SEDIMENT TRANSPORT PROCESSES AT A NOURISHED BEACH
PAUL A. WORK
A DISSERTATION PRESENTED TO THE GRADUATE SCHOOL
OF THE UNIVERSITY OF FLORIDA IN
PARTIAL FULFILLMENT OF THE REQUIREMENTS
FOR THE DEGREE OF DOCTOR OF PHILOSOPHY
UNIVERSITY OF FLORIDA
i,.~..r .d L L. *..\ :
To date I have been blessed with many excellent teachers, advisors, and mentors.
My advisor throughout my Ph.D. program, Dr. R.G. Dean, has set a standard which
is not likely to be surpassed, however. His unflagging support, enthusiasm, and
friendship will always be remembered, appreciated, and emulated. I also owe Mrs.
Dean thanks for her generous hospitality on many occasions.
Thanks are due to the staff of the Coastal and Oceanographic Engineering Lab-
oratory, particularly George Chappell, Don Mueller, and Mark Sutherland, for their
help with field work, and Sidney Schofield, for assistance in several areas, particu-
larly instrumentation. Many others assisted with field work at some point: Jorge
Abramian, Lynda Thompson, Mill Dowd, Dr. Dean, Jon Grant, Sam Houston, Feng
Jiang, A. Kadib, Li-Hwa Lin, Emre Otay, Rajesh Srinivas, and Victoria Stone.
I would also like to thank Professor Emeritus R.L. Wiegel of the University of
California at Berkeley for introducing me to the field of coastal engineering.
My final acknowledgement is reserved for those to whom I probably owe the most.
My parents instilled in me the work ethic and values that have allowed me to make
it this far, and have supported me in all my endeavors. And my friend Joni Schmidt
has helped me in more ways than she knows, providing me with everything from food
to a few lessons in what is and what should be important in life.
TABLE OF CONTENTS
ACKNOWLEDGEMENTS .. ........................
LIST OF TABLES .................................
LIST OF FIGURES ................................
ABSTRACT .. .. .. .. .. .. .. .. ... .. ... .. ... .
1 INTRODUCTION ...............................
2 LITERATURE REVIEW ...........................
2.1 Previous Studies of Beach Nourishment
2.2 Other Studies of the Perdido Key Area
2.3 Summary ................
3 FIELD DATA COLLECTION .......
Topographic/Wading Profile Surveys
Hydrographic Surveys .
Waves, Currents, and Tides .....
Sand Samples ............
Weather Data ......
Ft. Pickens Pier Tide Gage ......
Assessment of Errors ..
4 ANALYSIS AND INTERPRETATION OF FIELD DATA .........
4.1 Wave, Current, and Tide Data........................
4.2 Sediment Transport Processes: Observations Based on Survey Data
4.3 Longshore Sediment Transport Direction ...............
4.4 Nourishment Project Volumes ......................
5 MODELING LONGSHORE SEDIMENT TRANSPORT PROCESSES ..
Background .. .. .. .. ..... ... .. ... .
Analytical, "One-Line" Approach . .
Influence of Wave Refraction . .
Effective Wave Parameters . .
Comparison of Analytical Model to Measured Changes
Numerical Approach to One-Line Modeling .
5.7 Determination of Best-Fit Empirical Transport Coefficient ..... .
5.8 Discussion of Results . . .... .....
6 MODELING CROSS-SHORE SEDIMENT TRANSPORT PROCESSES
6.1 Background . . . .
6.2 Formation of Longshore Bars . . .
6.3 Previous Models for Beach Profile Evolution . .
6.4 Desired Characteristics of a Beach Profile Model . .
6.5 Cross-shore Transport Rate vs. Dissipated Wave Energy .
6.6 Previous Models Compared to Field Data . .
6.7 Sum m ary . . . . .
7 CONCLUSIONS ................................
REFERENCES ....... ......... .. ............. .. .
BIOGRAPHICAL SKETCH .................... ......
LIST OF TABLES
3.1 Chronology of Perdido Key Data Collection Efforts. ...... ..27
3.2 Coordinates, Elevations and Azimuths for DNR Monuments in
Escambia County, FL.................. ....... 33
3.3 Locations and depths of offshore sand samples. ... 49
3.4 Sieves used for grain size analysis. . ... 51
3.5 Cross-shore distribution of grain size at various contours (Gulf
side). Longshore standard deviation is given following median
grain size .. . . . 51
4.1 Monthly and annual statistics for significant wave height, H,. 68
4.2 Monthly and annual statistics for wave period corresponding to
peak of energy spectrum, Tp. . . 68
4.3 Monthly and annual statistics for magnitude of mean current, Uc. 69
4.4 Summary of available wave data for Perdido Key area. Data col-
lection for this study is ongoing. . . 81
5.1 Parameters selected for application of analytical solution for plan-
form evolution.................... .......... 127
5.2 Effect of various parameters on best-fit K value. Azx 3300 m,
At=1 day for all tests. an < 0 yields westward transport. 140
6.1 Forces and flows influencing cross-shore sediment transport. 159
LIST OF FIGURES
1.1 Schematic view of planform and profile evolution after beach nour-
ishment. Cross-shore width of project greatly exaggerated in top
figure, vertical scale exaggerated in lower figure. 5
1.2 Beach profile evolution in large-scale, two-dimensional wave tank.
Data from Kraus and Larson . . . 7
1.3 Beach profile evolution in field, after nourishment. Data from
Perdido Key, Florida. ................... ..... 8
1.4 Forces influencing beach profile evolution. . 9
3.1 Location of field project. ................... ... 25
3.2 Temporal availability of data at the field site. . .... 29
3.3 Locations of beach profile transects and data collection stations. 31
3.4 Generalized profile geometry and survey methodology. 36
3.5 Surveyed beach profiles at R-58, Perdido Key. ... 37
3.6 Raw bathymetry data for R-50, October, 1991. Two passes by
boat are superimposed. ................... .... 40
3.7 Bathymetry data for R-50, after spike removal, smoothing, adding
proper vertical and horizontal offsets, and patching with wading
profile data .................. ........... .. 41
3.8 Schematic of wave gage and tripod. . ... 42
3.9 Cross-shore distribution of grain sizes. . ... 52
3.10 Repeatability test at R-46, June, 1992. . .... 56
4.1 Spectra of noise in pressure sensor and current meter, based on
128 points sampled at 1 Hz, Hanning window. Gage in still air in
lab. . . . .. .. 63
4.2 Spectra from previous figure, transferred to free surface using an
assumed depth of 6 m.. .................... .64
4.3 Spectra of noise in pressure sensor, based on 1024 points sampled
at 100 Hz, Hanning window. Gage in still air in lab. ...... ..66
4.4 Number of valid pressure and current data bursts, sorted by month,
for the period 1/18/90-6/30/92 . .. 67
4.5 Maximum and mean monthly significant wave heights, (Hs)max
and H ............................ ..... 70
4.6 Maximum and mean monthly wave periods at spectral peak, (Tp)max
and T p. . . . .. .71
4.7 Maximum and mean monthly mean currents, (Uc)max and Uc. Cal-
ibration suspect for August, September, and October data. ... 72
4.8 Histogram of significant wave heights measured in the field. 73
4.9 Histogram of wave periods corresponding to spectral peak mea-
sured in the field. .................. ...... .. 74
4.10 Histogram of mean currents measured in the field. ... 75
4.11 Measured and predicted percent exceedance curves for significant
wave height. Measurements from Perdido Key, 1/90-6/92; theory
due to Thompson and Harris  . .... 77
4.12 Measured joint distribution of significant wave height and period
corresponding to spectral peak. Based on 2389 observations for
1/18/90-6/30/92. .................. ...... .. 78
4.13 Directional distribution of waves at the spectral peak, January,
1990-June, 1992. Top of plot corresponds to North. Each bin is
ten degrees in width .......................... 79
4.14 Directional distribution of mean currents, January, 1990-June,
1992. Top of plot corresponds to North. Each bin is ten degrees
in w idth. . . . .. 80
4.15 Coordinate system for sediment transport problem .... 83
4.16 Surveyed beach profiles at R-58, November, 1989, through May,
1991. . .. . ...... 85
4.17 Surveyed beach profiles at R-58, September, 1991, through June,
1992. . .. . ..... .. 86
4.18 Planform changes since beach nourishment. . ... 87
4.19 Movement of the -4 m contour since beach nourishment. 88
4.20 Planform changes associated with 1985 nourishment project. Data
collected by Rutgers University and reported in Dean . 91
4.21 Longshore gradient of longshore sediment transport rate, based
on measured profile changes. Positive gradients indicate erosion.
Pensacola Pass is adjacent to R-67; the western end of the survey
area is to the left .. ... .. .. .. .. .. .. .. ... 92
4.22 Average of profiles at ranges 45, 46, 48, 50, 52, 56, and 58 ... .93
4.23 Average of profiles at ranges 30, 32, 34, 36, and 38. ... 94
4.24 Cross-shore sediment transport rates computed based on average
profile changes within nourishment area for the 1990 and 1991
annual surveys. Offshore transport is positive, longshore gradients
of longshore sediment transport assumed negligible 95
4.25 Cross-shore sediment transport rates computed based on average
profile changes west of nourishment area for the 1990 and 1991
annual surveys. Effect of longshore gradients neglected. 96
4.26 Cross-shore sediment transport rates computed based on average
profile changes within nourishment area for intermediate survey
periods. Effect of longshore gradients neglected ... 98
4.27 Cross-shore sediment transport rates computed based on average
profile changes west of nourishment area for intermediate survey
periods. Effect of longshore gradients neglected. ... 99
4.28 Cross-shore transport signal after removal of longshore contribu-
tion, based on field measurements and computed "average" profiles
within the nourished beach...................... 101
4.29 Changes occurring at profile R-42 for first year. .... 103
4.30 Computed cross-shore distribution of the longshore gradient of
longshore sediment transport, q,(x, y)/x. . ... 104
4.31 Longshore component of radiation stress, Sxy(t) computed from all
suitable wave data. Sy, > 0 implies westward sediment transport. 107
5.1 Beach profile geometry for one-line model of planform evolution. 115
5.2 Error function .. .. ... .. .. .. .. .. .. .. 117
5.3 Beachfill evolution as predicted by analytical solution. K=0.77,
t=4000 m, Hb=0.5 m, (h. + B)=5 m, G=0.0113 m2/s. 118
5.4 Coordinate system for shoaling/refraction problem. Adapted from
Dean and Yoo [in press] ............ .... .... .. 123
5.5 Comparison of measured and idealized initial planforms. 127
5.6 Longshore gradient of longshore sediment transport rate, as pre-
dicted by one-line, analytical solution, vs. measurements, Septem-
ber, 1990, to October, 1991 ...................... 130
Finite-difference definitions for one-line model . .
5.8 Average retreat of waterline within nourished zone. Changes due,
almost exclusively, to cross-shore sediment transport. 135
5.9 "Error surface" indicating error between measured and numeri-
cally modeled longshore gradients of longshore transport as a func-
tion of K and (h. + B) ....................... 142
5.10 Best-fit (mimimum e') results for several combinations of K and
(h. + B). Curve 2: K=1.0, (h. + B)=2; 3: K=0.33, (h. + B)=4;
4: K=0.15, (h. + B)=8; 5: K=0.14, (h. + B)=10. All results
include 3.7 m/yr background erosion. . ... 143
5.11 Best one-line model results describing longshore gradient of long-
shore sediment transport, Q/x . .. 146
5.12 Computed longshore gradient of longshore sediment transport for
first and last daily time steps in "best" simulation. ... 147
5.13 Background erosion rate along Perdido Key, based on shoreline
positions indicated on historical maps. Pensacola Pass is to right. 150
6.1 Schematic view of three common distributions for the cross-shore
sediment transport rate. Note that each distribution has localized
zones of accretion and erosion. . . ... 162
6.2 Spatially-averaged minimum depth over bar vs. time, in field.
"Inside" implies nourished profiles (R41-R64); "Outside" means
"natural" profiles (R25-R40). . ...... 163
6.3 Spatially-averaged maximum depth in trough vs. time, in field.
"Inside" implies nourished profiles (R41-R64); "Outside" means
"natural" profiles (R25-R40). . ..... 164
6.4 Decay of cross-shore sediment transport rate with time, based on
surveys in large wave tank. (Case 300, Kraus and Larson ). 172
6.5 Total energy dissipated divided by surf zone volume, plotted against
average cross-shore transport rate within surf zone for large wave
tank data. Points are connected chronologically, with the maxi-
mum transport rate in each case corresponding to the first survey
interval..................... ........... 176
6.6 Total energy dissipated divided by surf zone volume, plotted against
average cross-shore transport rate within surf zone for field data. 177
6.7 Local cross-shore sediment transport rate vs. local energy dissi-
pation per unit volume for one large wave tank case. Results for
other cases exhibiting offshore transport similar. ... 178
6.8 Local cross-shore sediment transport rate vs. local energy dissipa-
tion per unit volume for selected intervals in the field. 179
6.9 Influence of slope term in modified Kriebel model for selected large
wave tank test. Hb=1.68 m, T=11.33 sec., D=0.22 mm, A=0.10
m1/3, K=l.lxl0-6 m4/N. "Alpha" is the coefficient in front of
the slope term in the transport equation. Calibration factor, K,
chosen to yield best fit for a=. . ... 184
6.10 R.M.S. error in depth as a function of transport coefficient, K, for
large wave tank case illustrated in Figure 6.9. . .... 185
6.11 Modified Kriebel model vs. average nourished profiles based on
measurements at Perdido Key, 9/90-2/91. K6bet=0.2xl0-6 m4/N.
R.M.S. error in depth=0.15 m. ................... 186
6.12 Modified Kriebel model vs. average nourished profiles based on
measurements at Perdido Key, 2/91-5/91. Kbest=2.0x0-6 m4/N.
R.M.S. error in depth=0.15 m. . . .. 187
6.13 Modified Kriebel model vs. average nourished profiles based on
measurements at Perdido Key, 10/91-1/92. Kbet=0.2x10-6 m4/N.
R.M.S. error in depth=0.14 m. ................... 188
6.14 Modified Kriebel model vs. average nourished profiles based on
measurements at Perdido Key, 1/92-6/92. Kbat=0.1x10-6 m4/N.
R.M.S. error in depth=0.26 m. ................... 189
6.15 R.M.S. error in depth as a function of transport coefficient, K, for
each field simulation. Note that K=0 corresponds to the R.M.S.
depth change measured in the field over the given interval 190
6.16 On- and offshore average slopes within nourished area, based on
measurements at Perdido Key. Onshore slope defined by +1 m
contour and waterline; offshore slope by -2 and -4 m contours.. 191
6.17 Deep water wave steepness, based on mean wave height (monthly
average), vs. dimensionless fall speed. Solid line defines switch
from offshore to onshore transport in SBEACH model. 193
6.18 Measured and modeled beach profiles, first post-nourishment sur-
vey interval at Perdido Key. Model result is based on Larson and
Kraus . K=l.5x10- m4/N, e=0.002 m/s .... 196
6.19 Measured and modeled beach profiles in field, profile recovery case.
Model result is based on Larson and Kraus . K=0.5x10-6
m4/N, E=0.001 m2/s. ......................... 197
Abstract of Dissertation Presented to the Graduate School
of the University of Florida in Partial Fulfillment of the
Requirements for the Degree of Doctor of Philosophy
SEDIMENT TRANSPORT PROCESSES AT A NOURISHED BEACH
Paul A. Work
Chairman: Dr. Robert G. Dean
Major Department: Coastal and Oceanographic Engineering
Data describing the evolution of a large beach nourishment project on the Gulf
of Mexico at Perdido Key, Florida, are analyzed to describe the sediment transport
processes governing the behavior of the beach. Analytical and numerical techniques
for prediction of the response of the nourished beach to physical forcing arising from
incident wind waves are tested.
Repetitive bathymetric and topographic surveys indicated placement of 4.1 mil-
lion m3 of sand in the nearshore zone between November, 1989, and August, 1990,
and a net loss of 7% from the monitored area after one year. Beach profiles were
surveyed every 3-4 months at 33 locations to monitor the evolving beach, providing
over 250 profiles describing the spatial and temporal changes at the site. Directional
wave data, sediment samples, tide, and weather data were also collected.
Cross-shore sediment transport rates and longshore gradients of longshore sed-
iment transport were computed from the measured changes in the beach profiles.
Results indicate that cross-shore sediment transport dominated much of the early
evolution of the project. Longshore gradients of longshore sediment transport were
found to be strongest on the "shoulders" of the beachfill, where shoreline curvature
changes most rapidly.
Longshore and cross-shore sediment transport processes were assumed indepen-
dent to allow separate investigations. A "one-line", numerical model for planform
evolution including the effects of background erosion, cross-shore sediment transport,
and spatial and temporal variation in the incident waves gave a reasonable description
of the longshore gradient of longshore sediment transport along the nourished beach.
Beach profile changes at the site were modeled by application of two previously
developed numerical models that simulate cross-shore sediment transport. One ap-
proach reasonably estimated the volumetric redistribution of sediment for the first
post-nourishment survey interval but yielded poor results for profile recovery events.
Performance of the second model was poor despite a more detailed description of the
cross-shore sediment transport rate.
The level of detail included in the assessment and description of long-term sedi-
ment transport processes at the site is largely unprecedented. The approach provides
information specific to beach nourishment projects as well as more general insight
into nearshore sediment transport problems.
The response of a beach to time-varying, physical forcing in the form of waves,
fluctuating water levels, nearshore currents, and winds is of significant interest in
both the applied engineering and research arenas. Nearshore sediment transport,
driven by the complicated and poorly-understood flows in and near the surf zone,
has proven to be a particularly durable research problem. This is evidenced in the
literature by numerous predictive models and laboratory and field studies regarding
coastal sediment transport. Yet no consensus has emerged of an adequate and general
Sediment transport in the nearshore zone has traditionally been divided, some-
what arbitrarily but conveniently, into shore-parallel longshoree) and shore-normal
(cross-shore) components. Although the two transport modes are clearly driven by
some, if not all, of the same forces, the complexity of the problem has generally re-
sulted in studies that either focus on only one mode or assume no coupling between
them. This has traditionally led to research that either addresses changes in the
planform or the profile of a beach.
The time scales for changes in the planform and profile of the beach are often
markedly different. A short period (hours to days) of very energetic waves, particu-
larly if coincident with enhanced water levels, can alter the beach profile drastically.
The retreat of the waterline that typically accompanies such an event is sometimes
only temporary, as pre-storm conditions may prevail within days of the event. This
has made field documentation of storm-induced beach profile changes difficult, since
conditions immediately before the storm are rarely known, and recovery has often
commenced by the time post-storm conditions are surveyed.
Major changes in the planform of the beach often occur on a time scale of seasons
to years. Such changes are usually attributed to longshore gradients of longshore sed-
iment transport. Many factors can lead to the presence of such gradients. Longshore
variation in wave height or wave direction, the presence of natural or man-made lit-
toral barriers, a change in shoreline orientation, and variability in the "erodibility" of
the beach (at one extreme because of the natural variability in sediment size, at the
other due to man-made structures such as bulkheads or revetments), can all lead to
Analytical and numerical models of beach planform evolution are often employed
to predict the effects of some human modification of the existing nearshore zone,
such as the construction or stabilization of a tidal inlet or placement of sand for
beach nourishment. The predictions often employ some simplifying assumptions and
require specification of empirically-evaluated coefficients. Field data for verification of
the coefficients are sparse, because of the difficulty and expense of the data collection
and the relatively large time scale involved. The present state-of-the-art for beach
planform modeling is, however, felt to be much better for quantitative predictions
than the models describing profile response.
A diverse range of approaches has been developed that fall under the general title
of beach profile modeling, from empirical representations of "equilibrium" beach pro-
files, to unsteady (sometimes unstable) models attempting to include detailed descrip-
tions of nearshore hydrodynamics. Engineering applications of beach profile modeling
include prediction of maximum storm erosion (both volumetric and in terms of water-
line recession), establishment of "set-back" lines for regulating coastal construction,
and investigation of the feasibility and predicted performance of beach nourishment
Most recent models for beach profile response have been tested against wave tank
data for calibration. Wave tank tests have several advantages, most significantly the
controlled conditions and the quasi-two-dimensional behavior due to the minimiza-
tion of longshore gradients. The ultimate test of any such model, however, is in
comparison to field data. The poor performance of many models for beach profile
response when attempting to describe events in the field is generally attributed to an
insufficient understanding of the complex physics of the surf zone.
An increase in both the popularity and cost of beach nourishment projects has
resulted in the need for better understanding of the processes that shape them over
their "lifetimes." The feasibility of a proposed project has often been assessed with
inadequate or untested hypotheses about how it will evolve over time. An increased
understanding of the post-nourishment behavior will help avoid costly mistakes.
Beach nourishment has been used as a tool to combat erosion for over half a
century, and many completed projects have been monitored to allow an assessment
of their performance. Monitoring has rarely been sufficient to allow any detailed in-
vestigation of the sediment transport processes governing the behavior of the project.
Often, only a qualitative assessment of project performance has been possible.
To realistically simulate the evolution of a nourished beach, both longshore and
cross-shore sediment transport processes must be addressed. A beach nourishment
project may be considered a man-made perturbation to a coast, which may already
be affected by human activity. Beach nourishment modifies both the planform and
the profile of the beach. This alters the pre-existing sediment transport processes,
inducing both longshore and cross-shore transport gradients that result in changes
in the shape of the beach.
There are two primary goals of this study. One is to present and interpret field
data describing both the evolution of a nourished beach and the forces driving the
response. The other is to assess the ability to predict the response, given the forcing.
A large-scale beach nourishment project at Perdido Key, Florida, initiated in Novem-
ber, 1989, provided an opportunity to collect appropriate data. Surveys immediately
before and after placement of fill material in the nearshore zone, and at roughly
four-month intervals afterward, provided a good description of the evolution of the
beachfill in both the profile and planform. Monitoring of environmental parameters
such as water level, mean currents, wave conditions, and winds provided knowledge
of the forces exerted on the evolving beach.
Most mechanical systems, given constant forcing, will approach an equilibrium
condition, often a dynamic equilibrium. Since beach nourishment is usually in re-
sponse to erosion, it is difficult to argue that the beach is in equilibrium before nour-
ishment, but the nourishment project generally increases the degree of disequilibrium
in both the longshore and cross-shore directions. Extending the analogy, it might
be expected that the rate of response is proportional to the degree of disequilibrium.
By this argument, the rate of change of the beach should decrease with time. It will
not reach zero, however, because the processes that made the nourishment project
necessary would still be eroding the beach after the perturbation vanishes.
Figure 1.1 provides a schematized view of the planform and profile perturbations
and response following a beach nourishment project. Because of the difference in time
scales, the profile changes become evident first. Planform changes immediately post-
nourishment are often idealized as rectangular, and the as-built, post-nourishment
profile is typically much steeper than the pre-nourishment profile. Offshore sediment
transport reduces the beach slope and width, while longshore gradients of the long-
shore sediment transport rate result in losses of sediment from the nourished area.
Figures 1.2 and 1.3 provide examples of wave tank and field measurements of
evolving beach profiles. A generalized, qualitative description of the processes causing
this evolution in the field provides some insight into the nature of the problem. Waves
from deep water propagate towards shore and begin shoaling, with possible energy
'.. ..'. ..* .
Figure 1.1: Schematic view of planform and profile evolution after beach nourish-
ment. Cross-shore width of project greatly exaggerated in top figure, vertical scale
exaggerated in lower figure.
inputs and losses from local winds, whitecapping, bottom friction, etc. Shoaling,
refraction, and diffraction can all be important in coastal waters and generally assume
increased importance with decreasing depth. Wind, tides, and wave setup/setdown
can all modify the local mean water level. The shoaling waves become increasingly
nonlinear with reduced relative depth until incipient breaking is reached. At this
point, organized wave motion is rapidly converted to turbulent water particle motions,
which can mobilize and entrain sediment. The applied stress of the wind and waves
acting on the water column will also result in the forcing of quasi-steady flows, both in
the longshore and cross-shore directions. These flows can then advect the sediment
suspended in the water column and mobilize additional sediment. Deposition of
the mobilized sediment provides feedback to the incident waves by modifying the
bathymetry responsible for shoaling and wave energy dissipation. These processes
are illustrated in Figure 1.4.
Many assumptions are often made to simplify the problem of describing the evolu-
tion of a nourished beach. The assumption regarding the independence of longshore
and cross-shore sediment transport modes was discussed previously. The incident
waves are often parameterized by specification of a single representative frequency
and wave height combination. With these simplifications, a fully satisfactory model
must still describe the spatial distribution of wave height and mean water level, wave
energy dissipation throughout the nearshore zone due to wave breaking, friction, and
turbulence, the nearshore velocity field, .and the resulting sediment transport.
Recent attempts to develop physics-based models for cross-shore sediment trans-
port have provided insight into the problem and some innovative approaches to de-
scribing the processes cited above, but poor quantitative agreement between measured
and predicted beach profile response. Additionally, all the models to date that have
attempted to retain the maximum amount of physics have been applied only for
Large Wave Tank Case 400
0 20 40
Distance from initial SWL (m)
Figure 1.2: Beach profile evolution in large-scale, two-dimensional wave tank. Data
from Kraus and Larson .
Perdido Key R-46
200 250 30(
Distance from Monument (m)
Figure 1.3: Beach profile evolution in field, after nourishment. Data from Perdido
: : : : : : 1 ; : : : .
: Bar Formation SUSF
:and Migration :. .:
Set Up/ Down
Figure 1.4: Forces influencing beach profile evolution.
In the chapters that follow, the field data collection techniques employed for this
study are described and their accuracy assessed. Information extracted from the
field data relevant to the sediment transport problem are discussed and summarized.
Results from "one-line" models for planform evolution, solved both analytically and
numerically, and two numerical models describing beach profile response to varying
wave conditions and water levels are compared to field measurements. The accuracy
and limitations of each model are evaluated.
The results presented here may be considered a step in the process that should
ultimately lead to a three-dimensional model for prediction of the evolution of a beach
nourishment project. Until sufficient knowledge and understanding are gained to
develop such a model, there will be a continuing need for field, laboratory, numerical,
and analytical studies that shed light on the sediment transport processes shaping a
beach after nourishment.
The problem of interest is quite complex since what is desired is the response
of a system to forces that are themselves poorly understood. Any comprehensive
model of coastal behavior must address the processes of wave shoaling and breaking,
the spatially-varying elevation of the mean water level (wave setup/setdown), some
details of both the fluctuating and mean motions of the fluid within the water col-
umn, and the resultant sediment transport. The recent trend in studies of evolving
beaches has been toward numerical models describing some or all of these phenomena.
Analytical models for special cases, such as planar beaches or rectangular beachfills,
are available in some instances, but numerical models are often employed when at-
tempting to simulate more realistic situations. Time-averaged quantities are almost
always employed, where the time-averaging is done over many wave periods, to reduce
Because of the importance of each of the component processes influencing the
evolution of a beach, an extensive review was made of the literature describing each
phenomenon. At this stage, however, the focus will be on prior studies that specifically
considered the problem of beach nourishment evolution in the field. Where necessary,
previous studies addressing the sub-processes (wave breaking, mean circulation, etc.)
will be reviewed.
It should be noted that the field data collected as part of this study do not include
any measure of the surf zone hydrodynamics. Wave heights, periods, and directions,
tidal stage and local wind speed and direction, which together are generally agreed to
include most of the important forces influencing the physical changes observed at a
beach, have all been measured. The response of the beach was also measured. But the
intermediate step, the environmentally-forced flows in and near the surf zone, which
induce the sediment transport causing the beach to change, have not been monitored.
This provides further justification for restricting the scope of the literature review to
studies specifically addressing the problem of beach nourishment evolution.
2.1 Previous Studies of Beach Nourishment Evolution
A review of some previous beach nourishment projects which were monitored
will help put the problem in perspective. Beach nourishment was being recognized
as a potential remedy to chronic erosion problems as early as the 1930's, but many
projects were not monitored adequately to quantify their evolution. The goal here is
not to assess the value or effectiveness of previous projects, but rather to determine
how they were monitored and what the monitoring programs revealed regarding the
physical processes important to the evolution.
The number of beach nourishment projects that have been completed is too large
to attempt a complete summary. Many projects were not monitored at all after con-
struction, or were monitored for only a short time. Only projects that were monitored
sufficiently to allow at least some assessment of their behavior after completion will
be discussed here. Many of the studies included surveys of the geometry of the beach
at varying levels of complexity, from measurements of beach width to wading profiles
to bathymetric surveys. Conclusions about changes in beach volume were often made
with little or no information about the subaqueous portion of the project.
Early reports by the Beach Erosion Board of the U.S. Army were visionary in
addressing the need for long-term survey data extending from the dry beach out to
beyond the limiting depth for sediment motion. Hall and Herron  provided an
early example of a well-monitored project at Long Branch, New Jersey. Tide, wave,
and wind data, sand samples, aerial photographs, and bathymetric surveys were all
employed to study the behavior of 460,000 m3 of sand placed in 11.5 m of water. The
area was monitored for roughly one year after placement, and no significant movement
of the sand was found. Results of similar experiments involving offshore placement of
sand for nearshore beach nourishment in 1935 at Santa Barbara, California, and 1935-
1942 at Atlantic City, New Jersey, were described. Findings at the three locations
were cited as evidence that nourishment material must be placed in shallower depths
to exhibit any net shoreward movement. A later investigation of the project at Long
Branch by Harris  which included additional data agreed with the findings of
the earlier study.
Later projects involved placement of sand on the beachface to increase dry beach
width. Hall  discussed beachfill design procedure and summarized over 50 beach
nourishment/sediment bypassing projects that had been completed at the time. The
majority of the projects were located in California, New Jersey, and Florida; they
ranged in volume from 50,000 m3 to over 14 million m3 (Silver Strand, Coronado
Beach, California, 1940-1944). Five case studies were considered to illustrate alter-
native approaches to beach nourishment.
Early monitoring projects often resulted in qualitative descriptions of the sediment
transport processes at the site, since the accuracy and temporal and spatial resolution
of the surveys made quantitative assessment speculative. Watts  described a
1.95 million m3 beach nourishment project at Ocean City, New Jersey, that was
monitored with the aid of bathymetric survey data, aerial photographs, and sand
samples. Numerous groins were constructed in the area for shoreline stabilization.
An estimated 95% of the fill material was lost from the nourished area in slightly less
than two years following nourishment. Lost material moved both up- and downcoast
to benefit adjacent areas. The finer grain size of the placed sand and the local shoreline
orientation were both concluded to contribute to the high loss rate.
Watts  evaluated a nourishment project in Harrison County, Mississippi,
with the aid of sand samples, bathymetric survey data, and wading profile data. A
terminal groin was built at one end of the 4.5 million m3 project for retention of
placed material. A hydrographic survey seven years after project completion revealed
that 90% of the 540,000 m3 "lost" from the beach above the -0.6 m contour (mean
low water) was merely transported offshore, adjusting the beach slope. This would
indicate a loss of roughly one percent in seven years, although survey errors must be
considered in this estimate. The benefit of this offshore material for shore protection
Watts  gave an account of the 1.0 million m3 beach nourishment project
completed in 1953 at Virginia Beach, Virginia. Emphasis was placed on the impor-
tance of adequate spatial resolution in the survey data for estimation of sediment
transport rates and directions.
Perdikis  summarized 79 beach nourishment projects in the northeastern
United States. Ten projects were discussed in detail, including changes in volumes
and beach slopes. The offshore extent of the survey was not given for all cases
and it appears that volumetric computations were based primarily on changes in the
subaerial beach. Many sites eroded much faster after nourishment than previously.
For many cases this was attributed to placement of material that was finer than
the native sediment or offshore sediment transport to result in a milder beach slope.
The difficulty of predicting post-fill behavior based solely on the knowledge of pre-
nourishment conditions was emphasized.
Vesper  monitored a nourished beach on Long Island Sound at Prospect
Beach, West Haven, Connecticut. Both the beach and the borrow area from which
335,000 m3 of sand for nourishment were obtained were studied for roughly three
years after placement. The borrow pit was located in only 1.5 m of water 300 m
offshore of the waterline. Beach nourishment material did not move into the borrow
pit within the three years of project monitoring, and no net loss of material from the
monitored area was found. The most significant adjustment of the fill was found to
occur during the first year after placement.
Kramer  noted the rapid initial reduction in beach width associated with
an unnaturally steep beach placed on the Island of Norderney in Germany. The
importance of seasonal trends was also noted. The later re-nourishment of this beach
and a further assessment of the first project was given by Kramer . The latter
paper included a qualitative prediction of the evolution of a proposed "sand groin" on
the Island of Sylt on the German North Sea coast. Kunz  discussed several later
nourishment projects at Norderney. Volumetric changes were found to be described
by an exponential curve. Beach nourishment has been employed at the site for nearly
forty years, with an average of 82,500 m3/yr being placed.
Several studies have included computations of changes in subaerial volumes of
nourished beaches, noting that much of the lost material is probably moving offshore
because of the unnaturally steep as-built profiles. Everts et al.  (also McCann
) monitored two beach nourishment projects at Atlantic City, New Jersey and
concluded that larger fill volumes generally implied more rapid loss rates. A nourish-
ment project at the same site in 1986 was monitored in a similar manner by Sorensen
et al. . The loss rate for the latter project was observed to decrease in the
second year of monitoring. This was attributed both to the characteristic behavior of
beach fills and a milder wave climate for the second year.
Fisher and Felder  studied a 465,000 m3 beach nourishment project at Cape
Hatteras, North Carolina. High-resolution beach profile data (bi-weekly, 150 m spac-
ing between profiles) and visual wave data were collected to monitor the subaerial por-
tions of the project for eighteen months after placement of the fill. Most of the changes
occurring during this time were found within the nourished area, where roughly half of
the initial volume above MSL remained after eighteen months. It was acknowledged
that much of the material lost from the subaerial beach was probably moving offshore
to deeper portions of the active profile.
A nourished beach at Treasure Island, Florida, was observed for roughly two
years to study post-nourishment behavior after placement of 600,000 m3 of sediment
(University of Florida ). Profiles were surveyed to the depth of closure to allow
computation of volumetric changes. Twenty months of monitoring revealed a loss of
25% of the project volume, much of it into and near the borrow pit. The fill appeared
to "stabilize" after one year, after which the erosion pattern was more uniform.
A study of a proposed 840,000 m3 beach nourishment project at Delray Beach,
Florida, employed monthly littoral drift "roses" (Walton ) and a finite-difference,
numerical simulation of planform changes to predict project performance (University
of Florida ). The annual volumetric losses from the nourished area were pre-
dicted to increase slightly over time for five years after construction.
Shemdin et al.  monitored a 2.6 million m3 beach nourishment project at
Jupiter Island, Florida. Shoreline recession and volumetric gain were both found at
both ends of the nourished area. This was attributed to a "groin effect." Predictions
of volumetric changes using an empirical longshore drift equation and measured wave
data were an order of magnitude larger than measured changes.
Dette  described the evolution of the 770,000 m3 "sand groin" (Kramer
) placed for beach nourishment purposes on the Island of Sylt in the North
Sea. Directional wave data and beach profile surveys to 5 m depths were used to
investigate the forcing and behavior. The project initially trapped sediment updrift
and thereby increased in volume.
Oertel et al.  monitored a nourished beach at Tybee Island, Georgia. The
project involved placement of 1.1 million m3 of sediment in 1976 on 6 km of beach.
Monitoring consisted of time-lapse still photography, movies, and bi-weekly beach
profile surveys using a modification of a method due to Emery . Rapid changes
to the subaerial beach were noted in the first six months after nourishment, with
much of the nourished area losing 40% of the subaerial volume. Much of this loss
(50-75%) was claimed to be a result of offshore sediment transport. Chu and Posey
 described a later project at the same site in 1987 involving 1.0 million m3
of sediment. Roughly 25% of the project volume was lost from the subaerial beach
during the first 21 months after placement.
Walton  summarized available data on beach nourishment projects in Florida
and on the lower Atlantic and Gulf coasts. Project performance was briefly assessed
where sufficient data existed. Hobson  provided a similar summary for twenty
projects located throughout the United States, including the well-known Miami Beach
project completed in the early 1980's.
Winton et al.  made a detailed investigation of the sediment transport pro-
cesses along 42 km of the southern coast of North Carolina. Observed changes after
beach nourishment at Wrightsville Beach and Carolina Beach were reported. An ex-
ponential loss of beachfill volume was found at Wrightsville Beach following the 1970
beach nourishment project. Eighty percent of the initial fill volume was lost during
the first 1-1/2 to 2 years after placement. This was attributed to sediment sorting,
slope readjustment, and lateral spreading of the fill.
Hushla  monitored two beach nourishment projects on the eastern coast of
Florida with the aid of aerial photography. Subaerial beach widths and areas were
used to assess performance. Five years of data for Port Canaveral Beach indicated a
loss of beach area equal to 30% of the increase resulting from project construction.
A similar calculation for Indialantic/Melbourne Beach based on changes observed in
one year revealed a loss of 50%.
Phillips et al.  studied a 720 m segment of South Beach, Sandy Hook, New
Jersey, after nourishment with 1.8 million m3 of sediment. Beach profiles were sur-
veyed and breaking wave and longshore current statistics recorded during site visits.
Volumetric changes were computed based on six surveyed profiles. Some profiles were
surveyed to the MSL contour and a constant slope assumed to allow extrapolation
to the -4 m contour for volume computations. The magnitude of the annual wind-
blown transport out of the study area was estimated at 26,554 m3/yr, although this
was qualified as an upper bound, order-of-magnitude estimate. The effects of wave
refraction, wave diffraction, cross-shore sediment transport, and measurement errors
were addressed qualitatively. An increased potential for longshore transport out of
the monitored area was noted as the shoreline prograded.
Kerkaert et al.  conducted a comprehensive study of a beach nourishment
project in Belgium. Predictions of post-nourishment planform changes were made
using the analytical approach of Pelnard-Considere . Repetitive bathymetric
surveys showed that six percent of the placed material moved either offshore or into
the dunes during two years of monitoring.
Ashley et al.  described the evolution of a nourished beach immediately
downdrift of Barnegat Inlet, New Jersey, a tidal inlet that was dredged and stabilized
with jetties. The beach was nourished with over one million m3 of sand and monitored
for over seven years, through profile surveys, sand samples, aerial photography, and
visual longshore current observations. Beach profile measurements were made at nine
stations, some of which were outside of the nourished area. Profiles were surveyed
bi-weekly for 3-1/2 years and annually thereafter, using the method due to Emery
. Measured changes in the subaerial beach volume were used to compute a
"residence time" for the beachfill of 3-1/2 years, compared to the pre-construction
estimate of 8 years.
Pilkey and several colleagues (Pilkey and Clayton , Leonard et al. [1990a],
Leonard et al. [1990b], Dixon and Pilkey ) summarized numerous beach nour-
ishment projects in the United States, reporting physical dimensions, costs, and
predicted and measured "lifetimes." The effects of various parameters on the beachfill
lifetime were discussed.
Stauble and Holem  documented an eight-year monitoring study of a nour-
ished beach at Indialantic/Melbourne Beach, Florida. The project spanned 3.4 km
and included 195,000 m3 of added sediment. The monitoring plan included wading
profile surveys, sand samples, aerial photography, and some wave data. Beach profiles
were surveyed weekly immediately after placement of the sand and then at increasing
intervals at later times. Offshore portions of the profiles were not monitored. Seven
years after placement of the beach nourishment material, the subaerial beach area
was found to be 13% greater than the pre-nourishment case.
Recent years have seen an explosion of reports on the evolution of beach nour-
ishment projects (Giardano et al. , Combe and Soileau , Jarrett ,
Jones and Kana , Aubrey and DeKimpe , Ibeet al. , Spadoni ).
Many of these document wave conditions, volume changes, etc., but do not attempt to
model the sediment transport processes determining the fate of the project. For econ-
omy of space, emphasis will be placed on studies that include empirical, analytical,
or numerical predictions of the future changes to the nourished beach.
Analytical and empirical predictions of volumetric changes in nourished beaches
have recently become a common component of monitoring studies. Wave height is
a dominant forcing term in most analytical investigations, but sufficient measured
wave data are rarely available. The analytical approach is often patterned after the
one-line theory of Pelnard-Considere . Volumetric changes for an entire project
imply an integrated approach (integrated in both the longshore and cross-shore di-
rections), so details of the sediment transport mechanisms causing the changes are
not usually addressed. Beachfill "lifetimes" are often of interest, although definitions
vary between studies and are not always given.
Skrabal et al.  estimated the lifetime of a 255,000 m3 nourishment project at
Fenwick Island, Delaware, based on historical shoreline change patterns. The lifetime
was defined as the time required for the nourished area to lose an amount of sediment
equal to the project volume. One year of post-nourishment data were available for
comparison, and revealed that 7.6% of the placed volume was lost from the monitored
area during this time. This loss rate was used to estimate a beachfill lifetime of 11.5
de Lange and Healy  monitored a small beach nourishment project (21,000
m3) in a tidally-influenced bay. Beach profile surveys spanning 3-1/2 years after
nourishment indicated that the beach volume above MLW decayed linearly, although
the loss rate appeared to be accelerating. It was predicted that all sediment added
during renourishment would be lost in 13 years.
Dean and Lin  discussed the measured and predicted performance of a
well-documented beach nourishment project at Redington Shores, Florida. A one-
line, numerical model for shoreline change, including background losses and measured
wave data as inputs, was used to simulate evolution of the project. Comparisons
were based on five surveys spanning eighteen months after project completion. The
approach is similar to Phlegar , who modeled, both analytically and numerically,
the volumetric changes occurring at ten nourished beaches in Florida, some of which
had been nourished multiple times.
DeKimpe et al.  modeled the fate of a small (81,000 m3) beach nourishment
project analytically and empirically. The analytical solution followed the one-line
approach of Pelnard-Considere  but did not yield volumetric changes as realistic
as an empirically-derived exponential curve. After five years of evolution, 1:43'f of the
subaerial project volume remained. The lifetime of the project, defined as the time
required for the fill to be reduced to 10% of its original volume, was estimated at 13
Larson and Kraus  discussed both analytical and numerical modeling of
post-nourishment beach changes. Longshore changes followed the one-line approach
of Pelnard-Considere , while cross-shore changes were modeled using a numer-
ical model based on observations from large-scale wave tanks (Larson and Kraus
, Larson, Kraus, and Byrnes ). The cross-shore model was tested against
field data describing beach profile evolution during a storm.
Mann et al.  modified the one-line approach of Pelnard-Considere  in
an ad hoc manner to account for an irregular pre-nourishment planform and back-
ground erosion. One post-nourishment survey from a 211,000 m3 beach nourishment
project at Buckroe Beach, Hampton, Virginia, was available for verification. Mea-
surements indicated a much more rapid loss than predicted for the first year after
Stauble et al. [1992, in press] made a detailed investigation of a beach nourishment
project at Ocean City, Maryland. Two phases of the project have been completed, the
first in 1988 (1.8 million m3) and the second during the summer of 1991. The scope
and goals of the monitoring project were very similar to the Perdido Key project that
is the focus of this study, involving bathymetric surveys, in situ wave data collection,
and sediment sampling. A detailed investigation of the temporal and spatial changes
in sediment size characteristics was made. Additional data collection and analysis
associated with this project is currently underway.
Some results became available as this study was nearing completion. Fernandez
et al.  described a monitoring program at a beach nourished with 1.5 million m3
of sand much larger than the native material. Visual wave data, bi-monthly beach
profile data to the -10 m contour, and sediment samples had been collected for over
two years at the time of publication to document changes in the 25 km project.
Ferrante et al.  discussed the construction and monitoring of a beach nour-
ished along 3 km with 1.2 million m3 of sand. A rubble mound berm breakwater was
built to retain the placed material, which was much coarser than native sediments.
The monitoring program included aerial photography, beach profile surveys, sediment
samples, and directional wave data. Surveyed profiles indicate an increase in eleva-
tion of the dry beach of close to one meter since construction. This gain was nearly
uniform in the cross-shore direction.
2.2 Other Studies of the Perdido Key Area
Few studies have addressed the physical forces acting on Perdido Key or the
response of the island to these forces. Balsillie et al.  determined long-term
shoreline change rates for the area based on historical charts. Psuty [1987, 1988]
monitored a 1.86 million m3 beachfill placed on the eastern end of Perdido Key in
1985. Stone et al.  estimated longshore sediment transport rates for much
of western Florida and southern Alabama based on hindcast and visual estimates of
wave conditions. A nodal point in the longshore sediment transport rate was predicted
on Perdido Key. Clark  provided a bibliography of selected reports regarding
coastal processes near Perdido Key. Most of the region monitored as part of the
present study was identified as an "erosion problem area."
Despite the cost associated with their construction, few beach nourishment projects
have been monitored sufficiently to allow a detailed, quantitative assessment of their
performance. Where monitoring studies have been sponsored, they have often been
of short duration and included a description only of the changes appearing on the dry
beach. Many studies have noted the importance of the offshore portions of the profile
but based performance assessment only on the condition of the subaerial beach.
The preceding literature review was not totally exhaustive, owing to the num-
ber of studies regarding the evolution of beach nourishment projects, but should
illustrate the deficiencies in the existing data sets. The complexity of the problem
demands a high level of care and expense associated with the data collection process.
High-quality bathymetric survey and directional wave data are obviously important
components of the data set.
Many recent studies have employed analytical and numerical tools for the predic-
tion of the response of a nourished beach to waves and currents. .lo-r have taken a
volumetric approach, which implies integration in both the longshore and cross-shore
directions. The remainder of this study will be devoted to a detailed investigation
of the spatially- and temporally-varying sediment transport processes that affect a
nourished beach after placement.
FIELD DATA COLLECTION
A significant portion of the research leading to this dissertation consisted of field
data collection at Perdido Key, Florida, and subsequent analysis and interpretation of
this data. Perdido Key is a barrier island on the Gulf of Mexico in the extreme western
"Panhandle" of Florida (Figure 3.1). It trends roughly east-west and is bounded on
the east by Pensacola Pass and on the west by Perdido Pass.
The U.S. Navy initiated a dredging project to increase the depth of Pensacola
Pass for navigation in November, 1989. Much of the dredge spoil was designated
for use as beach nourishment material for Perdido Key. This phase of the dredging
project resulted in approximately 4.1 million m3 of new material being placed on 7
km of the Gulf of Mexico shoreline of Perdido Key, increasing the dry beach width by
roughly 140 m. The native and fill sands are similar in color, composition, and size.
The entire nourishment area lies within the Perdido Key area of the Gulf Islands
National Seashore and is therefore under the jurisdiction of the U.S. National Park
Service. The park contains primarily undeveloped land and no coastal structures
except for a small structure abutting Pensacola Pass. The offshore bathymetry of
the study area is characterized by very mild slopes seaward of the -5 m contour;
the primary exceptions to this are the steep slope adjacent to the dredged channel
through Pensacola Pass, and Caucus Shoal, a long, thin, submerged sand spit adjacent
to Pensacola Pass at the eastern end of Perdido Key.
A five-year monitoring study containing several elements was sponsored by the
National Park Service and the U.S. Navy to investigate the evolution of the project
and its effect on pre-existing conditions. Biological, sedimentological, and physical
-~CZ--- II~L I 4
environmental conditions were all incorporated into the study. The Coastal and
Oceanographic Engineering Department of the University of Florida was contracted
to conduct the physical monitoring study, which will be the focus of this chapter.
The plans for the physical monitoring study called for the collection of several
types of data. A physical survey, including both subaerial and subaqueous features of
the island, was required yearly by the sponsors. To date, three such surveys have been
completed: the pre-nourishment survey of November, 1989, the post-nourishment
survey of September, 1990, and the survey of October, 1991. Intermediate surveys
extending to water depths of 4 to 5 m were also done to provide better temporal
resolution of changes. Five such surveys have been undertaken: February, May, and
September of 1991, and January and June of 1992.
Directional wave and current data have been collected offshore of the Perdido
Key Ranger Station, and a land-based weather station was installed at the Ranger
Station to monitor wind speed and direction, air temperature, and rainfall. A second
wave gage was installed near Caucus Shoal at the eastern end of Perdido Key in April,
1992, to investigate the sheltering effect of the shoal. Although the wave gages also
provide an estimate of tidal stage, an additional tide gage was installed at Ft. Pickens
pier, on Santa Rosa Island near Pensacola Pass. Locations for the instrumentation
are shown in Figure 3.3. Oblique photographs of the area were taken to qualitatively
document changes as they occurred.
Table 3.1 provides a history of the field work done to date for the physical mon-
itoring study; Figure 3.2 illustrates the temporal availability of the resulting data.
The following sections describe the data collection and analysis procedures. Details
not found here are available in Work et al. [1990a, 1990b, 1991a, 1991b, 1991c] and
Work and Dean . The wave and bathymetric data are also addressed in the
following chapter, because of their particular relevance to the problem of interest.
Table 3.1: Chronology of Perdido Key Data Collection Efforts.
Wading profiles (Gulf and Bay)
Sand samples, photos
Placement of nourishment material begins
Wave gage tripod and standalone gage installed
Tide gage with small stilling well installed
at Ft. Pickens Pier, Santa Rosa Island
Mechanical (analog) weather station installed
Large stilling well installed for Ft. Pickens tide gage
56 sand samples collected, to replace those destroyed or
not collected during pre-nourishment survey
Standalone wave data collection package retrieved;
fresh standalone package installed on tripod
Digital weather station installed
Standalone wave data collection package retrieved;
fresh standalone package installed on tripod
Placement of nourishment material completed
First post-nourishment survey:
Wading profiles (Gulf side)
Sand samples, photos
Standalone wave data collection package retrieved;
fresh standalone package installed on tripod
Ft. Pickens pier tide gage re-surveyed
Wading profile survey (Gulf side)
Standalone wave gage retrieved; tripod moved
Lightweight data/power transmission cable installed
Shore-connected wave gage installed on tripod
_________ Table 3.1: -continued.
Wave gage cable re-buried
Wave gage cable re-buried
Wading profile survey (Gulf side)
Shore-connected wave gage removed; cable cut
Standalone wave gage installed on tripod
Wind vane and anemometer replaced
Standalone wave gage removed from tripod
Fresh standalone wave gage installed on tripod
Wading profile survey (Gulf side)
Sand samples, photos
Ft. Pickens pier tide gage re-attached
Wading profiles and offshore bathymetry (Gulf side)
Installed heavyweight data/power cable for wave gage
Standalone wave gage removed from tripod
Shore-connected wave gage installed on tripod
Fresh standalone wave gage installed on tripod
Shore-connected wave gage replaced
Wading profile survey (Gulf side)
Replaced wind vane/anemometer
Replaced shore-connected wave gage on tripod
Replaced standalone wave gage on tripod
Replaced shore-connected wave gage on tripod
Removed standalone wave gage from tripod
Installed new standalone wave gage near Caucus Shoal
Wading profile survey (Gulf side)
Surveyed 7 profiles to wading depth on Bay side
Repeatability check for wading profiles
Z w o
// o I
.. U 0 o
- --- ,',', i A
~ --- 0
"^P- ^S 2
.^g ^5 o
7 ?? ; -
^I ^^ ^* s
Qoe ^ a -
The field data collection is an ongoing task; the results discussed in this disserta-
tion will cover the time period from November, 1989, through June, 1992. This allows
for the consideration of seven post-nourishment surveys and wave, current, tide, and
weather data from a 2-1/2 year time period:
3.1 Topographic/Wading Profile Surveys
A set of 33 beach profile transects was defined to monitor the evolution of the
nourished beach. Each transect originated at a State of Florida Department of Natural
Resources "permanent" benchmark monument and runs approximately perpendicular
to the local shoreline (Figure 3.3). Benchmark elevations and profile azimuths relative
to magnetic North were provided by the Department of Natural Resources (Table 3.2).
The monuments run the length of the study area and are separated by a nominal
distance of 300 m. For the pre-nourishment survey of November, 1989, 16 profiles
were also extended north into the bay inland of Perdido Key. An additional 7 profiles
were extended into the bay during the June, 1992, survey.
A tripod-mounted theodolite equipped with a magnetic compass was placed di-
rectly over each monument in order to establish the path to be followed during the
field survey. A wooden stake was used to mark the direction to ensure that later
surveys followed the same line. It is estimated, based on repetitive measurements,
that this method with the equipment described is accurate to within one degree of
the desired azimuth. The theodolite or a hand bearing compass was used to check the
azimuth defined by the wooden stake during later surveys to ensure that the stake
had not been moved.
The survey monuments at the site are typically found in or near the dunes, well
back from the post-nourishment waterline. A standard automatic level and a 7.5 m
telescoping, graduated, fiberglass leveling rod were used to measure elevations along
the profile. The "rodperson" traversed the profile, pausing at all significant changes
in slope for a reading to be made with the level. Elevations were recorded to the
Table 3.2: Coordinates, Elevations and Azimuths for DNR Monuments in Escambia
Monument Northing2 Easting Elevation Range Azimuth3
No.' (ft) (ft) (m, NGVD) (Degrees)
Table 3.2: -continued.
Monument Northing2 Easting Elevation Range Azimuth3
No.1 (ft) (ft) (m, NGVD) (Degrees)
R-55 489649.180 1102290.990 2.03 165
R-566,N 489603.500 1103328.000 2.48 165
R-57 489785.670 1104344.820 2.27 165
R-58b,N 489940.500 1105353.000 2.18 165
R-59 490080.500 1106356.500 1.91 165
R-60bN 490247.500 1107323.000 2.03 165
R-61,N 490350.500 1108298.000 2.68 165
R-62bN 490433.130 1109324.130 2.01 165
R-63b,N 490528.250 1110297.350 2.45 165
R-64b 490836.540 1111090.500 1.82 170
R-65b 491114.930 1111728.450 2.13 105
R-66Ab 492016.000 1112143.000 2.68 105
R-67b 492997.990 1112292.510 3.08 90
Notes: 1) Profiles surveyed by boat and wading are marked by a superscript b.
Profiles marked by superscript w were surveyed to wading depth only.
Superscript N marks those profiles extended north into the bay.
2) Monument coordinates are in units of feet for consistency with
common information sources.
3) Azimuths are measured clockwise from magnetic North and
correspond to the line-of-sight of an observer at the monument
looking offshore along the survey line.
nearest centimeter. Once the rodperson neared the waterline, a turning point was
established so that the level could be moved to a lower elevation near the waterline.
With a few exceptions, only one turning point per profile was required (Figure 3.4).
The leveling loop was not closed, but the waterline elevation and distance were noted
for purposes of error checking.
Distances were measured using a fiberglass tape, reading to the nearest decimeter.
Offshore distances exceeding 50 m were measured using a simple optical rangefinder.
Because of the nonlinear scale that is read by the rangefinder operator, the relative
error in the measurement increases with distance, but calibration tests indicated that
this error was within 2.5% for the distances measured in the field (<300 m). This
results in a maximum horizontal error of 7.5 m for distances measured by the optical
To maintain correct alignment along the profile while in the water, the rodperson
would align himself with two orange traffic cones placed along the upland portion of
the profile a distance apart and at different elevations. Strong currents occasionally
made exact alignment impossible, but the primary effect of improper alignment is a
"stretching" of the measured profile due to overestimates of the distances. The mild
bathymetric gradients in the longshore direction reduce the vertical error.
The rodperson would continue his traverse of the profile until the numbers on the
graduated rod became indistinguishable through the optics of the level. The profiles
measured in the manner described above generally terminated near the -5 m contour,
which fortunately lay near the toe of the beach nourishment material. The project site
was surveyed in this manner three to four times per year after placement of the beach
nourishment material. Figure 3.5 provides an example of a "typical" profile at the
site before and after nourishment. For this example, both the nearshore nourishment
and the offshore, "profile", nourishment changes are evident.
3.2 Hydrographic Surveys
Hydrographic surveys were conducted at nominal intervals of one year. An 8
m power boat was equipped with a Motorola Mini-Ranger Falcon 484 microwave
rangefinder and an Innerspace Technology Model 441 acoustic depth sounder and
digitizer for the measurement of distances and depths, respectively. In a test of four
currently feasible hydrographic survey techniques, this method was found to be the
least accurate (in terms of repeatability, with errors in the vertical of up to 0.30 m)
but fastest scheme (Clausner et al. ). Errors will be discussed in a section to
GJiU A l
1a uated n
..... Level and Optical / (7.5 m)
.'.. .... -*** \ -*' -*: : S- -
. ............: 0 ^ ^ r
On Board Data
. . .
Hydrographic Survey Technique
Figure 3.4: Generalized profile geometry and survey methodology.
Distance from Monument (m)
Figure 3.5: Surveyed beach profiles at R-58, Perdido Key.
The boat operator would maintain proper alignment by sighting two large, orange
flags placed at different elevations on the profile of interest. The profiling would
begin in the shallowest depth reachable by boat in order to provide maximum overlap
between wading profiles and boat data. The profile was extended offshore until the
depth was at least 8 m or the distance offshore at least 800 m. Distance was usually
the controlling factor, because of the very mild slopes offshore of the -5 m contour at
Perdido Key. The hydrographic survey technique is illustrated in Figure 3.4.
The electronic rangefinder system includes a battery-powered, shore-based "slave"
unit, and a "master" unit aboard the boat. Distances are computed from the time
required for a microwave signal to travel between the two units. A small IBM-
compatible laptop computer receives the signals from both the rangefinder and the
fathometer and stores data on an internal hard drive at a sampling rate of 1 Hz. With
a normal boat speed of 5 kt, this yields one data point every 2.5 m.
With the scheme employed, there is no way to determine how accurately the boat
followed the desired path for each profile. A second shore-based slave unit would
assist the boat operator in staying on-line and provide an additional quality check
by permitting computation of the distance between the desired and actual boat path.
Data points that were found to be too far "off-line" could then be excluded from
the result. This scheme requires a more sophisticated data acquisition program and
a faster computer to handle the additional data, so was not pursued for the present
The fathometer simply measures the time required for a signal to travel from the
transducer down to an acoustically reflective surface and back. The speed of sound
through the water, presumed not to vary temporally, vertically, or horizontally within
the survey domain, serves as the calibration constant and was adjusted immediately
before and after the survey by performing a "bar check." This requires the placement
of a large, flat object (in this case an octagonal "stop" sign) at a known depth beneath
the fathometer transducer and adjusting the calibration until the reported and known
depths agree. Bar checks done before and after the survey allow for correction for
temporal variation of the calibration factor; this was found to be unnecessary. Depths
measured by this system are estimated to be accurate to 3 cm when the system is
calibrated properly, exclusive of motion of the fathometer transducer (manufacturer's
Depths measured by fathometer must be converted to a known vertical datum by
removing the tidal signal and accounting for the known distance from the transducer
to the water surface. The survey boat will also exhibit some "squat", a tendency for
the stern to drop as speed increases, but this was considered negligible.
A minimum of two passes were made by the boat along each profile, in order to
cancel out the effects of waves. Fortunate timing allowed each hydrographic survey to
coincide with minimal seas. This often allowed the boat to cross over the breakpoint
bar, maximizing the overlap between the data collected by rod and level and that
collected by boat. This also reduced reliance on the optical rangefinder, which is the
least accurate of the devices used for distance measurement.
Tidal elevations during the hydrographic survey were measured using a simple
stilling well placed in shallow water within the survey area. Readings were recorded
manually every 15 minutes. The tidal stage was then removed from the fathometer
data to compute elevations relative to NGVD (National Geodetic Vertical Datum).
The raw bathymetry data resemble a noisy signal with a well-defined trend (Fig-
ure 3.6 illustrates a representative result). The first step in the data analysis was to
remove any obvious "spikes" in the reported distances and depths. The spikes were
defined as existing when the bottom slope and change in elevation between successive
data points simultaneously exceeded critical values. Spikes were often the result of
only one bad data point, so this method was effective for their removal.
Noise was filtered by placing a 10 m-wide "boxcar" window over each data point
and assigning the average of all values within the window to that point, conserving
sediment. Offsets for tidal stage and slave unit position were then added to the
depths and distances, respectively. The wading profile and boat data could then be
"patched" together and the results inspected visually for agreement (Figure 3.7).
0 100 200 300 400 500 600 700 800
Distance from slave (m)
Figure 3.6: Raw bathymetry data for R-50, October, 1991. Two passes by boat are
100 200 300 400 500 600 700
Distance from monument (m)
Figure 3.7: Bathymetry data for R-50, after spike removal, smoothing, adding proper
vertical and horizontal offsets, and patching with wading profile data.
3.3 Waves, Currents, and Tides
A submersible, self-contained electronic wave gage was installed in 6 m of water
in January, 1990. The gage is bolted to a heavy, steel, tetrahedron-shaped frame,
held in place on the sea floor by jetted piles (Figure 3.8). The frame has remained
on-site since its placement, with the gage being replaced periodically for servicing.
Housing for tape driver,
Figure 3.8: Schematic of wave gage and tripod.
Two types of wave data collection packages have been used: one self-contained
and the other connected by a cable to a shore station. The data collection and analysis
methods are the same for both models. A watertight cylinder, 1 m in length, houses
batteries, circuit boards, and some form of data storage device. Magnetic tape drives,
hard disk drives, and random-access memory (RAM) storage have all been used at
different times. An electromagnetic current meter is mounted externally on the top
of the cylinder; a pressure transducer is mounted on either the top or the bottom,
depending on the design.
Data were collected during the early stages of the project using self-contained
wave gages. These units must be replaced roughly every three months for battery
replacement and data retrieval. A serious drawback to the use of this type of unit
is that malfunctions are not evident until the gage is retrieved. For this reason, a
data and power transmission cable was installed, leading from the wave gage back to
a shore station equipped for modem communications. The shore power maintains a
charge on the batteries within the wave gage housing, and data can be downloaded
as often as desired over standard telephone lines. Additionally, parameters such as
sampling frequency and data burst interval can be altered without retrieval of the
gage. Marine fouling is a problem with either type of gage, so neither unit could be
Bursts of pressure and velocity data are collected once every six hours, using a
sampling rate of 1 Hz for 1024 seconds. This provides adequate resolution for the
waves of interest at the site by yielding a Nyquist frequency of 0.5 Hz, meaning
that waves having periods less than 2 seconds cannot be resolved. In practice, high-
frequency noise found in both the pressure and velocity time series imposes a more
restrictive limit. This is addressed further in the following chapter.
After checking the quality of the data, removing spikes, and correcting flat spots
in each time series, the pressure record is analyzed to compute the energy spectrum of
the sea surface using spectral analysis methods and linear wave theory. Each 102-1-
point time series (pressure, u-velocity, v-velocity) is treated as weakly stationary
and is time-averaged to determine and remove mean quantities, and then broken up
into smaller segments for computation of the energy spectra. Tidal stage is assumed
constant for this 17-minute interval, a reasonable assumption given the predominantly
diurnal tides and minimal range at the site.
The first 128 points of each time series are passed to a numerical Fast Fourier
Transform routine, allowing the computation of the (auto-) spectral density function
(energy spectrum) from Equation (3.1). The terms necessary for estimation of wave
direction are simultaneously computed. Only the pressure time series is used to
compute wave heights and periods; velocities are used for determination of wave
direction. A cosine2 window (or Hanning or Tukey window, Childers and Durling
) is applied to the first 10% and last 10% of the 128 points in each time series
segment during the analysis to minimize spectral leakage.
2(a? + b?) 1
E, (f) (3.1)
Here E,(f,) denotes the energy density of the free surface displacement at fre-
quency f, (Hz), Af is the frequency resolution, given by the inverse of the number
of records (i.e. 1/128), a, and b; are the real and imaginary components, respectively,
resulting from the Fourier transform of the pressure time-series, and r,(fi) is the
pressure response factor, given by linear wave theory:
= cosh k(h + z) (3.2)
where k = k(fi) is the wavenumber, h is the mean water depth, and z is the vertical
coordinate, with origin at the mean water level and positive upwards.
With the energy spectrum computed based on the first 128 points in the 1024-
point time series, the window is shifted 64 points and another spectrum is computed
based on the 128 points lying within the shifted window. This is repeated until the
end of the 1024-point series is reached. If each of the 1024 points passes the quality
test, the resulting spectrum is thus an average of fifteen spectra all computed from a
128-point subset of the 1024 available points.
The significant wave height, H,, is computed by integrating the energy spectrum
and applying Equation (3.3):
H, = 4.01/mo (3.3)
where mo is the zeroth moment of the energy spectrum, i.e. the area under the
spectrum. Equation (3.3) requires a narrow-banded spectrum and a Gaussian sea
surface displacement. These assumptions and the use of linear wave theory are not
strictly valid in shallow coastal waters, but are commonly invoked in practice.
During evaluation of the area under the spectrum, mo, energy found at frequencies
higher than a pre-defined cutoff frequency (0.33 Hz) is attributed to noise in the
instrumentation and therefore neglected. The effect of the noise and the chosen
cutoff frequency will be illustrated in the next chapter.
Computation of the directional wave spectrum largely follows the method of
Longuet-Higgins et al. . The directional spectrum, E(fi, 0), is assumed ex-
pressible as a truncated Fourier series in terms of wave direction, 0:
A 2 2
E(fi, 0) = + E A,(fi) cos nO + E B(f,) sin nO (3.4)
where the five Fourier coefficients may be expressed in terms of the available auto-
and cross-spectra, Spp(fi), S.,(fi), Sp(fi), S,,(f/), and S,,(fi). The auto-spectrum
of the v-component of velocity, S,,(fi), provides redundant information, but could
be used to obtain a second estimate of the one-dimensional energy spectrum, E,((fi).
Several subtleties present themselves when attempting to employ Equation (3.4)
to describe the directional spectrum. The result is that a directional spreading func-
tion, H(f, 0), is often introduced, such that:
E'(f, 0) = E(f)H(f, ) (3.5)
Several forms have been proposed for the spreading function, usually describing
H(f, 0) as a non-negative function, symmetric about some central direction, 0o, and
decreasing with increasing values of 10-0oI. A central wave direction could be assigned
to each frequency component, but for the present study it was desired to obtain one
representative direction for the entire spectrum. This direction was computed using
0o = tan-' [ (3.6)
where fm is the modal frequency, i.e. the frequency corresponding to the peak of the
The magnitude and direction of any quasi-steady currents, and tidal stage were
also computed during the wave data analysis procedure. Tidal stage and steady
currents are by-products of the de-meaning of the 1024-point bursts of pressure and
velocity data, respectively. The pressure gage is not vented, meaning that variations
in atmospheric pressure appear as changes in water surface elevation. This does not
corrupt the wave data analysis, however, because of the large difference in time scales
between the relevant gravity wave periods and atmospheric pressure fluctuations.
Computation of the direction of wave propagation and the direction of the mean
current requires knowledge of the orientation of the data collection package in the
field. The orientation was measured for each unit immediately after installation and
before removal by the diver hovering above it in the water column and sighting with
either a diver's magnetic hand bearing compass or a digital compass configured for un-
derwater use. Several of the packages used allow for only one orientation, so repeated
measurements aid in assessing the error in the measurement, which is estimated rather
subjectively at 10.
The results of the wave/current monitoring program and some additional sub-
tleties of the analysis procedure are discussed in a subsequent chapter. The wave
height, period, and direction data will be of primary concern.
3.4 Sand Samples
Close to one thousand sand samples have been collected to date to document
the spatial and temporal variability in sediment size within and bordering the study
area. Samples were collected during the wading profile surveys and may be considered
"grab" samples of surface sediments.
Many of the sampling locations varied slightly with time, since their positions were
defined by the locations of elevation contours. Samples were taken along each profile
in the dunes, at the berm, on the beachface, and at the -1 m and -2 m contours.
An additional sample was added at the "mid-beach" position (halfway across the
"'new" beach) after completion of the beach nourishment work. Samples were also
collected at the -5 m and -8 m contours (or at the end of the survey line) during
each hydrographic survey. The pre-nourishment survey of November, 1989, included
samples from the north (bay) side of the island, on the beachface and at the -1 m
and -2 m contours.
Samples from the -1 m and -2 m depths were collected by a swimmer using a
small can as a scoop. The -5 m and -8 m/end-of-line samples were taken from a
boat by lowering a steel bucket to the bottom and dragging it until sufficiently full.
The horizontal positions of the sampling points for the -5 m and -8 m samples
were established using a Loran navigation system. During the pre-nourishment sur-
vey, the boat crew noted the points on each profile where the depth reached 5 m or 8
m and stored the latitude and longitude as a waypoint in the Loran set. Some profiles
never reached the -8 m contour, so the coordinates of the offshore limit of the survey
line were stored instead. The "-5 m" samples either represent the approximate -5 m
contour or an intermediate point along the profile. The same Loran coordinates were
used for sampling during later surveys, with the depth recorded for each sample, al-
though the depths were not corrected for tidal stage. When used repeatedly to locate
the same points in this manner, accuracies of 50 m are commonly cited for Loran
navigation systems. Table 3.3 presents the location of each offshore sample and the
nominal and measured depths.
All sand samples were placed in labelled cloth sacks. They were dried in an oven
at 400C before sieve analysis. Organic content of most of the samples appeared neg-
ligible, although no attempt was made to quantify the percentage. Most samples
consisted primarily of quartz sand, with the shell fraction increasing near the water-
line. Some samples that resembled a fine "ooze" were taken after beach nourishment
at the -5 m and -8 m sampling points near the center of the beach nourishment
project. These hardened upon drying and were crushed for analysis.
Twelve sieves were used to determine the grain size distribution for each sample
(six sieves were used for the pre-nourishment samples). Sieve numbers and sizes are
listed in Table 3.4.
After sieving, mean and median grain sizes, sorting index, skewness, and kurtosis
were all computed for each sample. Except for the fine samples offshore of the center
of the nourishment area, the median grain sizes exhibit little variation in the cross-
shore direction (see Table 3.5 and Figure 3.9). Some fining is evident offshore, and the
coarsest sediment is found on or near the beachface. Longshore variation in sediment
size was also small, and the size characteristics of the "new" and native sands were
found to be similar. Sediment size statistics are discussed in more detail in Work et
al. [1991a] and Work and Dean .
3.5 Weather Data
A mechanical weather station was installed at the Perdido Key Ranger Station in
January, 1990. This unit recorded air temperature and wind speed and direction in
analog form on a strip chart. It was replaced in June, 1990, by an electronic weather
station that remains at the site. The electronic station includes a series of sensors
for wind speed, wind direction, air temperature, and rainfall. The wind sensors were
Table 3.3: Locations and depths of offshore sand samples.
Range Nominal 11/89 9/90 10/91 Latitude Longitude
No. Depth (m) Depth (m) Depth (m) Depth (m) (Deg., Min.) (Deg., Min.)
Range Nominal 11/89 9/90 10/91 Latitude Longitude
No. Depth (m) Depth (m) Depth (m) Depth (m) (Deg., Min.) (Deg., Min.)
R-58 5 5 5.0 5.2 30 18.69 87 20.09
8 5.8 5.8 4.6 30 18.45 87 20.02
R-60 5 4.5 4.6 4.6 30 18.73 87 19.71
8 5.6 5.6 4.6 30 18.51 87 19.66
R-61 5 4 4.3 4.3 30 18.75 87 19.53
8 5.0 5.2 5.2 30 18.53 87 19.47
R-62 5 3.5 3.8 4.0 30 18.78 87 19.34
8 4.6 4.4 4.6 30 18.55 87 19.27
R-63 5 3 2.4 3.0 30 18.84 87 19.14
8 3.5 3.5 3.4 30 18.58 87 19.05
R-64 5 2 5.5 2.4 30 18.83 87 19.02
8 3.1 7.1 3.0 30 18.67 87 18.99
R-65 5 2.7 5.5 4.0 30 19.03 87 18.70
8 8 9.4 7.3 30 19.00 87 18.61
R-66A 5 5 5 4.6 30 19.21 87 18.72
8 8 8 7.3 30 19.20 87 18.66
R-67 5 5 8 4.6 30 19.40 87 18.70
S 8 5 8 7.0 30 19.30 87 18.69
Notes: 1) All coordinates obtained during pre-nourishment survey
of October 28-November 3, 1989, except those for Ranges
66 and 67, which were taken during September, 1990 survey.
2) Measured depths are not corrected for tide, but merely indicate
the depth beneath the boat at the time of sampling.
Table 3.4: Sieves used for grain size analysis.
Sieve Number Sieve Opening
U.S. Standard (mm)
Table 3.5: Cross-shore distribution of grain size at various contours (Gulf side).
Longshore standard deviation is given following median grain size.
Location Average D50 (mm) Average D50 (mm) Average D50 (mm)
November, 1989 September, 1990 October, 1991
Dune 0.370.04 0.360.04 0.340.03
Berm 0.390.05 0.380.03 0.360.04
Beachface 0.390.05 0.400.06 0.360.06
-1 m 0.380.05 0.360.05 0.340.04
-2 m 0.300.03 0.350.07 0.280.03
-5 m 0.320.03 0.310.06 0.300.04
-8 m/End of line 0.320.05 0.3110.06 0.3110.06
D50 vs. Feature/Contour
0.38 .... ............. ------ ......
0 .3 8 ................................................ ......................... ......... .
0 3" ....... ------- .... .............. X -- ............... .
S0.36 -------A N----K-
S 0 .34 ......... ................................................. ......... ..........................................
Dune Mid-Bch Bern
1 Bchface -1 m
2 m -5 m -8 m/EOL
Figure 3.9: Cross-shore distribution of grain sizes.
mounted clear of the roof of the Ranger Station at an elevation of 11.7 m NGVD,
and the temperature sensors were shielded from direct sunlight.
Each sensor outputs either a pulse or a voltage to an analog-to-digital conversion
board. The digitized signal is sent to a small computer, which stores one set of
readings per hour in random-access memory. Once per day, the computer phones a
host computer and attempts to dump the stored data via modem. If successful, the
memory is cleared and data acquisition re-started. If unsuccessful, data acquisition
continues, and a new attempt at data transfer is made the following day.
Weather data are presented and compared with similar data collected by the
National Weather Service in Pensacola in Work et al. [1991b]. The weather data were
not used in the development or calibration of the beach evolution models presented
here, so will not be discussed further.
3.6 Ft. Pickens Pier Tide Gage
A mechanical tide gage was installed on Santa Rosa Island at the Ft. Pickens
pier (Figure 3.1). The gage uses a stilling well to damp out high frequency motion
and a spring-driven clock to drive a pen across a rotating drum covered with paper.
The gage was surveyed to refer all elevations to NGVD. The data do not form an
important part of the study because of the numerous gaps in the record and the
location of the gage inside Pensacola Bay.
3.7 Assessment of Errors
Many possible sources of error exist in the field data collected, both random
and systematic. A few were mentioned during the discussion of the data collection
methods. In several cases, it is possible to devise a scheme by which these errors
can logically be minimized. Most of the errors have been either treated during data
analysis or presumed negligible. It is important, however, to identify all significant
sources of error and attempt to assess the magnitude of each. Human errors, such as
reading a scale incorrectly, will not be addressed; such errors are often large enough
that they are easily identifiable, and there is rarely a rational way to compensate for
them. Emphasis will be placed on errors in the bathymetric and wave/current data,
since they are most relevant to the focus of this study.
The beach profiles are defined by horizontal and vertical distance measurements,
each having a different error, and each containing several contributions to the total
error. Some of the errors are proportional to the magnitude of the distance being
measured, while others are independent. Fortunately, most of the errors are not
cumulative. Because of the mild slopes involved in beach profile surveying, vertical
errors are particularly troublesome.
The portions of the beach profiles surveyed by rod and level contain roundoff
errors of up to 0.5 cm in the vertical and up to 5 cm in the horizontal. These
are presumed to be random, with zero mean, and will therefore cancel. Insufficient
tension in the tape will lead to overestimation of distances. Repeated measurements
of the same distance indicated that an error of less than 1.0% could be expected
when using the tape. Distances measured by the optical rangefinder were found to
be within 2.5%. The stated "probable range error" of the microwave rangefinder
system is 2 m, irrespective of range (manufacturer's literature). If the boat maintains
the correct alignment along the profile, the maximum horizontal error in the beach
profiles surveyed by boat is thus 3 to 4 m. Good overlap between wading profile data
and data collected by boat provided some verification and confidence in measured
Incorrect alignment along the profile introduces some error by leading to some
"stretching" of the profile, as distances are overestimated (assuming that the correct
alignment is perpendicular to the local shoreline). This effect is minimal until well
offshore. In some cases, large apparent changes in the profile appear in the dunes
when the profile line is not repeated exactly from one survey to the next. This is
simply because of the very strong topographic gradients in and near the dunes; i.e.
they are often steep. Fortunately, longshore bathymetric gradients offshore are small
for most of the study area. Therefore the primary effect of the boat being off-line
during the survey is to overestimate distances by a factor of 1/cos 0, where 0 is the
angular deviation of the boat from the correct path. With the bathymetric surveying
technique employed, there is no way to determine the magnitude of this error or
correct for it. The averaging involved in making several passes over the profile should
help remove errors of this type.
Vertical errors include those due to not holding the leveling rod vertical, possible
settling of the level or rod during the survey, and improper calibration of the level.
These sources were considered negligible, as periodic calibration of the level and
appropriate caution during each survey should have rendered them so.
An experiment was conducted in June, 1992, to assess the repeatability of the
wading profiles. The equipment was deployed at R-46 and the profile surveyed to
the maximum depth possible by swimming. The equipment was taken down, all
crew tasks were swapped, and the process repeated. This resulted in two surveys
of the same profile within one hour, during which wave activity was minimal. Any
differences between the two surveys may thus be assumed error.
The results of the repeatability experiment are shown in Figure 3.10. The two
realizations of the profile are superimposed and are nearly coincident. The difference
between the two surveys is also shown, plotted using a different vertical scale. Not
surprisingly, the maximum difference occurs near the dunes, where cross-line gradi-
ents are largest. This is not important for the present study, where changes in the
submerged portions of the profile are of primary interest. The average difference be-
tween the two profiles is slightly more than one millimeter. This average difference
will be important for any computations involving a sediment mass balance; the RMS
difference (2.6 cm) is more representative of the accuracy with which the elevation
of any particular point may be measured. Wave conditions during this test were
favorable, but the same may be said for all but two of the surveys.
Perdido Key: June, 1992
R46 Accuracy Check
4 ............ ....................................................................................................................... 1st T rial
3- -0.2 -
3 ................ ........... ...................................................... ..................... ..2...
> 2------\.._^ ^----------- ---------:---- ^ ^ .....
0 0.1 Difference
0 .- ... .............-................. ...... 0 0g-
S-2- --0.1 RMS=2.6em
----------------------------------------- F-. 1 I0S=3
0 50 100 150 200 250 300 350
Distance from Monument (m)
Figure 3.10: Repeatability test at R-46, June, 1992.
The largest potential errors in the surveys in the vertical are found in the regions
surveyed by boat. The accuracy of the tide data for correction of depths measured by
boat to a common vertical datum is critical. The tide gage was surveyed both before
and after the survey to establish its elevation and determine if it had settled during
the survey. Tide readings at fifteen-minute intervals provided sufficient temporal
resolution of the tidal stage throughout the survey. Error in the computation of the
elevation of the water surface at the tide gage is thought to be small, on the order of
one centimeter. Spatial variation in the instantaneous water surface elevation within
the survey area could be much larger than this (Blair ), but without additional
tide data, there is no reasonable way to account for this possibility. The small tidal
range at the site (0.3 m at Pensacola Pass, U.S. Department of Commerce ) and
their chiefly diurnal nature help to minimize any errors introduced in this manner.
Vertical or angular motion of the fathometer transducer, arising from wave-
induced motion of the survey boat, introduces some error in the measured depths.
Heaving under- or overestimates depths by the instantaneous vertical excursion of
the motion; rolling or pitching will always overestimate depths by the inverse of the
cosine of the angular displacement. Commercial systems for the measurement and
recording of the heave, pitch, and roll of the survey boat have recently become avail-
able and would allow compensation for these effects during data analysis (Downing
and Fagerburg ). Such equipment was not available for the present study. The
repeated passes over each profile during each bathymetric survey should cancel out
the effects of the heaving motion, while the minimal wave activity found during each
survey should render the effects of both the heaving and the angular displacement
Calibration of the fathometer was described earlier. The calibration was assumed
to be valid throughout the survey area, and the speed of sound through the water
column was considered independent of depth. Improper calibration would lead to
a systematic error, always over- or underpredicting depths, while density variations
would lead to a more random error. The accuracy of the fathometer, when used in the
manner described, was previously estimated at 3 cm, exclusive of errors arising from
heave, pitch, and roll of the survey boat. This assumes proper calibration; the largest
component of vertical error, however, is thought to be fathometer calibration. This is
because the bar check is typically done at depths of 1.5 to 3 m, with an uncertainty of
3-4 cm, or roughly 1-2%. During each survey, wading profiles were extended offshore
as far as possible to provide maximum overlap for quality checking.
The depths and distances recorded on board the survey vessel are collected inde-
pendently and both reported as functions of time. "Spikes" occasionally appear in
the depth or distance vs. time curves, as was shown in Figure 3.6. Spikes in the depth
curve always result in smaller reported depths and are caused by bubbles, fish, or tur-
bid water beneath the transducer. Excessive gain and turbid water are thought to be
the cause of the noisy offshore data collected in the middle section of the monitoring
area during the September, 1990, survey. Spikes in the distance curve are usually the
result of short time intervals when no communication is made between the slave and
master units of the range-finding system. Both types of errors are easily spotted and
removed during data analysis if the number of bad data points is not too large.
Upon examination of three sets of bathymetric survey data (November, 1989;
September, 1990; and October, 1991), some systematic error, most likely due to
calibration inaccuracies, was found. The data collected by boat for the October,
1991, survey were found to lie vertically between and nearly equidistant from the
profiles from the November, 1989, and September, 1990, surveys. The data collected
by boat were adjusted to remove any average error between the offshore portions of
successive surveys at one location. More sophisticated schemes for error removal could
be devised, such as increasing or reducing depths by a specified percentage, rather
than a single depth adjustment, but there will always be a degree of subjectivity
in selecting the region over which the profiles should agree, and this subjectivity
renders more complicated error correction schemes unjustifiable. Adjustment of the
September, 1990, bathymetric data was more difficult because of the noise found in
several of the data files.
The result of the vertical adjustment of the hydrographic data was an average
raising of the November, 1989, profiles by 14 cm (maximum of 29 cm; standard devia-
tion of 6.7 cm), and a lowering of the September, 1990, profiles by 8 cm (maximum of
20 cm; standard deviation of 6.9 cm). One of the October, 1991, profiles was lowered
by 6 cm, the rest were left unmodified. Note that only the data points collected by
boat were modified in this manner; wading profile data were unaffected.
It could be argued that the average of all three data sets should be computed,
and then each profile adjusted to agree with the computed average offshore. The
chosen method instead weights the most recent survey more heavily than the previ-
ous surveys, assuming that the quality of the data improved with experience in the
data collection process. The difference between the two adjustment methods is not
particularly relevant when considering sediment transport, since profile differences
are much more important than absolute depths.
Saville and Caldwell  and Nordstrom and Inman  both discussed the
vertical accuracy of beach profiles surveyed using acoustic fathometers. Based on
repetitive surveys of the same beach profile over a short time interval, Saville and
Caldwell  concluded that the probable vertical error in one profile is 2.5 cm,
but also conceded that this was probably an optimistic estimate. All measurements
were made in one day, tide range was minimal, no crew changes were made, and the
surveying equipment was mobilized only once. Nordstrom and Inman  compared
profile changes measured by fathometer and permanent vertical reference rods placed
at various locations in the surf zone and concluded that fathometer measurements
were accurate to 30 cm. The results of the Perdido Key monitoring program suggest
that the error estimated by Nordstrom and Inman  is a reasonable upper bound
for the absolute vertical error in the boat data. It is the author's belief, however,
that adjusting the profiles to agree in the offshore segments can reduce this error by
nearly an order of magnitude, leaving a vertical error of 3-5 cm.
The wave and current data have two primary sources of error: calibration and
orientation. The maximum error in the orientation of the data collection package
was earlier estimated at 100. Wave and current directions are directly dependent on
this orientation. This is a serious problem, because a ten degree difference in wave
direction can have a significant effect on nearshore sediment transport.
Errors in the calibration of the pressure sensor or current meter will affect the
computed mean quantities, such as tidal stage and mean currents, and significant
wave height. Calibration problems will in general not corrupt computed wave or
current directions or the wave period corresponding to the spectral peak. Pressure
sensors were calibrated prior to installation; the calibration factors supplied by the
manufacturer of the current meters were used for lack of suitable calibration facilities.
With one exception, the various current meters each reported similar steady currents
when deployed in the field, providing some confidence in the supplied calibration
Electronic noise in the pressure sensor and the hardware "downstream" from it
will lead to an overestimate of the energy in the wave spectrum. A simple experiment
was conducted to determine the amplitudes and frequencies associated with this noise.
The experiment is discussed in the following chapter.
Errors in the weather data are limited to calibration factors, although wind di-
rection is dependent on accurate alignment of the wind vane in the field. The error in
this angle is estimated at 50. The remaining sensors (wind speed, rainfall, tempera-
ture) were calibrated prior to deployment and presumed invariant. Two temperature
sensors were deployed side-by-side; they typically agreed within 1 F.
ANALYSIS AND INTERPRETATION OF FIELD DATA
The primary goal underlying much of the field program was to obtain data docu-
menting the sediment transport processes at a nourished beach, as well as the forces
influencing the processes. The sediment transport at the site has not been measured
directly, but can be inferred from the measured changes in the bathymetry. Many
environmental parameters have been monitored, but the data provided by the wave
gage (incident wave spectrum, water level, and mean currents) are thought to include
all the dominant environmental forces modifying the bathymetry at the site. The
goal of this chapter is to extract from the field data all qualitative and quantitative
trends that are important to the sediment transport problem of interest.
4.1 Wave, Current, and Tide Data
Analysis of the data from the offshore P-U-V (pressure and two components
of horizontal velocity) gage was described in the previous chapter. This analysis
provided, for every six hours of satisfactory operation, one realization of each of the
following parameters: energy spectrum (from which significant wave height, H,, and
wave frequency corresponding to the peak of the energy spectrum, Tp, are computed),
wave direction corresponding to the peak of the spectrum, magnitude and direction
of any mean current, and tidal stage.
Some quality checking was done in order to establish confidence limits on the data
collected by the wave gage. The first issue addressed was the level of noise in the
pressure sensor and the current meter and the associated electronics. To investigate
this, one of the wave gages was activated in the lab, in still air, and a 1024-point burst
of data collected. Figure 4.1 shows the resulting energy spectra of the pressure and
velocity signals, Ep(f,) and E,(f2), respectively. The spectra shown in Figure 4.1 have
not been transferred to the surface to enable computation of a significant wave height.
To determine the energy spectrum of the sea surface, each frequency contribution must
be divided by the square of its response function:
E,7(fi) p) ( 4f )
E,((,) = (4.1)
E,( f) = (4.2)
where Kp(fi) = cosh ki(h + z)/coshkih and K,(fi) = 2 fi cosh k,(h + z)/ sinh kh.
Since the response functions Kp(f) and x,(f,) are much smaller for high frequencies,
noise at high frequencies will be much more troublesome.
The noise shown in Figure 4.1 was transferred to an assumed free surface, using
Equations (4.1) and (4.2) with an assumed water depth of 6 m, to yield the surface
energy spectra shown in Figure 4.2. The energy in the velocity signal is seen to be
minimal and will not be of further concern since it is not used in the wave height or
wave frequency estimation.
The next step was to compute the significant wave height, H,, based on the
energy spectrum of the free surface, E,(fi), derived from the pressure time series.
The significant wave height is assumed to be equal to 4.01 times the square root of
the area under the spectrum. The cutoff frequency for evaluation of this area is clearly
important: with a cutoff frequency of 0.5 Hz (the Nyquist frequency), the significant
wave height for this case is 0.14 m, while a cutoff frequency of 0.33 Hz yields H,=0.05
m. Since the data collected in the field were analyzed with a cutoff frequency of 0.33
Hz, it was assumed at this stage that the contribution of the noise to the reported
significant wave heights was negligible.
Spectra of Noise
(1 Hz Sampling Rate)
4.5E -04 ....................................... .... .. .......... Pressure
4 .0 E -0 4 ..... ..................... ...............................................................
3.5E-04- -...-----.................. Velocity
3 .0 E -0 4 ..............................................................
2.5E -04 ........... ............. ....................
0 0.05 0.1 0.15 0.2 0.25 0.3 0.35 0.4 0.45 0.5
Figure 4.1: Spectra of noise in pressure sensor and current meter, based on 128 points
sampled at 1 Hz, Hanning window. Gage in still air in lab.
Spectra Transferred to Free Surface
1 .6 E -0 2 .................................................................................................... P re ssu re
1 4 E 0 2 --- --- -
1.4E-02 ........................................................................................ ..........Velocity
8 .0 E -0 3 ...........................................................................................
0 0.05 0.1 0.15 0.2 0.25 0.3 0.35 0.4 0.45 0.5
Figure 4.2: Spectra from previous figure, transferred to free surface using an assumed
depth of 6 m.
The noise can also bias the computation of the wave period corresponding to the
peak of the energy spectrum, Tp. If most of the signal is due to noise, and the noise is
distributed fairly evenly over all frequencies (i.e. white noise), then the reported value
for Tp will correspond to the cutoff frequency, regardless of the chosen value. Wave
and current directions can also be affected by noise: if the signal is very small, the
computed directions lose their significance. When interested in average values, these
problems can generally be avoided by weighting each quantity (Tp, wave direction, or
current direction) by the magnitude of the parameters used to estimate them. The
computed waves and mean currents may both be thought of as vectors, each having a
magnitude and direction. The confidence level of the computed directions decreases
as the magnitudes of the vectors decrease.
Some data files revealed failure of the current meter while the pressure sensor
was still functioning properly. Often this was attributable to fouling of the current
sensor by marine organisms. For this situation, only wave height and period data are
available. Wave direction is not available.
Although the results above suggested that the effect of the noise was negligible,
a second time series was generated by sampling the pressure at 100 Hz. The idea
was to determine if high-frequency noise was being aliased and appearing at lower
frequencies. The spectrum of this pressure time series is shown in Figure 4.3. A
dominant signal is evident at 40 Hz, most likely arising from the power supply or
amplification circuitry. Based on this result, sampling at 80 Hz or more and then
computing the significant wave height using a cutoff frequency lower than 40 Hz
could reduce the influence of noise on the result. This approach would greatly increase
storage requirements, however, and would not have a large effect on the results. A
much larger range of frequencies would be included in the analysis, but most of these,
corresponding to wave periods less than two seconds, are not of interest and can't be
scaled up to the free surface with confidence due to large amplification factors.
Only one gage was tested in the manner described above. Other gages might
display different noise characteristics, but the one tested is thought to be fairly repre-
sentative. The tests essentially investigated only the effects of electronic noise. Since
the gage itself modifies the flow that it is monitoring, hydrodynamic noise could also
introduce some error into computed wave and current characteristics. The magnitude
of this error cannot be assessed with the available data, but is thought to be relatively
Spectra of Noise in Pressure Signal
(100 Hz Sampling Rate)
I I I T I I I
5 10 15 20 25 30 35 40
Figure 4.3: Spectra of noise in pressure sensor, based on 1024 points sampled at 100
Hz, Hanning window. Gage in still air in lab.
Long-term behavior of the wave climate (seasons to years) at the site will be
important to describing the evolution of the beach nourishment project. Observa-
tion of the wave climate at the site for 2-1/2 years has provided some insight into
1... .. ... ..
. .. i i -
the typical conditions and seasonal trends at Perdido Key. Tables 4.1 through 4.3
summarize the monthly and annual statistics for significant wave height, wave period
at the spectral peak, and mean current from the wave gage offshore of Perdido Key.
Figure 4.4 illustrates the number of successful data bursts collected for each month.
Data Records: 1/18/90 -- 6/30/92
Figure 4.4: Number of valid pressure and current data bursts, sorted by month, for
the period 1/18/90-6/30/92.
Table 4.1: Monthly and annual statistics for significant wave height, H,.
Maximum Date of Mean Standard Number of
Month (m) maximum (m) Deviation (m) Records
January 2.34 1/14/92 0.46 0.34 267
February 2.61 2/3/91 0.59 0.36 225
March 1.76 3/29/91 0.51 0.35 285
April 1.46 4/9/90 0.48 0.32 231
May 1.45 5/9/90 0.38 0.26 239
June 0.99 6/30/92 0.27 0.21 231
July 1.00 7/5/90 0.33 0.18 123
August 0.58 8/23/90 0.27 0.09 96
September 1.74 9/30/91 0.40 0.30 201
October 1.58 10/30/91 0.68 0.33 181
November 1.51 11/1/91 0.44 0.36 94
December 2.20 12/20/91 0.48 0.36 216
All 2.61 2/3/91 0.45 0.33 2389
Table 4.2: Monthly and annual statistics for wave period corresponding to peak of
energy spectrum, Tp.
Maximum Date of Mean Standard Number of
Month (sec) maximum (sec) Deviation (sec) Records
January 9.8 1/23/90 5.8 1.6 267
February 10.6 2/6/92 6.0 1.5 225
March 9.8 3/19/90 6.4 1.7 285
April 9.2 4/1/91 5.8 1.5 231
May 9.2 5/8/90 5.7 1.8 239
June 9.8 6/24/90 5.6 1.9 231
July 9.8 7/30/90 5.8 1.8 123
August 11.6 8/8/90 6.4 2.3 96
September 9.8 9/14/91 4.7 1.5 201
October 8.0 10/31/91 4.8 1.4 181
November 9.8 11/4/91 5.1 1.9 94
December 12.8 12/8/90 6.0 1.9 216
All 12.8 12/8/90 5.7 1.8 2389
Table 4.3: Monthly and annual statistics for magnitude of mean current, Uc.
Maximum Date of Mean Standard Number of
Month (m/s) maximum (m/s) Deviation (m/s) Records
January 0.33 1/27/90 0.09 0.04 198
February 0.69 2/15/90 0.09 0.06 252
March 0.93 3/8/90 0.08 0.08 283
April 0.60 4/8/90 0.08 0.06 194
May 0.27 5/13/90 0.09 0.05 113
June 0.16 6/4/90 0.06 0.03 120
July 0.16 7/4/90 0.05 0.03 124
August 0.40 8/31/90 0.27* 0.06 96
September 0.40 9/22/90 0.29* 0.04 120
October 0.46 10/18/90 0.35* 0.05 81
December 0.22 12/27/90 0.08 0.04 101
All 0.93 3/8/90 0.12 0.10 1682
'All data for 8/90-11/90 from one deployment; current meter calibration
Seasonal trends in the data are illustrated in Figures 4.5, 4.6, and 4.7. Both mean
and maximum monthly significant wave heights are quite small during the summer;
the same is also true of the mean currents. Wave period exhibits little variation
Wave Heights: 1/18/90 -- 6/30/92
2 .5 .............. .................. .................................................
S 1 .5 ..................... ........... .................... ......
1 .0. .. ..... ........... ........ ......
0 .5 ............. .
J FMAMJ J ASOND
Figure 4.5: Maximum and mean monthly significant wave heights, (Hs)max and H,.
Wave Periods at Spectral Peak
1/18/90 -- 6/30/92
Figure 4.6: Maximum and mean monthly wave periods at spectral peak, (Tp)max and
Mean Currents: 1/18/90 -- 6/30/92
Figure 4.7: Maximum and mean monthly mean currents, (Uc)maz and Uc. Calibration
suspect for August, September, and October data.
Histogram of Significant Wave Heights
... : ................................ .................. ......................................................................
.......... --------------- ------- ....-------........................... ............
2 8- i I: II.. I I .il 8 ."
0.1 0.5 0.9 1.3 1.7 21 .5
0.3 0.7 1.1 1.5 1.9 2.3 2.7
Figure 4.8: Histogram of significant wave heights measured in the field.
Wave Periods at Spectral Peak
3 4.2 5.4 6.6 7.8 9 10.2 11.4 12.6
3.6 4.8 6 7.2 8.4 9.6 10.8 12
Figure 4.9: Histogram of wave periods corresponding to spectral peak measured in
Histogram of Mean Currents
S .................... .. ....................... ......................................................................
Figure 4.10: Histogram of mean currents measured in the field.
0.175 0.325 0.475 0.625 0.775 0.925
0.25 0.4 0.55 0.7 0.85 1
Mean Current (m/sec)
Figures 4.8, 4.9, and 4.10 present histograms of the significant wave heights, wave
periods at the spectral peak, and mean currents, respectively. Previous researchers
have addressed the long-term distribution of significant wave heights (Goda ,
Houmb ), either explicitly or implicitly corresponding to deep-water conditions.
Shallow-water waves are less linear and may not follow the same distribution func-
tion. The Shore Protection Manual (U.S. Army Corps of Engineers ) presents
a modified exponential distribution function for significant wave height, H,:
F(H, > H,) = exp ~61 (4.3)
where H 2 2.63H,,,, so that F(H, > H,,,,) = 1. H, and H8m,, are the mean and
minimum significant wave heights, respectively. Equation (4.3) has been compared
to nearshore wave measurements made in water depths of three to thirty meters and
was concluded to give a reliable wave height distribution function for as little as one
year of data containing six observations per day (Thompson and Harris ). The
results from this study are plotted and compared to Equation (4.3) in Figure 4.11.
Agreement is good except for the largest wave heights, where the effects of the limited
record length become apparent.
Prediction of the extreme wave climate at the site is not as important to the
purposes of this study as is knowledge of the conditions that have actually been
experienced. Estimation of extreme conditions will be important when attempting to
predict future changes.
Probability of Exceedance for Hs
.. ............. .
u u !:;::;^:...!!:;;;::.. ;.;.:;::;: .;::::: .:; .:;;;;:;?;::::::;;::;;:;.;::::::.;::;;::;;:
Figure 4.11: Measured and predicted percent exceedance curves for significant wave
height. Measurements from Perdido Key, 1/90-6/92; theory due to Thompson and
. .............................................. ....... ....................... ........ ..............................................
,........................ >,................ ................ ............................ ........ ....... ......
,................................... ............ ... .......... .............................. ........................ :......
,-- ........... .....................
The histogram of wave periods at the spectral peak is clearly influenced by the
chosen cutoff frequency. The cutoff frequency of 0.33 Hz (Tp = 3.03 sec) corresponds
to the wave period having the highest frequency of occurrence. This is due to the
large number of records where the wave energy is very small. As explained earlier,
very small waves will generally result in the computation of a peak frequency that
matches the cutoff frequency. The measured joint distribution of significant wave
height and peak period shown in Figure 4.12 supports this claim.
2.0 4.0 6.0
0.0 2.0 4.0 6.0 8.0 10.0 12.0 14.0
Modal Period (sec)
Figure 4.12: Measured joint distribution of significant wave height and period corre-
sponding to spectral peak. Based on 2389 observations for 1/18/90-6/30/92.
9 1.0 2
5 25 10 2
2 1 15 47 31 8 1
1 15 70 69 51 2 1 1
49 75 117112 72 6 2
210 157 144 132 105 23 28 1 1
314 53 56 80 145 62 55 3 1 1
IA I I I I
The distribution of the mean currents appears somewhat anomalous, but the
secondary maximum in the vicinity of 0.3 m/s is a result of the incorrectly calibrated
current meter that was in place during the summer of 1990. If not for this problem,
the shape of the distribution would resemble that for the wave heights.
Figures 4.13 and 4.14 illustrate the directional distributions of the waves at the
spectral peak and the mean currents. The results agree with published references
(see Work et al. [1991c]) and field observations in indicating that the waves and mean
currents are dominantly directed toward the compass quadrant lying between West
and North. This suggests that the net longshore transport of sediment at the site
will be directed westward. A subsequent section will be devoted to proof of this
Wave Direction Histogram: 1/90-6/92
-15 -10 -5 0 5 10 15
Figure 4.13: Directional distribution of waves at the spectral peak, January,
1990-June, 1992. Top of plot corresponds to North. Each bin is ten degrees in
Some additional sources of wave data for the Perdido Key/Pensacola area are
available. The quality and utility of the data vary widely, from offshore visual obser-
vations to nearshore, in situ pressure measurements.
Hindcast wave heights, directions, and periods for several offshore stations are
available through the Wave Information Study (WIS) sponsored by the Coastal En-
gineering Research Center of the U.S. Army Corps of Engineers (Hubertz and Brooks
). Wave parameters were reported for each three-hour interval during the years
1956-1975. A separate study addresses tropical storm and hurricane conditions (Abel
et al. ).
Current Direction Histogram
-15 -10 -5 0 5 10 15
Figure 4.14: Directional distribution of mean currents, January, 1990-June, 1992.
Top of plot corresponds to North. Each bin is ten degrees in width.
Table 4.4: Summary of available wave data for Perdido Key area. Data collection for
this study is ongoing.
Data Source Location Technique Period of H,, m Period, sec
Record Max. Avg. Max Avg.
USNWSC Area 26 Visual 1949-1971 8-10 1.1 >13
Balsillie Ft. Pickens Visual 9/69-8/70 0.66 5.78
(1975) State Park (Hb)
Hubertz/Brooks Station 29 Hindcast 1956-1975 3.7 1.0 9.6-10.5 5.2
This study Perdido Key In situ 1/90-6/92 2.6 0.45 12.8 5.7
Visual observations from ships are available through a program coordinated by the
U.S. Naval Weather Service Command (USNWC ). The Summary of Synoptic
Meteorological Observations (SSMO) data provide visual estimates of wave height,
direction, and period representative of regional conditions. Nearshore, visual obser-
vations were reported by Balsillie , for several sites on Santa Rosa Island, east
of Perdido Key. Wave heights, directions, periods, and the speeds and directions of
longshore currents were recorded 20 to 30 times per month for a total of eight months
in 1969 and 1970.
The only known wave data for the Perdido Key area that were measured in situ,
prior to this study, were collected by Psuty . Nearshore pressure and velocity
records were collected at several nearshore sites along Perdido Key for a total of
five days, making the overall record length too short for application to this study.
Statistics of the wave data sources are summarized in Table 4.4.
4.2 Sediment Transport Processes: Observations Based on Survey Data
It is generally convenient to divide sediment transport processes, somewhat arbi-
trarily, into longshore and cross-shore components. The relative importance of both
modes of transport will often vary both temporally and spatially when some form
of shoreline perturbation is produced. This will be shown to be the case for the
Perdido Key beach nourishment project, where both longshore and cross-shore sed-
iment transport gradients are responsible for the changes observed to date, with the
cross-shore processes exerting the most influence for the early stages of the evolution.
Although it was stated earlier that the study area is essentially undeveloped,
this does not imply that it has not been affected by human activities. Dredging of
Pensacola Pass has taken place for over 100 years (Dean ); the resulting channel
serves as a partial barrier to longshore sediment transport, serving more as a "sink"
than a "bridge" for the sediment. The reduced sediment supply to the eastern end
of Perdido Key has resulted in a positive longshore gradient of longshore sediment
transport and a corresponding erosional trend. One goal of the beach nourishment
project was to offset the erosion that is the by-product of the deepened navigation
channel. A smaller beach nourishment project (1.86 million m3) was undertaken
in 1985 (Psuty ) for the same reason. The history of the site is discussed in
more detail in Work et al. [1991c]. Some results from the earlier (1985) nourishment
project will be discussed, but the primary focus here will be on changes occurring
since completion of the latest beach nourishment project.
The concept of an "equilibrium" configuration for a beach is often useful, whether
or not it is ever actually reached. Cross-shore and longshore equilibria may be con-
sidered separately, in keeping with the arbitrary division of the sediment transport
into orthogonal components. Following this convention and the coordinate system of
Figure 4.15, the equation expressing continuity of sediment is given as follows:
Oh q Oqy
a = V + -- (4.4)
at 5x ay
where q, = q,(x,y,t) is the volumetric longshore sediment transport rate, q =
qy(x,y,t) the cross-shore component, and h(x,y,t) the depth. Note that positive
sediment transport gradients imply local erosion. At equilibrium, gradients in both
directions vanish. Conversely, large perturbations away from the equilibrium con-
edition, in either direction, would be expected to induce large sediment transport
.......................................q + Ay
qx + T Ax
Figure 4.15: Coordinate system for sediment transport problem.
If the same beach profile is surveyed at two different points in time, then a value
for the left-hand side of Equation (4.4), averaged in time over the survey interval,
may be determined for any point along the profile. If Equation (4.4) is integrated
between two points at the extreme onshore and offshore limits of the active profile,
such that there are no profile changes at either of the integration limits, then all
changes in the cross-sectional area of the profile may be assumed due to longshore
gradients of longshore sediment transport:
aA v 9h 9q 9Q,
A= y. = a-dy = (4.5)
at o at d- X ax
Here A = A(x, t) denotes the cross-sectional area of the profile and Qx(x, t) is the
longshore sediment transport rate, integrated across the active profile. The distance
to the point where the depth reaches the "depth of closure", or limiting depth for
sediment motion, is denoted by y.. Any temporal variation in A(x, t) is thus indicative
of a longshore gradient in longshore sediment transport, a9Q/ax.
If an assumption is made regarding the cross-shore dependency of the local long-
shore gradient of longshore sediment transport, Oq,(x, y)/9x, the sediment continuity
equation can be integrated with indefinite limits to determine the local cross-shore
sediment transport rate:
/,) 8 ah [ Sq
qy(y) = -dy -dy (4.6)
where a boundary condition of q,(0) = 0 has been invoked. Very little information
exists regarding the y-dependence of the longshore gradient of longshore sediment
transport, aqx(x,y)/Ox. The most convenient assumption is that this term does
not contribute to the mass balance, i.e. 9q,(x,y)/zx=0 for all y. If the cross-shore
mean value of aq,/9x is non-zero, then OQx/ax will also be non-zero, and a net
increase or decrease in cross-sectional area will be evident. Additional factors could
lead to apparent changes in the cross-sectional area of the profiles, such as sediment
compaction, survey error, and wind-blown sediment transport. These contributions
will be discussed in a subsequent section.
Figures 4.16 and 4.17 present examples of the surveyed beach profiles at Perdido
Key before and after the nourishment project. Most of the pre-nourishment profiles
had a similar shape, with a pronounced breakpoint bar and very mild slopes offshore
of the -5 m contour. The profile immediately post-nourishment is drastically differ-
ent and by the previous argument, assuming that the pre-nourishment condition is
more indicative of equilibrium cross-shore conditions, strong cross-shore transport
gradients would be expected. Profile data collected at later times indicate that there
has been some slumping and a breakpoint bar is forming; the profile shape is evolving
toward the pre-nourishment condition.
2 ...... ...... ....... .......... .... 1 1/8 9
^ ** * ** ^ ^ -*- - --* ---- ---* -* - -- -
S -2 -----................................. 5/9 1
Q 3 -
Figure 4.16: Surveyed beach profiles at R-58, November, 1989, through May, 1991.
50 100 150 200 250 300 350 400 450
Distance from Monument (m)
Examination of the planform changes of the new beach provides an indication of
the rates of cross-shore and longshore equilibration. Figure 4.18 illustrates the spatial
extent of the longshore perturbation that the beach nourishment project represents;
the cross-shore perturbation has already been illustrated. It can be seen that the
most dominant early change in the planform was a rapid retreat of the waterline
within the nourished area.
Evolution of Planform
1 6 0 .............................................. .............................................. ........ ............. .................................... ....................... 9 /W
g 1 2 0 .......................................... ................. ..... ... .... .
S100- ----------- 5/91
......... I ........................... ... .
"fi "" 1 14 i10/91
S 20- ..... 6/92
2 ..... *...... ............................................................................................. ........... 6 9 2
45 50 55
Figure 4.18: Planform changes since beach nourishment.
Figure 4.17: Surveyed beach profiles at R-58, September, 1991, through June, 1992.
50 100 150 200 250 300 350 400 450 500
Distance from Monument (m)
Figure 4.19 presents a similar result for the movement of the -4 m contour. Al-
though less pronounced, the -4 m contour has moved seaward within the project
domain. This is simply another way of looking at the "unsteepening" illustrated in
Figures 4.16 and 4.17: there is a net movement of sand offshore, causing erosion of
the waterline and accretion of the -4 m contour.
Movement of -4 m Contour
2 0 .............................................. ..... .................. ..... ..
S 2 0. .. .. .. ................................... .............................. -..........
Figure 4.19: Movement of the -4 m contour since beach nourishment.