a linear calibration derived from pump samples (Figure 10.3). Laser scattering data were inverted to produce particle size distributions as described by Agrawal and Pottsmith.46 The inversion process was applied with a modified version of a Matlab routine provided by Sequoia Scientific. Data quality was ensured by eliminating data with an optical transmission of less than 15% as advised by Sequoia Scientific. The inversion produced a size distribution reported as volume concentration (jxl l-1) in 32 logarithmically spaced size classes including particles from 2.5 to 500 ^m in diameter. Size classes for the LISST are fixed by the physical configuration of the scattering sensors. The cumulative volume concentration from all size classes was reported as total volume concentration (^.ll-1) and was used to calculate median particle diameter by volume (D50). Data from the 20 smallest classes were combined to calculate volume concentration <66.5 ^m. The remaining 12large sizeclasses were combined to calculate volume concentration >66.5 ^m. Bulk density (pb) was estimated as follows: The density of water (pwater) was calculated using temperature and salinity data from the CTD. Mineral solids density (psolid) was assumed to be 2.65 g cm-3. Solids fraction (<s) was calculated as TSS/Total Volume Concentration/psolid, and

FIGURE 10.4 (a) Photograph of M. Yates recovering the profiling rig during a tidal cycle anchor station. The VISTA settling tube is clearly visible, with the ADV to its left and the CTD immediately to its right in the image. (b) Captured image from VISTA from 10:10 am (EDT) on October 10, 2002 in ETM region. Sample depth is approximately 9.5 m. Field of view is 11.5 mm wide by 8.5 mm high.

FIGURE 10.4 (a) Photograph of M. Yates recovering the profiling rig during a tidal cycle anchor station. The VISTA settling tube is clearly visible, with the ADV to its left and the CTD immediately to its right in the image. (b) Captured image from VISTA from 10:10 am (EDT) on October 10, 2002 in ETM region. Sample depth is approximately 9.5 m. Field of view is 11.5 mm wide by 8.5 mm high.

Tidal cycle anchor stations were conducted as two vessel efforts over 12 daylight hours on one day of each seasonal cruise. A water column profiling rig with a trans-missometer, a LISST-100C, a CTD, an Acoustic Doppler Velocimeter (ADV), and a video settling tube was lowered repeatedly to 3, 7, and 10 m depths at 0.5 h per cycle from the Cape Henlopen (Figure 10.4). A modified Valeport settling tube (based on the Owen tube design47) was used to estimate settling velocities of relatively undisturbed particles collected just above the bottom at several times during the tidal cycle. A LISST-100C and a Precision Measurement Engineering Self-Contained Autonomous MicroProfiler (SCAMP) temperature microstructure profiler were deployed repeatedly from the Coot, with LISST-100C casts between each set of SCAMP profiles. Only the Coot LISST and SCAMP CTD data are presented here; analysis of the thermal dissipation rate data has been problematic. The different data types were analyzed individually and combined into depth-time contour plots using Surfer.

During the tidal cycle anchor stations, disaggregated size distributions of suspended sediment were evaluated by high volume pumping and filtering. Samples were collected at 1 m above the bottom, over both flood and ebb tides, using a submersible pump. The output of the pump was connected to three separate onboard filtration housings in parallel to enhance the speed of filtration, and immediately filtered through Millipore Durapore 0.65 ¡xm membrane filters. Filters were wrapped in aluminum foil and stored on ice until returned to the laboratory. Sediment was removed from the filters via sonication. Fractional pipette grain size analysis48 was performed on the combined material from all filters collected during each flood or ebb sampling period.

The Valeport settling tube, approximately 5 cm in diameter by 1 m long, was modified by constructing a flow-through water jacket around it. It was deployed horizontally in the water column at the desired depth and oriented into the flow by a vane. A messenger was sent to close the ends of the tube, which was then raised back to the surface, still in the horizontal. On deck, the tube was wrapped with reflective bubble wrap insulation and placed vertically into its stand, after which a hose running from a pump lowered to the sampling depth was connected to the water jacket. The combination of reflective insulation and flowing water jacket effectively eliminated internal circulations in the tube, one source of uncertainty in previous settling tube studies.49 Samples were withdrawn from the bottom for TSS analysis at specified, geometrically increasing time intervals starting 2 min and ending 80 min after inversion. The sequence of bottom-withdrawal TSS values was analyzed using a spreadsheet implementation of the procedure of Owen,47 similar to that described by Jones and Jago.50 This procedure yielded the distribution of settling particle mass in specific settling velocity intervals. Settling speed data fromETM surveys in 1996 and 2001 were collected and analyzed using the same methods.

Questions have been raised about data from both bottom withdrawal settling tubes and LISSTs in the literature,32,49,51-53 especially when used to characterize fine, flocculated sediment particles. For these reasons, and also because we hoped to be able to directly relate particle size and settling velocity, we developed a Video In situ Settling Tube Apparatus (VISTA) for use on the profiling rig (Figure 10.4). The VISTA used an on-deck pump to pull water through a tube attached to the profiling rig. The opening of the tube was directed into the flow by a fin. The pump pulled water into the mouth of the tube, through a 90° turn and up a vertical section of clear tubing at a flow rate approximating isokinetic sampling. The pump was run for 1 to 2 min at each depth to flush the settling tube, then valves at the top and bottom of the vertical section were triggered to trap a water sample and create a settling chamber. A PULNiX CCD underwater video camera was used to record images of the settling particles. The camera was mounted on the side of the settling tube with its faceplate about 5 mm from the side of the tube. The in-focus field of view in the center of the sampling tube was 8.3 mm high by 11.2 mm wide. Lighting was provided by a collimated underwater video light, shown in a sheet through the side of the tube that overlapped the focal plane of the camera. The camera output an analog signal that was recorded on Hi-8 tapes. An example of the image is shown in Figure 10.4.

For particle sizing, images were digitized and captured using the frame grabber software DVStill at a resolution of approximately 640 x 480 pixels. Selected frames were saved as bitmap files and analyzed using the Image Toolbox in MATLAB. The image processing technique converted the image to a 256 shade grayscale, sharpened it to enhance the edges of the particles, removed the background by use of a threshold filter, and then linearly enhanced the contrast to cover the range of the grayscale. Levels of sharpening and thresholding were adjusted depending on lighting conditions. The particles were identified using a Prewitt edge detector, and the final size and shape were determined using dilation and erosion functions. Equivalent particle diameters were calculated by taking the average of the minor and major axis lengths, and particle volumes were calculated using the equivalent diameters. Particles under the size of three pixels (approximately 30 jxm) were not considered. Approximately ten frames from each video clip were processed. All analyzed particle diameters and volumes were used to calculate the overall size distribution and D50 for comparison with simultaneously collected LISST-100C data, while individual particle diameters were used for comparisons to estimated floc settling speed.

Particle settling speeds were estimated using an ExpertVision (EV) Motion Analysis system. The EV Video Processor captured analog video data directly from the Hi-8 tapes, then created digitized outlines to facilitate particle tracking. The sub-sampled field of view was represented as a grid 256 pixels wide by 240 pixels tall. A distance scaling of 0.035 mm/pixel was used. The EV Motion Analysis software was used to calculate the vertical motion of particles. Due to the 90° turn at the tube's intake and rapid closure of the valves, some turbulence was created in the settling chamber. Small fluctuations due to ship motion were also present. Thus, it was necessary to estimate a background velocity by calculating the speeds of small particles that moved with the flow. Floc settling speeds were calculated by subtracting interpolated background vertical velocity from the vertical velocity of each large particle.

Relationships between particle size and particle settling speed were assessed by comparison with the floc fractal relationship defined by Hill54 as

WSr \drj where wsf, df and wsr, dr are the settling speed and diameter of a floc and the smallest particle in the floc (the reference particle), respectively, and D3 is the fractal dimension. The settling speed of the reference particle is given by Stoke's equation:

18 v where g is the acceleration due to gravity, v is the kinematic viscosity of the water, and sr is the specific gravity of the reference particle. Defining p = g(sr - 1)/(18v), assuming sr = 2.65 g cm-3 and v = 0.012 cm2 sec-1 and substituting (10.2) into (10.1), we obtain

which was fit to the data in a least squares sense to obtain estimates of D3 and dr.

Estimates of turbulent shear were made using data collected with a 10 MHz Sontek Acoustic Doppler Velocimeter (ADV) deployed on the profiling rig during the tidal cycle anchor station studies. The ADV on the profiling rig measured velocity in three dimensions in a single sampling volume (0.25 cm3), burst sampling 4096 samples at 10 Hz every 10 min. This allowed time for the profiling rig to be repositioned at the next depth between bursts. The ADV was mounted such that the sampling volume was at the same elevation as the LISST sampling volume and the intake for the VISTA. Vertical velocity spectra were used to estimate turbulent dissipation rate (e) by the inertial subrange technique,55,56 removing the (minimal) effects of ship motion with the technique developed by Gross et al.57 for estimating dissipation in the presence of wave motions. Small scale shear, y, was calculated as y = (e/v)1/2[sec-1], where v is the kinematic viscosity.

Time series current velocity profiles were collected for the duration of each seasonal cruise using a 1,200 kHz Broadband Acoustic Doppler Current Profiler (ADCP) mounted in the hull of a surface buoy in a downward looking configuration. The ADCP buoy was deployed at 39° 22.098' N, 76° 07.523' W in approximately 11 m of water. The ADCP collected data at 5-min intervals, sampling 0.5-m bins through the water column starting at 1.56 m below the surface and extending to approximately 1 m above the seabed. For each sample, 120 water column pings were averaged, which gives an estimated standard deviation of the measured velocity of about 0.5 cm sec-1. ADCP data was post-processed using MATLAB software, where all bad data points (including all data below 0.75 m above the seabed) were removed. North and east velocities were translated into along and cross-channel components, using a projected angle for the along channel direction of 230° T (ebb positive).

A microcosm erosion testing system was used to test the erodibility of bottom sediments collected using the Cape Henlopen Ocean Instruments multicorer. The erosion system consisted of two 10 cm Gust Microcosms,58 which use a spinning disk with central suction to generate a controllable, nearly uniform shear stress. A Campbell Datalogger controlled the system and stored data. During erosion experiments, a sequence of increasing levels of shear stress was applied to undisturbed cores. The effluent from each Microcosm was passed through a turbidimeter and time series of turbidity were measured. The effluent was collected, filtered, and weighed to determine the actual mass eroded during each step, which was used to calibrate the turbidimeter. Erosion rate was subsequently calculated as the product of pumping rate and suspended sediment concentration, and the data were analyzed according to the formulation of Sanford and Maa.59

0 0

Post a comment