March 91 satellite image and the January 86 aerial photograph
were manually digitized. The January 86 aerial was taken at low
tide, whereas the February 86 SPOT image was acquired at nearly
high tide. The only substantial difference between the hand
digitized and computer created vector coverages was that the
computer converted a small boat in the March 91 image in to a
shoal. Inspection of the other computer created coverages
showed that this was an intermittent problem, depending on the
speed of the boat. (Faster boats left larger, brighter wakes.) The
boat wakes, of which 4 were converted to "shoals", were
eliminated by hand from the remaining images. None of the boat
wakes occurred in the primary study areas. Figure 2 shows
several results for the Joiner Bank Area. The distance of the tip
of the hook to the low tide line was measured for each of the five
data sets. One surprising result of this study is that Joiner bank
appeared to move fairly consistently over the period from 1974 to
1989, averaging just over 70 meters per year of landward
migration. For SPOT multispectral data at 20 meters/pixel, this
would be 3.5 pixels of movement per year, an easily detectable
amount. From comparisons between the 1986 aerial photograph
and SPOT image, it appears that the position of the low tide line
on a shoal can be determined to approximately 1/3 pixel, or about
+/- 3 meters. Panchromatic data, at 10 meters/pixel, lends greater
accuracy to the movement rates and positions. However, it should
be noted that this increase in apparent accuracy may not be
entirely relevant, as these shoals can move several meters during
a single mild storm event, and due to the variation in low tide
levels. The movement rates and estimated positional accuracy's
for Joiner Bank are summarized in Table 1. From this data it
would appear that the rate of movement of the shoal has
increased since the end of the renourishment project. Although
only additional study over time will verify this, there are other
indications of increased wave energy in the Joiner bank area
which will be described later. The same process was performed
on the Barrett and Grenadier shoals at the south end of the island.
Very little migration was detected in that area. Conversations with
local boaters seem to indicate that the south end is not nearly as
dynamic as the northern areas of the island.
Date Source of Distance Estimate Ave Rate
Data to Tip of Accuracy of motion
Spit of Dist.
Mid 74 NOS Chrt 2195 m :i-10m ----—-—--—
Jan. 84 Aerial 1524m +/- 01m 67 m/yr
Feb. 86 SPOTXS 1373m +/- 03 m 72 m/yr
May 89 SPOTXS 1148m +/-03 m 70 m/yr
Mar. 91 SPOTPan 948m +/- 02 m 109 m/yr
Table 1: Movements of the End of Joiner Bank.
To summarize, analysis of the satellite data resulted in reasonably
accurate positions and geometry of emergent shoals. The tide
level at the time of image acquisition is an important factor in the
analysis, and must be corrected for. There was very little
positional difference in manual versus automated raster to vector
conversion. As with any image classification, inspection for
spurious data (such as boat wakes) was important. Overall, this
method was considered to be very successful for ongoing shoal
location and geometry monitoring.
Updating and Using Derived Bathymetric Data
The waters off the Atlantic shoreline of Hilton Head are quite
shallow. Waves coming in from the deep ocean are refracted
significantly before striking the shoreline, either concentrating or
scattering wave energy. Since wave energy is the primary driving
force of near shore sediment movement, determining wave
refraction patterns is a key to understanding shoreline erosion.
Given the dynamic nature of the shoals around Hilton Head,
traditional nautical charts should probably not be used as the only
source of bathymetric data for use in computer models of wave
refraction. The most significant refraction effects occur when the
wave enters water shallower than 1/25 the wavelength. For the
highest energy waves striking Hilton Head, this corresponds to a
depth of around 5 meters. (US Army COE, 1973) The following
assumptions and methods were used in developing a bathymetric
data set for use in the wave refraction model:
322
e Given the sandy characteristics of the area, a bottom
reflection based model would be accurate to a depth of 3
meters. Band 1 was used to calculate depth, with band 2
being used as a check.
e Reflective characteristics of the sand across the area are fairly
uniform. :
e Changes between 3 and 5 meters were less likely to occur
than those above 3 meters, and NOS data could be used in
these areas if sediment loads precluded the use of the satellite
model.
The 1989 multispectral image was used to derive the bathymetric
data for this study. Throughout the study area, comparisons were
made between the satellite derived model, NOS charts, and
soundings used for the design of the renourishment project. The
correlation between known sites and the satellite derived depths
was generally good, with the average error less than .3 meters.
The composite bathymetric model was used as the basis for a
wave refraction study, which was compared to the wave refraction
study submitted by the Town's consulting engineer. (Olsen Assoc,
1987) The model used by the consultant was based on a 366
meter grid, while the model derived for this study used the 20
meter grid size of the original satellite image. This point is
important, since the grid size for the consultant's model was 3
times larger than the wavelength of the waves under
consideration, while the GIS grid was about 1/5 wavelength. While
the results of the two models generally agreed, there were some
significant differences. The finer grid model predicted somewhat
higher wave energies impacting the Port Royal shoreline opposite
Joiner Bank due to the dredging. Also, the finer grid indicated
other possible disruptions to that area, such as changes to the
tidal currents from Port Royal sound. (The consultant study did
not consider tidal effects.) The finer grid was very sensitive to
wave input conditions, with variations resulting in more changes to
the impact on the beach than coarser grids. Work is still
progressing on the integration of the high resolution refraction
model and the results of beach profiles taken during and after the
project.
Detection of Refraction Patterns
A variety of image processing algorithms were tried to enhance
features which may be visible on the satellite images. One feature
visible in certain areas was the actual wave refraction patterns.
Simple edge enhancement of a contrast stretched image was
sufficient to bring out these patterns. Ocean waves at the time of
the 1991 image were from the south east, 0.6 to 1.0 meter in
height, with a wavelength of approximately 120 meters. These
parameters were used as input to the wave refraction model.
Refraction patterns were output as a GIS vector coverage. On a
digital overlay of the computer model patterns on the enhanced
image the patterns matched quite closely. Although aerial
photography would show shorter wavelength waves not resolved
by satellite, with further work this method may be useful for the
verification of wave refraction models over large areas.
Monitoring Underwater Features
The original reason for attempting to derive an updated
bathymetric model was to update NOS charts for areas near the
dynamic shoal systems. Study of the images revealed a number
of entirely submerged features. Mapping underwater formations
was the third task undertaken during this study. Most significant
among these was imaging the actual borrow sites used for the
beach renourishment project. The March 1991 panchromatic
image clearly showed both sites. It was, however, apparent that
the two sites were recovering in different ways. The Gaskin Bank
site appeared to be infilling with material of the same type as the
surrounding sandy sediment. The reflectance values were
consistent with those of the surrounding areas after correction for
the attenuation due to being 2 meters deeper. There was some
suspended sediment apparent. The Joiner Bank area, however,
showed a great deal of suspended sediment in the area of the site.
The site itself was darker than could be accounted for by
attenuation due to depth. Patterns in the sediment plume
appeared to originate from Port Royal sound. These
interpretations were confirmed by a study of the site by the South
NA MA
