USGS CoastCam at Sand Key, Florida: Timestack Imagery and Coordinate Data (Camera 2)

Before mounting the CoastCam at the beach, the camera (FLIR, BFLY-PGE-50S5C-C) was used to take intrinsic camera calibration images. Please refer to the sandkey_c2_20180413_IO zip file and the sandkey_c2_calibration_metadata record included in Brown and others (2024) for access to the intrinsic camera calibration data and more information on the calibration process. After calibration was complete, the camera mount was installed on the rooftop of Dan’s Island Condominium building (1600 Gulf Blvd, Clearwater, FL 33767) on 20180413. The camera was installed and connected to a control computer, wall powered, and wired ethernet connection in an interior room of the facility. The control computer was connected to a hard drive to transfer the images upon collection.

The camera was programmed to collect three-band (RGB) images at 2 Hertz (Hz) for a total of approximately 17 minutes (2048 images), every hour during daylight hours. The images included in this data release were collected from May 29, 2018, to November 5, 2021. Gaps may exist in the dataset due to power disruptions or camera malfunction. Additionally, some images may have been obscured by water droplets, insects, birds, salt, sand and dirt, poor visibility from adverse weather conditions, or sun glare, and thus may not reflect the true state of the beach and ocean. Each hour the camera was activated, a timestack image product was produced by sampling intensity values at a single transect of pixels, defined by (U,V) pixel coordinates, from the 2048 images and saved in Tag Image File Format (TIFF, .tiff). The (U,V) sampling locations were determined by identifying a cross-shore transect of locations in local horizontal (X,Y) coordinate system, which aligned with the local shoreline orientation. The elevation of samples was chosen as Z = 0 m, which is an elevation assumed to be near the water line. The vertical reference frame was NAVD 88 coordinate system. XYZ-sampling locations were projected into (U,V) image coordinates following established methods (Palmsten and Brodie, 2022; Bruder and Brodie, 2020) and using camera extrinsic calibration information (sandkey_c2_20180419_EO) from Brown and others (2024). To create the timestacks, the raw images were saved on the control computer’s hard drive, then processed locally to produce one timestack per hour. The timestacks were then sent via internet to a U.S. Geological Survey internal File Transfer Protocol (ftp) site, where they were then automatically downloaded from the ftp site to the SPCMSC servers for archiving.

The TIFF timestacks were processed to add additional information using the netCDF4 Python library version 1.6.4. The times of the collection, sampled at 2 Hz, were stored as a one-dimensional (1D) array, where each index of the array contained a time represented as epoch time, seconds elapsed since 00:00:00 UTC on January 1, 1970. The sampling locations, saved as local XYZ values, were stored as a two-dimensional array. Because pixels were stored assuming an elevation of Z = 0 m and the true beach surface may vary, and additional processing step was needed to project the sampling locations onto a topo-bathymetric surface representative of the beach at the time of the image data collection following previously described methods (Palmsten and Brodie, 2023; and Holland and others, 1997). The topo-bathymetric surface was developed from GNSS survey data collected within the field-of-view of the camera (Birchler and others, 2024) and lidar data (National Geodetic Survey, 2024). The point data was interpolated onto a grid using the method described by Plant and others (2002) with smoothing length scales of 1 m in the cross-shore direction and 40 m in the alongshore direction. The local (X,Y) coordinates were rotated and translated into UTM coordinates using information describing the local origin and rotation angle for the Sand Key site, where the horizontal reference frame (X,Y) was NAD83(2011)/UTM Zone 17N. The (X,Y) coordinates were also translated from UTM to WGS84 latitude and longitude coordinates using the UTM Python library version 0.7.0. One netCDF(.nc) file was created for each timestack image using the netCDF Python library. The data included in each netCDF file includes the timestack image; the UV image coordinates, the XY coordinates in UTM; the XY local coordinates; the latitude and longitude coordinates; the Z-elevation coordinates projected onto the beach surface; and an array of the epoch times as previously described. Additional metadata was included in the netCDF files as global and variable attributes using the Climate and Forecast (CF)-compliant metadata convention (version CF-1.6). These attributes describe the data collection, as well as how the different coordinate systems are used within the data.

A total of 1,404 netCDF timestack files (*.nc) are included in this release and can be viewed and downloaded on the data release webpage . The netCDF files were renamed using the following convention, with ten period-separated elements: 1) the epoch time--seconds elapsed since 00:00:00 UTC on January 1 1970--the timestack was collected 2) The day of the week the timestack was collected, abbreviated to three letters 3) The month the timestack was collected, abbreviated to three letters 4) The date when the timestack was collected in the format dd_hh_mm_ss where 'dd' is the day of the month, 'hh' is the hour in 24-hour format, 'mm' is the minute, and 'ss' is the second 5) The timezone for date previously described 6) The year the timestack was collected 7) The site nickname where the timestack was collected 8) The camera the timestack was collected from, where 'cx' represents the timestack 9) the name of the transect sampled to create the timestack 10) the file extension, .nc. An example of a filename formatted in this way would be: 1527631200.Tue.May.29_22_00_00.GMT.2018.sandkey.cx.runup100.nc.