Video data collections during SandyDuck consisted of image-based products as well as pixel and stack time series. These are explained below, as well as collection schedules.

Image-based products: Standard image-based collections consisted of snapshot, 10 minute time exposure images and 10 minute variance images (figure 1). The latter two are collected synchronously. In other words, the same frames are used for each. However the time exposure consisted of a simple averaging of frames while the variance image is computed by calculated appropriate intensity moments then finding the variance at each pixel. In fact, variance images show standard deviation, not variance.

Five cameras allowed coverage of the entire region from north to south (figure 2, upper). Data from these could be merged to give a full panoramic view (figure 2, lower).

Figure 2. Image collections were from 5 cameras looking from north, through offshore, to south. Knowing the photogrammetry, these could be merged into a single panoramic view.

From merged information from the cameras, a full plan (map) view could be calculated (figure 3). Merged views were also calculated from variance images.

Time series of video intensity were collected for a suite of pixels. All locations were specified by the horizontal location of interest (an x-y pair) and by a vertical elevation equal to the predicted tide elevation (corrections for actual surge, etc, will have to be accounted for after the fact).

Pixels of interest corresponded to the following:

• locations of all pressure sensors and current meters
• runup transects at y = {703, 778, 815, 828, 855, 905}
• longshore time stacks for longshore current estimation centered at the locations of SIO instruments {v11, v12, v13, v21, v22, v23, v24, v25, v32, and two other offshore sites}.
• Cross-shore arrays at every north CRAB line using x = [100:2.5:500].
• Longshore arrays at x = [125: 25:375] spanning the CRAB lines with 2.5 m longshore spacing.

Figure 3. Merged panoramic view (lower) and accompanying plan view showing the shoreline and bar morphology over a 2 km section of beach centered around the pier. The rip channel group near y=1000 was unexpected (although visually appealing).

Pixel collection points for cameras 0 and 3 are shown in figures 4 and 5, respectively. Pixel time series are 17 minutes long with a sampling interval of 0.5 s. Colections from the two cameras are 0.25 s displaced, but are generally interpreted onto the cam 0 time stamps using a linear interpolation.

Figure 4. Pixel array for camera 0, shown in oblique (upper) and plan (lower) views.

Figure 5. Pixel array for camera 3.

Collections were made hourly from 0600 EST until 1800 EST from 09/24 until 10/31/97.

Each collection consisted of a snapshot, then timex/variance pair, going for cameras (0,1) then (2,3), then cam 4. At 34 minutes past the hour, the pixel time series began. Time exposure images began on 9/21.

970930 had bad pixel collects.

971001 – pixel spacing for Hil changed from 5 m to 2.5 m.

971010 – first two hours of pixel collects missing

971015 – stacks bad all day due to Holman flare for programming.