Tidal constituents from remote sensing image sequences

Abstract We address the problem of obtaining tidal constituents from a series of shoreline positions derived from time-sequential airborne imagery of the intertidal zone. A multi-sensor experiment ( Bachmann et al., 2012 ) conducted at the Virginia Coast Reserve Long-Term Ecological Research (VCR LTER) site 1 supplied in-situ and airborne data. Hyperspectral and infrared shoreline imagery of the VCR LTER barrier islands, shallow lagoons, and mainland coastal areas as well as topographic LiDAR data were obtained from the same air platform over a nine-day period, and LiDAR and in-situ topographic beach surveys were used to specify the beach bathymetry and derive a Digital Elevation Model (DEM) over the primary area focused on in this paper at Wreck Island, VA. The DEM and shoreline positions are used to obtain a time sequence of tidal heights; these are found to be consistent with a time series from a GPS-equipped tide gage float anchored nearby. A least-squares fit of the five largest tidal constituents (M 2 , N 2 , S 2 , O 1 , K 1 ) to this one-week record indicates that the first three can be retrieved with only a ±15–20 % error, as determined by the one-year sea-surface height time series from a nearby NOAA Sea Surface Height (SSH) station at Wachapreague, VA. The broader question of how many images are needed to obtain tidal constituents from an annual record is also addressed. Using SSH values from a NOAA tide gage station as a surrogate for image-derived SSH values, we progressively decimate a yearlong tide record, calculating its tidal constituents at each level of decimation. We find that that only one image every ten days is needed to retrieve the five largest tidal constituents to within a 10% error.