations of a(675) for shallow [(a)–(c)] and deep [(d)–(f)] case studies are indicated for Black Sea [(a) and (d)], Monterrey Bay [(b) and (e)], and East Sound [(c) and (f)]. ψR1 is a dimensionless variable defining the spatial skewness of Rrs(443)/Rrs(490) (lower-right corner in each panel), two standard errors of ψR1 are indicated between parentheses, FOD is the first optical depth in m (black dashed line), PD is the pycnocline depth in m (black solid line). Missing data are depicted in white. Plots are based on 1-m vertically averaged a(675) data interpolated between Scanfish dives.

Given the relatively shallow detection depth [i.e., smaller than depth of subsurface a(675) layer] of passive sensors in these waters, the aforementioned spatial changes in ocean color could be mostly explained by optical changes within the layer above the main “opticline.” Indeed, we found that skewness of inherent optical properties measured in this layer and R1 may covary at specific wavelengths (Table 1). For each experiment, deepening of the subsurface optical layer and associated decrease of ψR1 was always accompanied by a decrease on a(490)/b(490) above the main opticline. This pattern was not observed at shorter wavelengths or using single optical variables instead of ratios. Thus, we suggest that dissolved colored components (i.e., constituents strongly absorbing in the UV-blue spectral range) had a smaller influence on ψR1 variability with respect to that associated with particulates (e.g., phytoplankton cells, detritus, bacteria). Also, the above results imply that ψR1 variability is largely modulated by light absorption and scattering.

Table Grahic Jump Location
Spatial statistics of opticals properties above the main opticline. For each case study, spatial skewness is computed for different IOPs and wavelengths influencing skewness of Rrs(440)/Rrs(488).

The aforementioned covariability between ψR1 and depth of the optical layer was also attributed to light spectral changes with depth. When layers are relatively shallow, the variance in the reflectance at the wavelength that is attenuated faster dominates the variance in the reflectance ratio. When the layer is relatively deeper, this wavelength stops varying since light is mostly attenuated in the layer nearest the sea surface. Thus, the variance in the reflectance in the other wavelength dominates the variance of the ratio. This change exhibits itself as a change of sign in the skewness of R1. The variance in reflectance at a given wavelength itself is driven by changes in the depth of the layer with the added issue of exponential weighting of the optical properties contributing to the reflectance. Hence, a meter of deepening contributes differently to variance than a meter of shallowing.

Thermohaline structure was usually coupled to depth changes of pigmented particulates as inferred from a(675) (Fig. 3). These results emphasize the importance of stratification on vertical distribution of phytoplankton, and are consistent with other studies highlighting the major association (i.e., >70% of cases) between the depth distribution of optical layers and water column stability.11 Despite this overall correspondence, decoupling between optical, biological, and physical variables may occur (e.g., ES in Fig. 2) when optical constituents do not behave as passive tracers (e.g., migrant phytoplankton), or they form aggregates that escape the pycnocline barrier. These ecological scenarios may introduce large uncertainties when vertical localization of submarine optical layers is based on hydrographic profiles.12 Conversely, ψR1 was very sensitive to vertical changes of subsurface optical layers for a broad range of vertical mixing and ecological conditions. However, we recommend more comparisons in stratified coastal waters with different phytoplankton communities before the use of ψR1 can be generalized as a noninvasive optical proxy for screening depth changes on subsurface optical layers.

Graphic Jump LocationF3 :

Biological versus physical mechanisms modulating vertical distribution of optical properties. Vertical variation (δ) of a(675) and seawater density (ρ) per meter is plotted in log10 scale. Each data point corresponds to the arithmetic average within 1 m bin. Ascending and descending Scanfish measurements are included.

We appreciate the valuable scientific comments given by Dr. Emmanuel Boss that helped to improve the original manuscript version. This work was supported by the NRL internal project “3D Remote Sensing with a Multiple-Band Active and Passive System: Theoretical Basis,” UNSPECIFIED PE0601153N .

Churnside  J. H., and Donaghay  P., “ Thin scattering layers observed by airborne lidar. ,” ICES J. Mar. Sci.. 66, , 778–789  ((2009)).
Hoge  F. E., , Wright  W., , Krabill  W. B., , Buntzen  R. R., , Gilbert  G. D., , Swift  R. N., , Yungel  J. K., , and Berry  R. E., “ Airborne lidar detection of subsurface oceanic scattering layers. ,” Appl. Opt.. 27, , 3969–3977  ((1988)).
Barbini  R., , Colao  F., , Fantoni  R., , Ferrari  G. M., , Lai  A., , and Palucci  A., “ Application of a lidar fluorosensor system to the continuous and remote monitoring of the Southern Ocean and Antarctic Ross Sea: Results collected during the XIII and XV Italian Oceanographic campaigns. ,” Int. J. Remote Sens.. 24, , 3191–3204  ((2003)).
Montes-Hugo  M., , Churnside  J. H., , Gould  R. W., , Arnone  R. A., , and Foy  R., “ Spatial coherence between remotely sensed ocean color data and vertical distribution of lidar backscattering in coastal stratified waters. ,” Remote Sens. Environ.. 114, , 2584–2593  ((2010)).
Pegau  W. S., , Gray  D., , and Zanaveld  J. R. V., “ Absorption and attenuation of visible and near-infrared light in water: Dependence on temperature and salinity. ,” Appl. Opt.. 36, , 6035–6046  ((1997)).
Langford  V. S., , McKinley  A. J., , and Quickenden  T. I., “ Temperature dependence of the visible–near infrared absorption spectrum of liquid water. ,” J. Phys. Chem. A. 105, , 8916–8921  ((2001)).
Zanaveld  J. R. V., , Kitchen  J. C., , and Moore  C. C., “ Scattering error correction of reflecting tube absorption meters. ,” Proc. SPIE. 2258, , 44–55  ((1994)).
Millero  F. J., , Chen  C. T., , Bradshaw  A., , and Schleicher  K., “ A new high pressure equation of state for seawater. ,” Deep-Sea Res.. 27A, , 255–264  ((1980)).
Fofonoff  P., and Millard  R. C., “ Algorithms for computation of fundamental properties of seawater. ,” UNESCO Technical papers. 44, , 53  ((1983)).
Mobley  C. D.,  Light and Water: Radiative Transfer in Natural Waters. ,  Academic ,  San Diego, California  ((1994)).
Dekshenieks  M. M., , Donaghay  P. L., , Sullivan  J. M., , Rines  J. E. B., , Osborn  T. R., , and Twardowski  M. S., “ Temporal and spatial ocurrence of thin phytoplankton layers in relation with physical processes. ,” Mar. Ecol.: Prog. Ser.. 223, , 61–71  ((2001)).
Zawada  D. G., , Zanavell  R. V., , Boss  E., , Gardner  W. D., , Richardson  M. J., , and Mishonov  A. V., “ A comparison of hydrographically and optically derived mixed layer depths. ,” J. Geophy. Res.. 110, , C11001  ((2005)).

Biographies and photographs of the authors not available.

© 2011 Society of Photo-Optical Instrumentation Engineers (SPIE)

Citation

Martín A. Montes-Hugo ; Alan Weidemann ; Richard Gould ; Robert Arnone ; James H. Churnside, et al.
"Ocean color patterns help to predict depth of optical layers in stratified coastal waters", J. Appl. Remote Sens. 5(1), 053548 (September 02, 2011). ; http://dx.doi.org/10.1117/1.3634055


Figures

Graphic Jump LocationF3 :

Biological versus physical mechanisms modulating vertical distribution of optical properties. Vertical variation (δ) of a(675) and seawater density (ρ) per meter is plotted in log10 scale. Each data point corresponds to the arithmetic average within 1 m bin. Ascending and descending Scanfish measurements are included.

Graphic Jump LocationF2 :

Vertical position of subsurface optical layers and spatial patterns of Rrs. Depth variations of a(675) for shallow [(a)–(c)] and deep [(d)–(f)] case studies are indicated for Black Sea [(a) and (d)], Monterrey Bay [(b) and (e)], and East Sound [(c) and (f)]. ψR1 is a dimensionless variable defining the spatial skewness of Rrs(443)/Rrs(490) (lower-right corner in each panel), two standard errors of ψR1 are indicated between parentheses, FOD is the first optical depth in m (black dashed line), PD is the pycnocline depth in m (black solid line). Missing data are depicted in white. Plots are based on 1-m vertically averaged a(675) data interpolated between Scanfish dives.

Graphic Jump LocationF1 :

An example of Scanfish MK II dives (ups and downs) in the BS, MB, and ES. BS, MB and ES sampling locations had an averaged bottom depth of 125, 120, and 27 m, respectively.

Tables

Table Grahic Jump Location
Spatial statistics of opticals properties above the main opticline. For each case study, spatial skewness is computed for different IOPs and wavelengths influencing skewness of Rrs(440)/Rrs(488).

References

Churnside  J. H., and Donaghay  P., “ Thin scattering layers observed by airborne lidar. ,” ICES J. Mar. Sci.. 66, , 778–789  ((2009)).
Hoge  F. E., , Wright  W., , Krabill  W. B., , Buntzen  R. R., , Gilbert  G. D., , Swift  R. N., , Yungel  J. K., , and Berry  R. E., “ Airborne lidar detection of subsurface oceanic scattering layers. ,” Appl. Opt.. 27, , 3969–3977  ((1988)).
Barbini  R., , Colao  F., , Fantoni  R., , Ferrari  G. M., , Lai  A., , and Palucci  A., “ Application of a lidar fluorosensor system to the continuous and remote monitoring of the Southern Ocean and Antarctic Ross Sea: Results collected during the XIII and XV Italian Oceanographic campaigns. ,” Int. J. Remote Sens.. 24, , 3191–3204  ((2003)).
Montes-Hugo  M., , Churnside  J. H., , Gould  R. W., , Arnone  R. A., , and Foy  R., “ Spatial coherence between remotely sensed ocean color data and vertical distribution of lidar backscattering in coastal stratified waters. ,” Remote Sens. Environ.. 114, , 2584–2593  ((2010)).
Pegau  W. S., , Gray  D., , and Zanaveld  J. R. V., “ Absorption and attenuation of visible and near-infrared light in water: Dependence on temperature and salinity. ,” Appl. Opt.. 36, , 6035–6046  ((1997)).
Langford  V. S., , McKinley  A. J., , and Quickenden  T. I., “ Temperature dependence of the visible–near infrared absorption spectrum of liquid water. ,” J. Phys. Chem. A. 105, , 8916–8921  ((2001)).
Zanaveld  J. R. V., , Kitchen  J. C., , and Moore  C. C., “ Scattering error correction of reflecting tube absorption meters. ,” Proc. SPIE. 2258, , 44–55  ((1994)).
Millero  F. J., , Chen  C. T., , Bradshaw  A., , and Schleicher  K., “ A new high pressure equation of state for seawater. ,” Deep-Sea Res.. 27A, , 255–264  ((1980)).
Fofonoff  P., and Millard  R. C., “ Algorithms for computation of fundamental properties of seawater. ,” UNESCO Technical papers. 44, , 53  ((1983)).
Mobley  C. D.,  Light and Water: Radiative Transfer in Natural Waters. ,  Academic ,  San Diego, California  ((1994)).
Dekshenieks  M. M., , Donaghay  P. L., , Sullivan  J. M., , Rines  J. E. B., , Osborn  T. R., , and Twardowski  M. S., “ Temporal and spatial ocurrence of thin phytoplankton layers in relation with physical processes. ,” Mar. Ecol.: Prog. Ser.. 223, , 61–71  ((2001)).
Zawada  D. G., , Zanavell  R. V., , Boss  E., , Gardner  W. D., , Richardson  M. J., , and Mishonov  A. V., “ A comparison of hydrographically and optically derived mixed layer depths. ,” J. Geophy. Res.. 110, , C11001  ((2005)).

Some tools below are only available to our subscribers or users with an online account.

Related Content

Customize your page view by dragging & repositioning the boxes below.

Related Book Chapters

Topic Collections

PubMed Articles
Advertisement


 

  • Don't have an account?
  • Subscribe to the SPIE Digital Library
  • Create a FREE account to sign up for Digital Library content alerts and gain access to institutional subscriptions remotely.
Access This Article
Sign in or Create a personal account to Buy this article ($20 for members, $25 for non-members).
Access This Proceeding
Sign in or Create a personal account to Buy this article ($15 for members, $18 for non-members).
Access This Chapter

Access to SPIE eBooks is limited to subscribing institutions and is not available as part of a personal subscription. Print or electronic versions of individual SPIE books may be purchased via SPIE.org.