• TABLE OF CONTENTS
HIDE
 Front Cover
 Title Page
 Acknowledgement
 Table of Contents
 List of Figures
 List of Tables
 List of symbols
 Abstract
 Introduction
 Data collection
 Image analysis
 Results and discussion
 Conclusions and recommendation...
 Appendix A: Listing of equipment...
 Appendix B: Program listings
 References
 Biographical sketch






Group Title: UFL/COEL (University of Florida. Coastal and Oceanographic Engineering Laboratory) ; 93/006
Title: Video monitoring techniques in the coastal environment
CITATION THUMBNAILS PAGE IMAGE ZOOMABLE
Full Citation
STANDARD VIEW MARC VIEW
Permanent Link: http://ufdc.ufl.edu/UF00079959/00001
 Material Information
Title: Video monitoring techniques in the coastal environment
Physical Description: x, 94 leaves : ill. ; 29 cm.
Language: English
Creator: Mason, Timothy P., 1968-
Publication Date: 1993
 Subjects
Subject: Coastal and Oceanographic Engineering thesis M.E   ( lcsh )
Dissertations, Academic -- Coastal and Oceanographic Engineering -- UF   ( lcsh )
Genre: bibliography   ( marcgt )
non-fiction   ( marcgt )
 Notes
Thesis: Thesis (M.E.)--University of Florida, 1993.
Bibliography: Includes bibliographical references (leaves 92-93).
Statement of Responsibility: by Timothy P. Mason.
General Note: Typescript.
General Note: Vita.
Funding: This publication is being made available as part of the report series written by the faculty, staff, and students of the Coastal and Oceanographic Program of the Department of Civil and Coastal Engineering.
 Record Information
Bibliographic ID: UF00079959
Volume ID: VID00001
Source Institution: University of Florida
Holding Location: University of Florida
Rights Management: All rights reserved by the source institution and holding location.
Resource Identifier: aleph - 001950915
oclc - 31200902
notis - AKC7457

Table of Contents
    Front Cover
        Front Cover
    Title Page
        Page i
    Acknowledgement
        Page ii
    Table of Contents
        Page iii
        Page iv
    List of Figures
        Page v
        Page vi
    List of Tables
        Page vii
    List of symbols
        Page viii
    Abstract
        Page ix
        Page x
    Introduction
        Page 1
        Page 2
        Page 3
        Page 4
        Page 5
        Page 6
        Page 7
        Page 8
    Data collection
        Page 9
        Page 10
        Page 11
        Page 12
        Page 13
        Page 14
        Page 15
    Image analysis
        Page 16
        Page 17
        Page 18
        Page 19
        Page 20
        Page 21
        Page 22
        Page 23
        Page 24
        Page 25
        Page 26
        Page 27
        Page 28
        Page 29
        Page 30
        Page 31
        Page 32
        Page 33
    Results and discussion
        Page 34
        Page 35
        Page 36
        Page 37
        Page 38
        Page 39
        Page 40
        Page 41
        Page 42
        Page 43
        Page 44
        Page 45
        Page 46
        Page 47
        Page 48
        Page 49
        Page 50
        Page 51
        Page 52
        Page 53
        Page 54
        Page 55
        Page 56
        Page 57
        Page 58
        Page 59
        Page 60
        Page 61
        Page 62
        Page 63
        Page 64
        Page 65
        Page 66
        Page 67
        Page 68
        Page 69
        Page 70
        Page 71
        Page 72
        Page 73
        Page 74
        Page 75
        Page 76
        Page 77
        Page 78
        Page 79
        Page 80
    Conclusions and recommendations
        Page 81
        Page 82
        Page 83
        Page 84
    Appendix A: Listing of equipment and vendors
        Page 85
        Page 86
    Appendix B: Program listings
        Page 87
        Page 88
        Page 89
        Page 90
        Page 91
    References
        Page 92
        Page 93
    Biographical sketch
        Page 94
Full Text



UFL/COEL-93/006


VIDEO MONITORING TECHNIQUES IN THE
COASTAL ENVIRONMENT








by



Timothy P. Mason






Thesis


1993














VIDEO MONITORING TECHNIQUES IN THE
COASTAL ENVIRONMENT













By

TIMOTHY P. MASON


A THESIS PRESENTED TO THE GRADUATE SCHOOL
OF THE UNIVERSITY OF FLORIDA IN PARTIAL FULFILLMENT
OF THE REQUIREMENTS FOR THE DEGREE OF
MASTER OF ENGINEERING

UNIVERSITY OF FLORIDA


1993














ACKNOWLEDGMENTS


I wish to express a great deal of thanks to my advisor and supervisory committee

chairman, Dr. Daniel M. Hanes, for his support and guidance during my stay at the

University of Florida. I would also like to thank the members of my supervisory

committee, Dr. Robert G. Dean and Dr. Robert J. Thieke. Also many thanks go to Dr.

Rusty Erdman, without whom this endeavor into the video realm would not have been

possible.

I would also like to thank the field crew at the Coastal Engineering Laboratory,

Mark Sutherland and Vic Adams, and fellow students)Phil Dompe and Chris Jettefor

assistance during field deployments and surveys. I am greatly indebted to the coastal

secretarial staff, Becky, Sandra, Cynthia, and Lucy, for their crucial clerical support.

Thanks also go to Helen for the goodies that keep the coastal working hard. I need to

thank my parents Faye and Richard Badman for providing me the mind, body, and

support to exceed in all my endeavors. I am also indebted to Paula and Bob Kelly for

much-needed summer employment, financial assistance, and many great times. Mr.

Robert Janssen, wherever he may be, is responsible making my desire to ride waves a

reality and an obsession.

Miscellaneous kudos go to my friends who kept me entertained during this work:

Quazi, Paul (the man), Dagwood, Joe, and Eric. Thanks for entertainment and sanity

checks go to Dominical and Tamarindo, Reggie and Mike, Rob (for the awesome boards

and adventures), Market Street Pub and St. George's Tavern, Squeeze, the Mons and the

Diamond Club, the greatest drummers of the world, rock and roll, Beavis and Butt-head

(that would be cool), Dos Equis, Imperial, Fosters Lager, and (most importantly) waves.















TABLE OF CONTENTS

Page
ACKNOWLEDGMENTS ....................................................... ......................... ii

LIST OF FIGURES ............................................................................................ v

LIST OF TABLES .................................................................................................... vii

LIST OF SYMBOLS ................................................................................................ viii

ABSTRACT ........................................................................................................ ix

CHAPTERS

1 INTRODUCTION ........................................................ .............................. 1

1.1 Objectives, Scope, Tasks .................................................. ..................... 1
1.2 Upcoming Chapters ........................................ .............. .......................... 2
1.3 Literature Review..................................................... ............................... 3

2 DATA COLLECTION ........................................ .............. ......................... 9

2.1 The Video Monitoring System........................................... ..................... 9
2.1.1 Introduction.............................................. ....................................... 9
2.1.2 Description.......................................................................................... 9
2.1.3 Capabilities ........................................................ ........................... 10
2.2 Field Deployments......................................................................................... 11
2.2.1 Hollywood Beach................................................. ......................... 11
2.2.2 M iami Beach...................................................................................... 13
2.2.3 Longboat Key..................................................... ........................... 14

3 IMAGE ANALYSIS...................................................................................... 16

3.1 Digital Image Concepts............................................... .. .................... 16
3.2 Photogrammetry Basics ..................................... .............................. ...... 18
3.3 Image Rectification................................................. ................................ 21
3.3.1 Introduction.................................................. ................................. 21
3.3.2 Program Background .................................................................... 23
3.3.3 Camera View and Setup........................................... ..................... 23
3.3.4 Surveys and Measurements................................................................ 24









3.3.5 Rectification Program ...................................................................... 25
3.3.6 Program Calibration......... ................................. ......................... 27
3.4 Other Methods .......................................................................................... 32
3.4.1 Global Lab Image ............................................................................ 32
3.4.2 Overlay Grids................................................................................... 33

4 RESULTS AND DISCUSSION..................................................................... 34

4.1 Introduction....................... ........................................................................ 34
4.2 Site Documentation...................................... ........................................... 34
4.2.1 Large Image Data Sets..................................................................... 34
4.2.2 Event Documentation....................................................................... 36
4.2.2.1 Local conditions.................................................................. 36
4.2.2.2 Beach traffic...................................................................... 38
4.2.2.3 Beach activities ................................................................. 38
4.2.2.4 Shoreline change ............................................................... 40
4.2.2.5 Predominant coastal features ................................................. 41
4.2.2.6 Strange events ................................................................... 41
4.3 Nearshore Features...................................................................................... 47
4.3.1 Shoreline Position.............................................................................. 47
4.3.2 Bars and Surf Zone Width............................................................... 53
4.4 Turbidity Phenomena................................................................................ 59
4.4.1 Natural Turbidity Structures ............................................................ 59
4.4.1.1 Lobes................................................................................. 60
4.4.1.2 Rip currents....................................................................... 70
4.4.2 Man-induced Turbidity Structures................................................... 74
4.4.2.1 Discharge snakes ............................................................... 74
4.4.2.2 Dredge overflow ............................................................... 79

5 CONCLUSIONS AND RECOMMENDATIONS .............................................. 81

5.1 General Conclusions.................................................................................. 81
5.2 Future Developments................................................................................ 83


APPENDIX A: LISTING OF EQUIPMENT AND VENDORS .......................... 85

APPENDIX B: PROGRAM LISTINGS ............................................................ 87

REFERENCES ..................................................................................................... 92

BIOGRAPHICAL SKETCH ................................................................................ 94


I















LIST OF FIGURES

Figure


2.1 Typical image from Hollywood Beach data set........................................ 12

2.2 Camera location on Hollywood Beach, FL.............................. ........... 13

2.3 Typical image from Miami Beach data set................................................. 15

2.4 Camera location on Miami Beach, FL.................................... .......... .. 15

3.1 Exposure station geometry..................................................................... 19

3.2 Auxiliary image coordinate system ........................................ .......... .. 22

3.3 Oblique image, Hollywood Beach 5-24-91 ............................................. 26

3.4 Rectified image, Hollywood Beach 5-24-91........................................ 27

3.5 Test setup schem atic .............................................................................. 28

3.6 Oblique test im age ................................................................................. 29

3.7 Rectified test im age................................................................................ 29

3.8 Florida Field oblique shot...................................................................... 31

3.9 Florida Field rectified ............................................................................ 31

3.10 O verlay grid ........................................................................................... 33

4.1 Hollywood Beach under typical conditions........................................... 36

4.2 Hollywood Beach under storm conditions............................. ........... .. 37

4.3 In-situ data from Hollywood Beach, FL.............................................. 38

4.4 Hollywood Beach renourishment, June 16, 1991................................... 39

4.5 Zoom of Hollywood Beach .................................................................... 40

4.6 Hollywood beach cusps, April 26, 1991.................................. ........... 42

4.7 Hollywood beach cusps, May 23, 1991 .................................. ............ 42










4.8 Cargo ship grounded on Miami Beach ................................................. 43

4.9 Cargo ship rem oval................................................................................ 43

4.10 Surf break after ship rem oval...................................................................... 44

4.11 Bungee jumper in mid-flight.................................................................. 44

4.12 Halloween storm waves, October 31, 1991 ........................................... 45

4.13 Halloween storm swash and flooding .................................................... 46

4.14 Nearshore turbidity during Halloween storm......................................... 46

4.15 Hollywood Beach nourishment sequence................................................ 48

4.16 Hollywood Beach shoreline pre- and post-nourishment........................... 49

4.17 Hollywood Beach shoreline during nourishment ..................................... 49

4.18 Hollywood Beach construction profiles...................................................... 51

4.19 Shoreline location from construction surveys.......................................... 51

4.20 Typical time average from Miami Beach ............................................... 55

4.21 Miami Beach bar progression..................................................................... 56

4.22 Comparison of image intensity and beach profile .................................. 58

4.23 Turbidity lobes propagation, April 26, 1991 ........................................... 61

4.24 Plotted turbidity lobes progression, 4/26/91.......................................... 62

4.25 Turbidity lobe progression, May 24, 1991........................................... 66

4.26 Plotted turbidity lobes progression, 5/24/91 ............................................... 67

4.27 Rip current at Miami Beach, 12:00 PM............................................... 72

4.28 Rip current at Miami Beach, 1:00 PM..................................... ............ 72

4.29 Time averaged image, Miami Beach...................................... ........... 73

4.30 Discharge snake progression, 6/12/91 ................................................... 75

4.31 Plotted dredge discharge snake progression, 6/12/91 ............................... 76

4.32 Dredge overflow turbidity plume....................................................... 80


I















LIST OF TABLES


Table Pg


4.1 Shoreline position comparison........................................................... 52

4.2 Data summary of turbidity lobes, April 26, 1991 ................................... 63

4.3 In-situ data summary, April 26, 1991 ...................................... .......... 64

4.4 Wind data from April 26, 1991............................................. ........... 64

4.5 Data summary of turbidity lobes, May 24, 1991 ................................... 68

4.6 In-situ data summary, May 24, 1991 ...................................... ........... 69

4.7 Wind data from May 24, 1991 .............................................. ........... 69

4.8 Data summary of discharge snake, June 12, 1991 .................................. 77

4.9 In-situ data summary, June 12, 1991 ..................................... ............ 78

4.10 Wind data from June 12, 1991 .............................................. ........... 78














LIST OF SYMBOLS


fc focal length

camera azimuth angle

H exposure station elevation

Hs significant wave height

NTU turbidity, in Nephelometric Turbidity Units

P ground location coordinate

p image location coordinate
s camera swing angle

- camera tilt angle

0 horizon tilt angle

peak wave direction

U cross shore current velocity

V longshore current velocity

Zp ground point elevation















Abstract of Thesis Presented to the Graduate School
of the University of Florida in Partial Fulfillment of the
Requirements for the Degree of Master of Engineering

VIDEO MONITORING TECHNIQUES IN THE
COASTAL ENVIRONMENT
By

Timothy P. Mason

December 1993

Chairman: Daniel M. Hanes
Major Department: Coastal and Oceanographic Engineering


The field of coastal engineering has primarily utilized traditional measurement

techniques such as boat surveys and beach profiles. Aerial photography has played a

major role in the mapping of coastlines and ocean currents. More recent technology has

provided the ability to make field measurements with precision electronic

instrumentation, but the problem of making large-scale measurements within economic

bounds remains. The video revolution and, in particular, desktop video has greatly

enhanced the ability of scientists and engineers to visualize events over greater periods of

time and larger scales than ever before.

The uses of video, benefits, shortcomings, and future expectations are presented

as applicable to the field of coastal engineering. This feasibility study incorporates some

applications which have previously been explored, as well as others which result from

two unique video data sets that were collected between 1991 and the present.

Using a rectification procedure and image processing system, video estimates of

general length scales and shoreline position have been found to be within 5% of survey

data. Accuracy depends on the camera and camera set up, type of image used, view


I /










geometry, and subsequent image manipulations. Nearshore sand bar dimensions and

location are observed using a time averaging technique. Beach fill progress during a

beach renourishment is monitored by measuring changes in shoreline location. Rip

currents and other turbidity structures, both natural and man-induced, are observed and

mapped over time. Although the forcing of such turbidity events is not fully explained,

the influence of waves and coastal advective currents is shown. The turbidity patterns

tend to form after periods of increased wave activity or during beach renourishment.

Cross shore and longshore velocities of observed turbidity plumes are similar to in-situ

currents measured during the Hollywood Beach renourishment of summer 1991.

Turbidity plumes mapped at Hollywood Beach may extend offshore over 3 times the

width of the surf zone and have typical wavelengths around 5 times the width of the surf

zone.

Video monitoring has proven to be a useful and cost-effective way to perform

traditional observations and also to document events that would not normally be possible

to record with standard instrumentation in the coastal environment. The nature of the

typical oblique views and digital imagery places some limitations on the accuracy of the

applications, but the large spatial scale and flexible temporal scale make video

monitoring an attractive option.


I














CHAPTER 1
INTRODUCTION




The old adage "a picture is worth a thousand words" is quite appropriate to

summarize the concept of video monitoring. The field of coastal engineering, still in

relative infancy, has traditionally utilized visual measurements coupled with standard

measurement techniques. These techniques include surveys with differential levelling,

hydrographic surveys with transducers, and wave staffs, to name a few. Aerial

photography and acoustic instrumentation have also contributed to a more comprehensive

understanding of coastal processes.

The use of video monitoring, particularly in the coastal environment, has resulted

from several circumstances. The video revolution and recent advances in desktop video

have put video equipment within the reach of consumers. People are familiar with video,

which produces a quality image that can convey information much faster than the written

word. A video image (for example, of a segment of the coast) transmits much

information about the given scene, over a large spatial scale. Weather, sea conditions,

beach traffic, ship traffic, and coastal features are some of the obvious pieces of

information which can be visually assessed from video images. With time lapse video,

the long term events and processes can be studied.


1.1 Objectives. Scope. Tasks

The primary purpose of this paper is to introduce the concept of video monitoring

to the coastal scientist and engineer. Few investigators have used video for coastal










research applications over the last decade, so a literature review is provided in Section

1.3. The developments in video monitoring techniques are presented. The basic

principles of images and their analysis are discussed. The video monitoring system

hardware and software, installation, and uses are presented.

Techniques of image interpretation for coastal applications are discussed, but the

highest emphasis is placed on the feasibility of using video for coastal observations.

From our preliminary (and prototype) systems and data collections, the "dos and don't"

of video monitoring have become clear. This report is an introduction to video

monitoring, to present the benefits and drawbacks of such systems, and to inspire interest

in others to pursue the suggested techniques in greater detail.



1.2 Upcoming Chapters
Section 1.3 contains a short literature review of prominent work that has been

done with video monitoring (and related subjects) over the last two decades. Important

entries are merely summarized. Interested readers are strongly suggested to consult the

references provided for details.

Chapter 2 contains the data collection information from the past two years of

research. The video monitoring system and the field deployments are described. Chapter

3 presents the basics of image analysis. An introduction to digital image concepts,

photogrammetry, image rectification and other methods are discussed. Chapter 4

contains the results and discussion of the gathered and analyzed data. Some areas

considered have been previously covered, while others are unique to this research.

Chapter 5 contains conclusions and a summary of this study, as well as recommendations

for current and future video monitoring endeavors. Equipment, vendors, and computer

program listings are presented in the appendices.










1.3 Literature Review

As far back as the 1940s, coastal engineers were beginning to use photography for

qualitative assessment of coastal features. Aerial photographs and satellite images have

been integral in the mapping of the world coastlines and other large scale features of the

oceans. Presented herein are several of the papers from the past two decades which

introduced video monitoring and brought it to the level of today. It is noted that the field

is limited; there is not much reference literature available.

Prior to the video revolution, which can be dated back to the early 1980s, the

emphasis was placed on the use of terrestrial photographs for interpretation.

Photogrammetric techniques for interpretation of oblique photographs have been well

known for many years. Loomer and Wolf (1974) used land-based cameras to map water

surface velocities on Lake Michigan. A camera was set up on a bluff at 30 m above the

lake water level. Oblique photographs were taken of drogues which were deployed on

the lake to map outfall currents. Using traditional photogrammetric techniques, Loomer

and Wolf determined that errors in measurements increase with distance from the camera.

Errors in velocity estimates were up to 10% for drogues that were 213 m from the

camera.

In 1976, Maresca and Seibel used a similar technique to measure breaking waves

and currents in the nearshore region. Two techniques were investigated: single oblique

photos and stereo pairs (the concept of stereo image analysis extends beyond the scope of

this paper and will not be included). Several important points were made. Maresca and

Seibel pointed out that if the tilt of the apparent horizon is less than 1, then no corrective

measures need to be taken to account for a slightly tilted horizon (which would introduce

roll effects in the images). Using a 35 mm camera on a voice timer, snapshots'were taken

and analyzed to determine breaker location, nearshore wavelength, runup, and ice ridge

location. With the camera mounted on an 8 m bluff, the maximum useful range was

found to be 250 m. Accuracies of measured distances were better than 1% horizontally










and 10% vertically. Breaker height estimates were found to have accuracies around 10%,

but the difficulty was in determining the true location of the trough of the waves from the

photographs. Some current mapping was also investigated with the use of dye packs

placed in the surf zone. For single oblique shots, Maresca and Seibel note that no

reference stakes are required for image calibration if the horizon, camera elevation, and

focal length are known.

The most prolific group on the subject of video monitoring for coastal

applications is under Dr. Rob Holman at Oregon State University. Holman and Guza

(1984) used time-lapse photography to measure run-up. Photo measurements were

"intercalibrated" with standard dual-resistance wire sensors. At this early stage, the

photos (slides) were projected on a screen and digitized manually. This subjectivity and

tedious nature of the manual digitization gave way to further improvements, but the

compared results were quite accurate (a small difference in set-up measurements, but up

to 83% difference in swash variance). The main advantage of the photographic technique

was the ability to make measurements in storm conditions, when coastal engineering

interests peak.

Lippmann and Holman (1987) developed a technique for modeling and measuring

3D morphology of submerged features during the DUCK'85 field experiment. 35 mm

cameras were set up on a 14 m tower, but 10 minute time exposures were taken instead of

single snapshots. The time exposure essentially averages over the modulations of the

incoming, random waves. Again, photographs were analyzed with the subjective

digitization process. A small data set was used to test the model which was based on the

preferential breaking of incoming waves over shallower areas and sand bars. This

technique, like Maresca and Seibel, was operational during all types of conditions (except

darkness) and covered a much larger spatial area than practical with traditional survey

techniques. Lippmann and Holman's initial tests showed that errors in measurements

made with the photographic technique are within 2% of the distance from the camera and


I










up to 15% on bars with an incoming tide. No initial effect on the time exposure

technique due to wave size was initially seen.

Lippmann and Holman (1989) presented results from the SUPERDUCK

experiment (1986). They improved upon their earlier concept in several ways. The

previous model was updated and based on a random wave model, with the assumption

that light intensity (and hence, image intensity) is proportional to the energy dissipated by

breaking waves. The dissipation model was found to work best for waves that just break

over the bar and at lower tides.

Of significant importance was the medium transition from standard photography

to video. A black and white video camera was mounted on a 40 m tower and recorded 20

minute records each hour. A computer-based image processing system was introduced to

simulate a time exposure by time averaging successive frames. The image processing

system was also used to generate intensity plots for selected cross shore and long shore

profiles. The intensity plots were compared with substantial survey data which was

collected during SUPERDUCK. Also presented were the photogrammetric equations of

geometrical transformation for measurement from oblique images and rectification.

The results of analysis showed that for a camera elevation of 40 m and tilt of 85,

the worst-case error in measurements was less than 5% of the distance to the camera.

When the wave field became saturated (during large wave conditions), it was observed

that the location of the bar crest was consistently displaced seaward with up to a 35%

error in determination of the crest location. Accurate bar crest location was also affected

by high onshore winds and persistent foam which remains shoreward of the bar area after

waves break.

To remove the "noisy" signal of the foam, Lippmann and Holman employed a

differencing technique. This technique is basically the reverse process of time averaging.

Instead of adding frames, successive frames (0.5 to 1.0 second apart) were subtracted. In

the resulting image, areas of small change in contrast exhibit no difference. Breaking

i










areas exhibit large intensity changes over time, therefore large difference signals. The

differencing method improved the accuracy of the bar crest location determination.

General conclusions of Lippmann and Holman (1989) are as follows. First, the

time averaging technique presents a generally good method of determining submerged

features and their cross shore and longshore scales. This technique is also excellent for

determination of sea level location. Large wave breaking tends to skew the bar crest

location offshore, and residual foam tends to skew crest location onshore.

Aagaard and Jorgen (1989) used a video technique to measure run-up. This was

primarily an extension of Holman and Guza's work (1984).. Aagard and Jorgen made use

of a video camera which was mounted on a tripod and set to look alongshore. Stakes

were set at specific intervals for later profile identification. The camera range was

determined to be about 75 to 100 m for 15 m swash excursions and 0.5 inch diameter

stakes. Frames were digitized with a PCVision frame grabber, and the profiles were

scanned by the computer at selected intervals. An operator subjectively digitized the

shoreline. A 2048-point time series could be generated in about 75 minutes. The authors

found this method to be more time-consuming than Holman and Guza (1984) but more

accurate and easier than using photographs.

The group at Oregon State University continued work into the 1990s but began to

move into other applications of video monitoring. Holman et al. (1990) improved upon

the previous run-up experiments during SUPERDUCK'86. Three video cameras were

mounted 43.2 m above mean sea level. Video records of 1 hour and 55 minutes were

taken to coincide with in-situ current meter data runs.

An Imaging Technology 150 Series image processing system was used for

analysis. The appropriate geometry of the image was solved, cross shore profiles were

taken at 10 m intervals, and the location of the swash excursion located each second.

This use minimized the operator subjectivity of other experiments to date. Typical


I










horizontal resolutions were found to be 20 cm at close range (2 cm swash elevation) and

73 cm for the most distant transects (7.3 cm swash elevation).

Holman, Liplmann, O'Neill, and Hathaway (1991) investigated a video method

for measuring beach profiles. This technique relied heavily on the photogrammetric

transformations for the quantification of three dimensional quantities from a two

dimensional image. Where most other applications had been used to determine quantities

on a plane (the sea surface), this technique required an extra piece of information. A line

had to be drawn across the beach profile in order to successfully extract the three

dimensional data desired. A fire hose was used for initial tests, but suggestions were

made for use of a light beam or shadow which could easily be automated for

computerized measurements.

Results for the typical camera setup (angle of tilt = 750, elevation = 44.02 m

relative to MSL, field of view = 30) show a 5 cm accuracy at a range of 100 m. The

advantage was noted to be the ease of measurements on a large scale.

Lippmann and Holman (1991) used video to observe phase speed and angle of

breaking waves. Experiments were conducted during the DELILAH experiment in fall

1990. Up to 8 cameras were deployed, but the data presented made use of only one

particular camera view. Intensity records were sampled over 2 hours at 10 Hz, and later

resampled at 8 Hz for easier correlation with other in-situ instrumentation. In-situ

instrumentation included an array of pressure sensors and current meters.

Results of this experiment showed a good correlation between the in-situ wave

measurements (phase, celerity, and spectra) and the video intensity time series. This

result allows the video technique to be used for measurement of nearshore wave

parameters, without the normal problems of deployment and maintenance of standard in-

situ instrumentation.

A final technique was tested in 1991. Holland, Holman, and Sallenger (1991)

used video to measure barrier island overwash in Isles Demieres, Louisiana. Data were










taken along with in-situ instruments. The new data set included a major event courtesy of

Hurricane Gilbert, September 1988. Overwash velocities were mapped from the video

data, by calculating bore speeds from observations of gradients in the video intensity data

time series.

Many coastal measurement techniques are available with video monitoring. Some

areas have been studied in detail, primarily the basics of video photogrammetry and the

time exposure techniques, by Lippmann and Holman (1987 and 1989). Most of the

investigations have concluded with as many questions as solid results. The results are

based on relatively small data sets. Comprehensive, high quality video data sets and

enthusiastic investigators are required to propose and evaluate new analysis techniques.














CHAPTER 2
DATA COLLECTION


2.1 The Video Monitoring System

2.1.1 Introduction

The video monitoring system came into existence with the desire to use visual

media to make observations and measurements on the assembly line or in the field. Still

terrestrial and aerial photography have both been utilized for many years. For the

measurement of static subjects, these methods are fine. Problems arise, however, when a

dynamic event has to be documented. Video allows data to be taken continuously or in

bursts. The other factor influencing the development of the video monitoring system is

the financial advantage. Aerial photography can get quite expensive, with a conservative

minimum around $75 per hour of flying time plus camera equipment. If multiple photos

are required on a regular basis, the cost will easily exceed the cost of the video

monitoring system. A video monitoring system provides the opportunity to acquire

images, still or moving, at any desired sampling frequency. Maintenance is minimal after

initial deployment and mainly involves periodic data off loading.

2.1.2 Description

The video monitoring system has gone through a major metamorphosis during

this project. The video monitoring systems (hereafter referred to as VMS) which have

been used during this project have been designed by Erdman Video Systems of Miami

Beach, Florida. Erdman's VMS consists primarily of off-the-shelf consumer and

industrial components which are controlled with commercial and custom software. VMS

primary components include a Hi8 video camera with variable zoom, a pan and tilt

mechanism which consists of a digitally-controlled stepper motor, and a personal










computer (PC). Our first VMS was run by a Databank 386 PC with only 40 megabytes

(MB) of hard disk space. The current configuration uses a 200 MB hard drive, which can

store up to 7000 compressed images.

VMS software allows the user full control of the camera. Up to 34 different

scenes can be programmed at a time. The user sets the desired pan, tilt, and zoom for

each scene. Repeatability is better than 0.10 for all angles. Sampling frequency is also

set for each scene. All of the programming functions can be adjusted from a remote PC

connected to the VMS by a modem. Images are acquired and digitized in 24-bit color

Targa format. Acquired image data can also be transmitted to and viewed on a remote

PC via the modem connection. Images are compressed with JPEG compression to

maximize storage and minimize modem transmission time. All that is required for

viewing purposes is an appropriate SuperVGA display. Images are decompressed and

converted to standard 8-bit GIF format for display on PCs with limited display capability.

2.1.3 Capabilities

The VMS is versatile. The camera can be mounted on any wall or roof. The

housing is weatherproof for outdoor deployments and includes a cooling fan and sun

shroud. The only extra protection to be provided might include a lightning protection

device if the system is mounted on a rooftop or lightning prone area, like Florida. Views

can be set for any zoom, pan, and sampling scheme.

The first VMS (VM-1) did not include a weatherproof camera and had to be

deployed inside a building. Problems arose from the occasional dew formation on the

windows and typical afternoon glares off the glass windows. VM-1 recorded all video

data on Hi8 videotape. Several scenes were digitized with the frame grabber for

transmittal to the university for site evaluation at regular intervals. Consequently, data

sequences used for analysis had to be digitized from the source tapes.

SThe latest version of the VMS includes a surveillance-type camera housing which

is fully weatherproof. In addition to recording all data on tape, the images are digitized,


I










compressed, and stored on the computer hard drive. The only disadvantage exists if the

user desires continuous video or a very high sampling frequency. In this case, the Hi8

taping system would be modified accordingly. The latest VMS also includes an

improved pan and tilt mechanism for reliable repeatability of pan and tilt values.

Field applications of the VMS are not limited to subjects in the coastal

environment. The nature of the VMS lends itself to long term monitoring projects, at a

minimal cost. Other applications include construction site monitoring, breakwater

stability monitoring, marine and terrestrial traffic monitoring, surface pollution dispersal,

sea turtle nesting, and bird nesting behavior. The Department of Coastal and

Oceanographic Engineering is currently investigating the uses of video monitoring

underwater, for observing field instrumentation sites during deployments.



2.2 Field Deployments

2.2.1 Hollywood Beach

The video monitoring data from Hollywood Beach, Florida was collected as part

of a study on the influences of beach nourishment on nearshore turbidity. The study,

sponsored in part by Florida Sea Grant, took place between January 1990 and April 1992

in order to make measurements before, during, and after a beach nourishment during the

summer of 1991. In-situ measurements of wave climate and turbidity in the lower 1

meter of the water column were recorded during 30 minute bursts at 4 hour intervals.

The VMS was initially installed as a test for future projects. At the onset of the project,

there was no specific goal in mind for the video data, other than perhaps site

documentation. Here it is stressed that since no specific goal existed, the required

attention to the details of installation, camera geometry, and ground surveys was not

given. As the investigators made routine remote logons and observed the VMS data, it

was realized that this was an excellent source of information pertaining to events in the

coastal environment.


I









The Hollywood VMS was set up on the 21st floor of the Summit Condominium,

located at 1201 S. Ocean Drive, Hollywood Beach, Florida. As stated previously, the

camera was not weatherproof and was erected inside a private residence in the

condominium south tower, 55 m above mean sea level (MSL). The camera view was

pointed in a northeastward direction, over a 6.5 mile stretch of coastline including part of

the project area and Dania Beach. The wide camera view (8.5 mm zoom) provided for a

"useful" video window that spanned DNR R-Monuments 114 through 117 on Hollywood

Beach. A maximum angle directed offshore (relative to longshore, generally N 50 E on

Hollywood Beach) was about 750, limited by the north-facing sliding glass door. Data

was sampled at the following rates. Eight scenes were set at various intervals up the

beach (northeast) with appropriate zooms. Video records of 0.7 seconds each were

sampled at 30 minute or 1 hour intervals from April 1991 through November 1991.

Figures 2.1 and 2.2 show a typical view from the Hollywood data set and an aerial

snapshot with the VMS location indicated on it, respectively.


gure L.1 I typical image rrom nollywooa 1ei















HOLLYWOOD BEACH
.. A .. _


rida Dept. Natural Resources)


2.2.2 Miami Beach

The purpose of the Miami Beach VMS deployment was to obtain a video data set

of nearshore features and their changes over several months. This deployment included

several upgrades (hardware and software) to the VMS. Hardware upgrades included a

weatherproofed camera housing, a much improved pan-and-tilt mechanism, and a larger,

200 MByte hard drive on which to store images. Software upgrades were fourfold. The

remote access program was modified to allow the remote user to actually set and reset

any scene, or add scenes as desired. The image transfer options were expanded to include

transmittal of sequences of images or mosaics of images. The 8mm tape was abandoned.

Images were digitized in real time and compressed and saved on the computer hard drive.

This eliminated the extra step of re-digitizing desired images from tape at a later date.

Perhaps the most important modification was the implementation of a time averaging

scheme in order to eliminate high frequency signals and enhance quasi-static nearshore










features, like shoreline position and bars. This concept of time averaging is discussed in

more detail in Section 3.1, Digital Image Concepts.

The VMS was erected outside an apartment on the 16th floor of the south wing of

the Roney Plaza, located at 2301 Collins Avenue, Miami Beach, Florida. The camera

height was about 50 meters above MSL. The other camera view parameters were similar

to that of the Hollywood Beach deployment. Cameras are generally set for the best view

during any deployment, so parameters may vary accordingly. Instantaneous images were

digitized once per hour for each of 5 scenes. Time averaged images of each scene were

also sampled at hourly intervals. The time averages were sampled as follows, dictated by

the computer hardware and software speeds: 8 images were sampled over a 5 minute

period and digitally added to obtain the final image. The image data were collected over

a six month period, from December 1992 through May 1993. It is noted here that no in-

situ instrumentation was deployed during the video deployment. A nearby University of

Florida Coastal Data Network (CDN) instrument package was not in operation during the

experiment. The primary objective was to test the application of the VMS, not to study

any particularly detailed mechanics ofnearshore processes. Figures 2.3 and 2.4 present a

typical view from the Miami Beach data set, and an aerial snapshot with the camera

location indicated on it.

2.2.3 Longboat Key

In June 1993, a video monitoring system was erected on a condominium on

Longboat Key, Florida, as part of a beach nourishment monitoring program. The video

monitoring program was mandated by Florida Department of Environmental Regulation.

The study is designed to provide a comprehensive turbidity analysis before, during, and

after the beach nourishment which occurred during summer of 1993. This paper does not

cover any of the experimental details or future analysis of the Longboat Key results, but

the techniques discussed herein are being applied to that data set.











































figure 2.) typical image irom Miami


r figure L.4 camera locanon on Miami beacn, rL (courtesy riorlaa uept. Natural Kesources)














CHAPTER 3
IMAGE ANALYSIS


3.1 Digital Image Concepts
There is a great deal of literature on digital images and image processing. One

general image processing reference is Gonzalez and Wintz (1987). In this section, a brief

introduction to key concepts and terminology needed for understanding this paper is

presented. Readers are strongly recommended to check the reference listing for further

information which goes beyond the scope of this paper.

Digital imagery results from the desire to enhance existing pictorial information

and allow computers to interpret pictorial information. A digital image is created by

segmenting a picture into discrete spatial elements, hereafter referred to as pixels. Each

pixel is assigned an intensity level or color. A monochrome or grayscale image consists

of a matrix of pixels that have brightness or intensity values ranging from black to white.

We call this a purely black and white image if only 2 grayscale shades are present: black

and white. Typical monochrome images contain 256 grayscales, where 0 corresponds to

black and 256 corresponds to white. Using computer terminology, an image with 256

grayscale is referred to as an 8-bit image (28 shades of gray), which is an industry

standard. Color 8-bit images allow use of 256 colors out of a possible 16 million. Most

image processing involves only monochrome images, since color adds an extra

complication.

The typical image file types which were used in this study are of three types,

TIFF, GIF, and BIF. TIFF (Tagged Interchange Format File) and GIF files are industry

standards for personal computers, paint packages, scanners, and image processing

packages. Each file contains coded pixel intensity or color data and a header with



16










information about the image size and other characteristics. Most images used correspond

to standard VGA or NTSC monitor resolution with a size of 640x480 pixels. A 640x480

TIFF or GIF 8-bit monochrome image occupies 307 KBytes of memory. Some

applications produce 512x480 images, which occupy 245 KBytes. Image storage and

handling can present problems for computers with limited memory and speed.

To access pixel intensity data, these standard file formats must be converted to

raw data files. BIF (Binary Information File) files contain only raw data. This binary

data can then be converted to ASCII format and accessed directly by the particular user.

All image conversions of this type can be done with software programs like Image

Alchemy, by Handmade Software, Inc.

To convert videotape or live video camera signals to digital images, a frame

grabber is used. A frame grabber digitizes an incoming video signal by converting the

analog input to a digital output. Many frame grabbers have onboard memory buffers

which can convert and store images in real time, or 30 Hz for standard video. This study

utilized a Data Translation DT-3851 Frame Grabber in a Gateway 2000 486DX/50E for

digitization from the Hollywood Beach data videotapes (480x640 images) and a Digital

Vision Computer Eyes Color Frame Grabber for real time digitization in Miami Beach.

Miami Beach images were later converted to monochrome for analysis.

Image processing involves basic operations on images, like contrast and

brightness adjustments, as well as more complicated operations like filtering and edge

detection. Our image processing system centers around a Data Translation software

package, Global Lab Image. This software operates under a windows environment and

provides simple display and advanced processing functions. It is also programmable for

user customization.

Some important image processing functions include equalization and contrast

stretching, intensity profiling, distance determination, image arithmetic, filtering, and

edge detection. Equalization and contrast stretching are important for human perception









of image data, by allowing the user to adjust contrast and brightness for optimal

visualization of image features. The computer can equalize the image using linear

histogram equalization. Intensity profiling provides a plot of image intensities along a

line of pixels in an image. Distance determination is used to calculate pixel-to-pixel

distances for calibration with real world values. Image arithmetic includes addition

(averaging), subtraction, multiplication, division, and logical operations on images.

Filtering essentially applies a mathematical filter (like low pass or Sobel) to image data,

which can help extract information that is otherwise difficult to see. Edge detection is a

filtering process which locates edges of objects that have similar intensity or size

characteristics, or user defined threshold values, which are characteristic of edges in

images.

Several of the above mentioned image processing techniques will be explained in

greater detail in upcoming sections. Many of the particular applications of image

processing require custom implementation. Most programmers favor the C programming

language, since it is well suited for high speed numerical calculations. The computational

software package MATLAB, by The Math Works, Inc., is somewhat slower than C but

has advantages including an ability to store and perform operations on large data sets

(like images), as well as numerous routines for signal processing and image manipulation

and display.

3.2 Photogrammetry Basics

The image interpretation process is based on the geometry of photogrammetric

principles. The location of any point on the image is related to the location of its

corresponding point on the ground, the focal length, the camera tilt, swing, and azimuth,

and the exposure station elevation. This can be expressed as

(x,y) = f(X,Y,Z, fc, T, s, H) (3.1)
where x and y are the image coordinates (relative to the fiducial coordinate system), X, Y,

and Z are the corresponding ground coordinates to be imaged, fc is the camera focal










length, r is the camera tilt (upward from vertical), ) is the camera azimuth, s is the swing

angle (or roll angle) of the camera, and H is the elevation of the exposure station above

the ground plane origin O, called the nadir point. Focal length fc is defined as the

perpendicular distance (in mm) from the optical center of the camera lens to the imaging

surface of the camera. For a standard camera, the imaging surface is film. For a video

camera, the imaging surface is a charge-coupled device (CCD). A short focal length, say

8 mm, results in a wide field of view, while a longer focal length like 80 mm results in a

"zoomed" view. Figure 3.1 illustrates the geometry of the exposure station.


Figure 3.1 Exposure station geometry


For the simplest case, we intend on imaging (and rectifying) the nearshore sea

surface. In this case the ocean surface is assumed to be a plane. The three-









dimensionality of the onshore features in the oblique image can be used for determination

of the parameters of exterior orientation, but will not be accurately mapped to the

rectified image (remember that this transformation applies only to a plane). Radial lens

distortion should not play a major role in the rectification, so long as one does not travel

too far into the distance and the edges of the oblique are avoided. An exposure height of

55 m 180 ft is too low to result in any significant radial lens distortion, according to
Wolf Figure 5-17, p.102 (1983). Lippmann and Holman (1989) presented the equations

for geometrical transformation from image to ground coordinates. For the simple case,

image coordinates p(x,y) corresponding to any ground coordinate P(X,Y,Z) can be

expressed by:



Stn(tan ( ) (3.2)



2 + f2 X (3.3)
Z2 +y2


where Z denotes the relative height of the particular point P with respect to the camera
elevation H, or Z = H Zp. Zp is defined as the elevation of point P with respect to the

ground plane. The application of one known pair of points (x,y) and (X,Y,Z) can be used
to solve for the unknowns fc and T. The camera parameters fc and T are difficult to

measure accurately. The focal length of the camera is unknown because one does not

measure quantities directly from the CCD, but from a computer screen which has an
enlarged display. Thus the camera focal length is altered by an unknown value. By

applying several of the surveyed control points to eqns. (3.2) and (3.3), the unknowns can
be determined with the method of least squares which results in the "best" solution.

More complicated solutions result when the scene geometry deviates from the

simplest case. For our case, most of the useful images from Hollywood Beach fall into









this complicated category. The horizon line is not parallel to the reference axis (say, the

top edge of the image), but is tilted at some angle. Using the rigorous but generalized

equations for the solution of high oblique images (Wolf, pp. 442-453), the geometry can

be solved numerically. However, Wolf points out that the relationships governing the

image-to-ground transformation problem can be based on three independent angles. The

idealized problem equations (3.2) and (3.3) account for the tilt angle. The other angles,

swing and azimuth, can be accounted for outside equations (3.2) and (3.3).

The azimuth angle ( is determined from survey data or estimated and solved as

done previously with control points and least squares. Solution of fc, z, and with these

methods results in typical errors up to 0.5 percent, 0.250, and 0.50 respectively (Holman

and Lippmann, 1989). The swing angle can be accounted for with a basic rotation of the

oblique image to effectively untill" the horizon. Technically, this process is called

converting the fiducial coordinates (x,y) of the image to the appropriate auxiliary axes

(x',y'). Figure 3.2 defines the image coordinate system. Since (3.2) and (3.3) perform the

transformation in the idealized camera coordinate system, the results of their application

lie in the auxiliary image coordinate system, and must be converted back to the fiducial

coordinates. The swing angle s is defined as the angle between the fiducial vertical axis

and the downward end of the auxiliary axis that is perpendicular to the apparent horizon

line. The swing compensation results from a simple coordinate rotation.


x'= xcosO-ysinO
(3.4)
y'= xsin + ycos0
The application of (3.4) accounts for the swing angle, if present, where 0 = s 180.

3.3 Image Rectification
3.3.1 Introduction

Rectification of images is not a new concept, but it is essential to the successful

analysis and interpretation of oblique images. The basics of image rectification can be










found in any text on photogrammetry. Rectification involves the transformation of an

oblique image into a plan (or map) view by graphical or analytical methods.


Figure 3.2 Image auxiliary coordinate system


Oblique images are difficult to take measurements from, due to the varying

degrees of perspective distortion. What one believes to be, for example, a line running

east to west may be at some unknown angle. For this reason, a rectified image can

provide much needed spatial information which cannot be accurately determined from an

oblique image. The current goal is to produce a rectified image of the sea surface which

is on a scaled, square grid. Spatial variations of nearshore features can then be easily

observed and measured for research in the coastal environment.

The rectification process is fairly straightforward once the geometry of Section

3.1.2 is understood. The parameters of exterior orientation are solved, based on the

control points and method of least squares. The next step is to determine the desired

output resolution and location of the rectified view on the original image. This step is


yy
I /
Horizon





S--- --
x
01xi









best left to trial and error. After the output image "grid" and scaling are set, the

rectification can begin. Each rectified grid point P(X,Y) is transformed through (3.2) and

(3.3) to determine the corresponding image location p(x,y). The intensity found at

location (x,y) is mapped to (X,Y). This process is repeated for each grid point, or each

pixel. The loss of resolution is readily apparent with an understanding of the rectification

procedure. Many points in the far field of the rectified grid will receive intensities

mapped from a single point on the oblique image. Results are satisfactory but may be

improved with enhanced image processing techniques (averaging or bilinear

interpolation) during the mapping process.

The final rectified image is on a scaled grid of known dimension. Direct

measurements of desired length scales can be made with the use of graphical techniques

or an image processing package.

3.3.2 Program background

The program rectify, m was written for math package MATLAB for several

reasons. Primarily, MATLAB can support the large data arrays (three 640x480 image

matrices) which are required to solve the problem. When using a PC-486 computer, a

typical black and white image takes about 25 minutes and 9 MB of active memory to

rectify. Careful attention to survey work for the ground control (or ground "truthing") of

the images is imperative to obtaining the most accurate rectified images. Resolution of

the rectified image depends on the height of the camera above the plane of rectification

(mean sea level for my application), the tilt angle of the camera, and the focal length

(zoom) of the camera lens. The following paragraphs describe each step of the procedure

and program.

3.3.3 Camera and view setup

The camera setup and views are left to each user, but several factors must be kept

in mind. First, the camera should be weatherized and placed outside so reflective effects

of windows and dew do not ruin potentially important data (the author recommends









Erdman Video Systems, Miami Beach, FL for the solution of all video installation and

sampling requirements). Second, the camera view should be set so the apparent horizon

line is tangent to some horizontal reference to avoid roll effects in the rectified images.

This roll effect complicates rectification and had to be addressed in the Hollywood Beach

data set. Third, the study area should be split into several longshore/cross-shore segments

and the zoom power of the camera utilized. A zoomed image in the distance will produce

better results, although the ground survey work will increase with each new scene. Care

should be taken when using zoom features of cameras that are not purchased with video

monitoring packages. Each scene's zoom factor must be consistently duplicated to avoid

re-calibration of focal length and tilt for all images.

3.3.4 Surveys and measurements

After setting up the camera and determining the sampling scheme, measurements

of the so-called parameters of exterior orientation should be performed as carefully as

possible. The desired parameters include: camera elevation above a standard reference

elevation (1929 NGVD for our purposes), camera focal length for each scene, and the tilt

angle of the camera axis upward from vertical. Next, ground control points must be

determined (as applicable to the particular setup geometry) and surveyed. For

completeness, one should survey a minimum of 4 points with as much spatial coverage of

the camera view as possible. It is best to utilize permanent features which are easily seen

in the view and tied into standard survey monuments (like DNR R-monuments). The

azimuth angle between the camera system x-axis Xg (perpendicular to the camera view)

and the ground system x-axis E (usually directed offshore) needs to be determined.

Estimate angle j and survey several points relative to both coordinate systems (these can

correspond to the above ground control points). Finally, survey the point which

corresponds to the center of the camera view, if possible. As a general rule, the more

points surveyed will produce the best final results. The author again emphasizes the

importance of thorough surveys.









Another useful tool for the verification of the rectified images is the aerial

photograph. If recent aerials are not available, it is recommended that several aerials of

the study area be taken. The study site and camera location should be clearly indicated or

obvious to the naked eye on the photographs. Flags, bright buoys, and boats are good site

markers. These photographs serve as a basis upon which the rectified images can be

ground truthed.

3.3.5 Rectification program

The rectification program is versatile; it can be used for any image size, typically

640x480 or 512x480 images. The program has 2 parts: rectify.m can be run when a new

camera setup or view is implemented. The parameters of exterior orientation and output

grids are determined and saved as two MATLAB files inp.mat, containing the camera

focal length fc, tilt angle T, elevation H, image size and resolution, camera azimuth angle

* and the offsets of the output grid in camera coordinates, xo and yo. The offsets xo and

yo are distances from the camera coordinate origin to the standard ground coordinate

system origin, measured in the camera coordinate system. File inp2.mat contains the

ground location grids (Xg and Yg) of the rectified image in camera coordinates. The

second part, remap.m, can be used after the initial run of rectify.m. Remap.m loads the

two input files so the input phase of the original program is bypassed.

When running rectify. m, the user is prompted for various inputs, including initial

guesses for fc, (if required), and control points. All input data must be entered in

array column form in consistent units. The first part of the program solves for the

parameters of exterior orientation by applying the method of least squares to the known

survey control points. More points spread over the expanse of the screen increases the

accuracy of the results. The second part of the program sets up the rectified image grid

and loads the input image. The final part of the program applies the image transformation

equations and maps the correct pixel intensities to the final image grid.









The calculated focal length will depend on the image size being used. If one was

to measure distances directly from a 1:1 positive, the focal length calculated by the

program would be the focal length set on the camera's zoom lens. However, we will

generally measure image coordinates in pixels from a digitized image on the computer

screen, so the calculated focal length will be considerably larger than the camera's actual

focal length. The particular computer screen will have to be calibrated. In doing so, one

essentially determines the size of each pixel. Image coordinates measured in pixels can

then be converted to proper units within the program.

The output resolution and location of the final grid origin are user inputs. The

output image resolution (meters/pixel) can be any value depending on the detail required.

I have used resolutions of 1 m/pixel and 2 m/pixel for most cases. Figures 3.3 and 3.4

present an example of a digitized oblique image from the Hollywood data set and the

corresponding rectified image at 2 m/pixel resolution, respectively. The output grid

origin is moveable so that one can effectively "window" an entire image to create a

mosaic of more detailed images. Remember that the output resolution degrades fairly

quickly with increasing distance from the camera nadir point and increasing camera tilt.
































3.3.6 Program Calibration

The rectification program was tested and calibrated before applying it to any field

data. A test case was set up inside a lab at the University of Florida. A rectangle was

marked on the floor, and a Hi8 video camera was set with zoom, tilt, and azimuth similar

to normal field conditions. Figure 3.5 shows the test case geometry. The camera was

0.692 m above the ground plane (the floor), the zoom was set to 11 mm, and the azimuth

4 was -27.5. An image was digitized as shown in Figure 3.6. A straight edge was

placed along the principal line of the camera view for clarification. With known control

points on the ground and image, the image was rectified using Rectify.m. The output

resolution was set to 2.5 mm/pixel on a square pixel grid. The rectified image is shown

in Figure 3.7. The rectified image was loaded into Global Lab Image, a windows-based

image processing software package, and checked for errors.

It was found that the RMS normalized difference between rectified image

estimates and actual measurements was 5.2%. The distortion of the image in the far field

is evident by observing the crispness of the bottom edges of the rectangle and the relative









fuzziness of the top edge. This distortion is due to the break down of resolution in the
image as we approach the "horizon" and possibly some lens curvature effects. It is noted

that although the crispness of the lines degrades, the overall thickness of the lines remains

constant. The actual thickness of the tape was 1.9 cm, and the average measured
thickness of the tape on the rectified image was 2.0 cm. Similarly, the actual and
measured lengths of the dark-colored ruler were 0.61 m. Therefore, precise edges of
subjects in the far field of the rectified images may not be accurately detected, but their
location in the ground plane and relative size should be accurate. Also, the original is
accurately represented after rectification. The rectangle was not precisely squared off
when it was set, so the angles between the sides are not exactly 90. Note also that the
black wedges on each side of the rectified image result from the edges of the original
image and define the camera field of view.


-- 0.8 m


Tape Rectangle


+


Camera Field of View


1.06 m




Camera
0.3m


Figure 3.5 Test setup schematic


27.50
\7


i


r





I










A second test case was set up at Florida Field. To simulate typical field

conditions, the camera was placed 9 m above the level of the field and an oblique view

was taken. An Omni Total Station was used to survey locations of the known control

points to determine the parameters of exterior orientation. The oblique image is shown in

Figure 3.8. The image was rectified as shown in Figure 3.9. Two difficulties arose

during the test which affect the result. First, the field has an 8 inch crown which is the

difference in elevation between the sidelines and the centerline of the field. The camera

location and large tilt angle (840) cause a distortion of the lines as we move towards the

upper right of the image in Figure 3.8. The crown causes the lines to appear more curved

than a normal perspective on a plane. Second, the camera mount on the tripod was not

level and was not determined until after the test was completed. The swing angle

compensation (Equation 3.4) was based on the assumption that the angle of the wall on

the far end of the field was parallel to the horizon line, since no other reference was

available.

The rectified output reflects the true locations of points on the field with the loss

of resolution typical of the rectification process and the influence of the crown of the

field. Measurements were made on the rectified image to check the accuracy of the

rectification program. Errors were assessed by evaluation of the RMS normalized

difference between actual measurements and video-based estimates:



RMS error = (3.5)
n


where xi are the normalized errors between image estimates and actual quantities and n is

the number of observations. For the Florida Field test case, errors were calculated with

equation (3.5) as 0.77% for undistorted angular measurements and 4.17% for length

measurements.




































Figure 3.8 Florida Field oblique shot










3.4 Other Methods

Alternatives to the image rectification method are available for quantification of

features in images. The two suggested methods include a calibration and measurement

method with a commercial image processing software package, Global Lab Image, and

generation of a perspective grid which is superimposed over image data. These methods

are mentioned for completeness but are not actually applied. Further investigation is

required to use these methods in the future.


3.4.1 Global Lab Image

One alternative method to rectification is the straight implementation of an image

processing package that allows calibration. Data Translation's software package Global

Lab Image includes functions for perspective calibration of images. The operator can

pick up to eight known control points on an image and input real world two-dimensional

values for them. A least squares method solves image perspective transformations, based

on the input values, to produce correct distance measurements. The limitation is that the

function only solves the perspective transformation in a plane. Thus, all control points

and measured points must be coplanar.

This technique was attempted during the early phases of this research, but the

limitations proved to be too restrictive. Measurement problems result from the large

perspective distortion of the high oblique images. Even if the calibration is successfully

implemented, making distance measurements from the obliques is very difficult. In the

coastal views which are used here, one cannot ascertain the cross shore and longshore

directions at each point desired. Drawing cross shore transects at several points along an

oblique coastline view is nearly impossible. The oblique image calibration method was

abandoned for this study, but may prove more useful when the perspective problem is

more subtle.










3.4.2 Overlay Grids

A second measurement technique that was considered during the early phases of

research was overlay grids. These grids can be constructed from analytic formulae or

graphical packages which generate them. Perspective grids, unlike square grids, account

for the distortion that accompanies an oblique image. By selecting an appropriate


Figure 3.10 Overlay grid


reference axis (like the shoreline), perpendiculars can be constructed based on

photogrammetric principles (Wolf, 1983). Figure 3.10 is an example of a perspective

grid, generated by Autocad. To generate a grid of this type, the camera elevation and

location from the reference axis must be known. Problems result during overlay

procedures, as limited software is available for image overlay. This technique was also

abandoned for the more versatile rectification approach of Section 3.3.















CHAPTER 4
RESULTS AND DISCUSSION


4.1 Introduction

The video data were collected at the sites and times as described in Chapter 2.

Pertinent images and events were then selected for analysis, based on observation of the

recorded data. Images were analyzed with the rectification method outlined in Chapter 3.

As a result of the large amount of quality data collected and varied interests of the

sponsoring parties, the analysis was broken down into three major sections. First, site

documentation deals with the generally qualitative uses of video monitoring and the

handling of such large data sets. Second, nearshore features presents some measurement

capabilities of coastal morphology which are of interest to coastal engineers, including

shoreline position and sandbar identification. Third, turbidity phenomena investigates the

use of video monitoring for mapping of natural and man-induced turbidity events.



4.2 Site Documentation

The first and most obvious use of video monitoring is site documentation. The

simple gathering and archiving of images provides a database of valuable information.

The problems which must be addressed include image storage and archiving, data

extraction, and then presentation methods.

4.2.1 Image data sets

Video monitoring generally results in large data sets, since sampling can be set to

virtually any frequency desired. Especially in the applications of coastal engineering,

many of the processes which are desirable to record occur during relatively long time










periods. For example, sandbar morphology and shoreline change are monitored over

weeks or months. Investigators want a thorough data set, so records may be taken each

hour for several weeks or months. In shorter-term investigations, like rip currents or

swash observations, the overall time period may be only several hours. The image

sampling frequency may be once per minute, or even once per second. To be more

specific, the Hollywood Beach image data set is used as an example.

The Hollywood data set consists of 11 scenes of 0.7 second video records which

were recorded at 15 minute intervals. Data were collected over a 7 month period from

April through November 1991. Each day recording began at 7 AM and continued until 7

PM. If we take each 0.7 second record as being one "image," (since one would probably

digitize one representative frame from each 0.7 second record) then the total number of

acquired images is 110,880. Due to this copious data set, only one scene was used for

further analysis. The video records fill over 10 two-hour Hi8 videotapes. Original

records were recorded in series on tape and encoded with date and time information. The

master tapes were later "decoded," a process that involves placement of the records from

each scene in chronological order. This computer-assisted process creates a time lapse

video of each scene. Time lapse records allow quick viewing of scene conditions and

changes over a relatively long period of time. Further image manipulation from the tape

records is discussed in forthcoming sections.

The Miami Beach data set consists of eight scenes of purely digital images which

are compressed with JPEG compression to a size of 30 KBytes. No tape was used, so

storage and handling problems result from digital data files. In order to use images, each

file must be decompressed (expanded). Image viewing software like Image Alchemy or

Vuimage Plus is then used to look at the data. Several software packages allow the

creation of slide shows from image sets. The advantage of purely digital images is the

ease of access to any particular date and time. The disadvantages include disk storage

space and the inability to see real-time activity.
































!ach under typical conditions


4.2.2 Event documentation

Without entering the quantitative realm of video monitoring, the qualitative

aspects of the video data are numerous. Site documentation includes assessment of:

1. Local conditions

2. Beach traffic

3. Beach activities

4. Shoreline change

5. Predominant coastal features

6. Strange events.

Each of these items can be qualitatively estimated from the video data. Presented next

are some examples of each.

4.2.2.1 Local conditions

Figures 4.1 and 4.2 present a comparison of images from Hollywood Beach to

illustrate the assessment of local conditions from video data. Image 4.1 shows an up










coast zoom from late May 1991, under typical weather and crowd conditions. Typical

conditions for South Florida in late May include sunny skies and calm seas. Image 4.2 is

a similar view of Hollywood Beach during a large storm system that occurred between


May 19 and 22. Note the wide surf zone and numerous offshore whitecaps. Winds from

the southeast at 20 to 28 mph and large waves of 7 to 10 feet were reported in the

dredger's log during this time period, as nourishment operations had to be postponed. An

in-situ record of significant wave height (Dompe and Hanes, 1992) during late May is

shown in Figure 4.3. A peak significant wave height of nearly 2 meters is quite large for

south Florida. This ability to qualitatively verify in-situ measurements with the video

records is a useful application of video monitoring. The video data permits a visual

comparison of conditions like wave size and direction with in-situ measurements. A











video monitoring system was also utilized during the Vilano Beach experiment in spring

1992 for verification of in-situ instrument data.


Significant Wave Height




0 ------- ------i----------------- i ----------
E 1


18 19 20 21 22 23 24 25
May 1991

Figure 4.3 In-situ data from Hollywood Beach, FL (after Dompe and Hanes, 1992)





4.2.2.2 Beach traffic

Another general use of site documentation video records is the ability to measure

beach use. In Figure 4.1, a typical day at Hollywood Beach, the number of beach goers

can be estimated. Using the video or digital original for optimal clarity, people on the

beach and in the water can be counted. This particular scene is far in the distance. A

much better result can be obtained for the traffic application if the camera is closer to the

point of interest. Knowing the area of a given beach segment, the average number of

people per unit area can be determined. Of interest to government agencies may be the

number of beachgoers, for cost and benefit analysis. With the beach nourishment in

mind, the benefits of a wider beach can be easily quantified with the use of video records

of beach traffic before and after the nourishment.

4.2.2.3 Beach activities

In addition to beach use by people, the video records present a comprehensive

view of various beach activities. First is the routine beach maintenance, which includes

garbage removal and sand grading. Second is the ability to document the beach










renourishment procedures (in the Hollywood Beach case). Figure 4.4 is an image from

June 16, 1991. The dredge discharge was placed around 800 feet south of DNR

monument T-1 15 from 3 PM until midnight. In Figure 4.4, one can see the discharge

point in the lower right. The sand/water slurry mixture is deposited so that the dike

system allows the water (and fine sediments, resulting in coastal turbidity) to escape to

sea, while the higher quality sediment remains on the beach. On the left are bulldozers

which grade out the newly placed sand to the specified design slope. Also in the image

are the discharge pipe and temporary construction office on the beach. Figure 4.4 is only

a single image. The time lapse video allows the viewer to actually see the discharge point

operation and movement, the bulldozers at work, and the new beach being built.










4.2.2.4 Shoreline change

From video records we can also make simple, qualitative analysis of shoreline

change. From wide views or close zooms we can see the long-term shoreline changes as

well as tidal fluctuations throughout the day. Individual images are difficult to use for

this purpose and the time lapse videos are much easier for qualitative analysis. A more

quantitative shoreline position analysis is discussed in Section 4.3. Figure 4.5 presents an

image of the beach down close. These images were not calibrated and were not used for

subsequent analysis, but the potential for detailed measurements is apparent. Swash

motions are easily recorded and quantified. A detailed method is presented in Holman

and Guza (1984) and Aagaard and Holm (1989).










4.2.2.5 Predominant coastal features

Coastal features are of interest to the coastal engineer because an understanding of

their formation and forcing mechanisms may reveal coastal circulation and nearshore

sediment transport patterns. Field data have been collected with standard

instrumentation, but the use of time-lapse video adds the advantage of being able to see

changes over any period of time. Sampling frequency can be set for any observation

requirements. For example, the Hollywood data set contains several views. Figures 4.6

and 4.7 present similar images taken over about a month time period. Figure 4.6 is an

image from Hollywood Beach, in front of the Hollywood Beach Hotel complex, taken on

April 26, 1991. Four large scale, symmetrical beach cusps can be seen along the

shoreline. A large storm which occurred between May 19 and 22 effectively wiped out

the cusps. Figure 4.7 shows the effect of the storm on the cusps. Figure 4.7 was taken on

May 23, 1991, after the storm passed and the seas started to resume typical South Florida

conditions. Other predominant features include rip currents and sandbars.

4.2.2.6 Strange events

Documentation of strange or random events is another area of qualitative video

monitoring. Several examples have been discovered during this research. Most of the

time, these events are not discovered until the video deployment is over and the tapes or

images are reviewed. One instance occurred during a test deployment in Miami Beach.

On March 24, 1991, a freighter ran aground on South Beach during a storm. The video

captured the events surrounding the grounding and removal of the ship on March 31.

Figures 4.8 and 4.9 show the ship when it ran aground and its removal, respectively.

During the eight days that the ship remained aground, a tombolo formed between the

beach and the ship. This event also created a good surf break as evidenced by the large

number of surfers seen in several images (Figure 4.10). Figure 4.11 is another scene from

the Miami test deployment including a bungee jumper caught in free fall. Perhaps the











































wood beach cusps, April


nywuuu oeacn cusps, May Lj, i I




































Figure 4.8 Cargo ship grounded on















03-31-1991


155f-










most pertinent event to coastal engineering is the data from Hollywood Beach that

includes the great Halloween storm of 1991. The in-situ instrumentation was not

deployed during the storm, but the video monitoring system did capture the event.

Figures 4.12 through 4.14 present several images from the Halloween storm. Figure

4.12 shows the large waves generated by the storm, with four discrete breaker lines. This

type of wave breaking is uncharacteristic in South Florida, with the exception of winter

cold fronts (which can generate large northeast swells) and tropical activity. The

shorebreak and setup which caused coastal flooding are obvious in Figure 4.13.

Increased fine sediment suspension results in nearshore turbidity, as shown in Figure

4.14. The greater relative intensity in the nearshore area corresponds to elevated

turbidity. Also note the periodic wave trains approaching the shoreline.


yure ,4.1z Ialloween storm waves, uctooer 1i,






































een storm swash and flooding










4.3 Nearshore Features
This section deals with the more quantitative aspects of video monitoring. The

video monitoring system could be used for any specific study of nearshore features, but

only the most general are presented here. The applications include mapping of the

shoreline position, bars, and surf zone width. The Hollywood Beach data is used for the

shoreline position analysis. The Miami Beach data, which incorporates a time averaging

technique, is used for the study of bars and surf zone width.

4.3.1 Shoreline Position

Shoreline position and shape are important to coastal engineers for obvious

reasons. Shoreline location is an indicator of erosion or accretion rates at a specific place.

The shape of the shoreline can reveal much information about the local wave and current

conditions of a particular beach. It is desired to map the shoreline remotely, over any

given period. Using video monitoring, this can be done without performing time-

consuming profiles or expensive aerial photographs.

The Hollywood Beach data set provided an excellent opportunity during the beach

renourishment of summer 1991. Select video images were rectified at a resolution of 1

m/pixel for this section. The beach renourishment began to the south of the camera

location in late April and finished well north of the site in August. Selected images

include April 26, June 13 through 19 (when the dredge discharge passed by the camera

site), and August 16. Figure 4.15 shows a series of images over this time, with dates

indicated in the upper right of each image. The corresponding shoreline plots are

graphically represented in Figures 4.16 and 4.17. Figure 4.17 shows representative

shoreline positions for April, June, and August 1991. Figure 4.18 shows a representative

shoreline for each day of interest when the nourishment passed the site. The cross shore

distance shown is taken from a baseline along the west edge of the broadwalk on

Hollywood Beach, close to the east-west coordinate of DNR Monument R-1 16.












































































Figure 4.15 Hollywood Beach nourishment sequence


420





















61/15





















6/19











-Waw.",













The longshore coordinate (in meters) is arbitrary, but approximate longshore locations of

DNR/CP&E Monuments are shown as reference points.


Figure 4.16 Hollywood Beach shoreline pre- and post-nourishment




Shoreline Position
100
6/13/91
x 6/14/91
+ 6/15/91
90 o 6/17/91
S* 6/19/91

80-



. 70



60 -



50
0 100 200 300 400 500 600 700

Longshore Coordinate (n)

Figure 4.17 Hollywood Beach shoreline during nourishment


Shoreline Position
95

90 4/26/91
6/13/91
85
85 8/16/91
E
80-

E 75

60
E 70

0
55-

5 T-115 R-116

50
0 100 200 300 400

Longshore Coordinate (m)


R-117


I I










The reader must remember that the plots shown are digitized from a representative

image on each day chosen. Therefore, the effect of swash motion and tides are not taken

into account. To better represent the mean shoreline position for any particular day, an

average of all the day's images should be taken. The averaged image would remove the

effect of the tidal and swash motions around the mean shoreline. This method would not

be suitable in this case, since the shoreline position may change drastically over only a

couple of hours during the nourishment construction phase.

Qualitatively, the shoreline had a wavy or rhythmic appearance in April. The

widest points occurred around T-115 and R-116.5. Significant storms occurred in May

and June which reshaped the coastline as shown on June 13. The widest areas remained

around T-1 15 and shifted from R-l 16.5 to R-116. After the project was completed, the

area near T-l 15 had received the least amount of material, while the previously narrow

areas received a larger fill volume. This is in general agreement with the construction

plans, which show a wider toe of the fill around R- 116 relative to the surrounding

monuments (CP&E, 1990). According to Coastal Planning & Engineering's 6 Month

Follow-Up Study (1992), moderate erosion occurred at R-116 and accretion occurred at

T-115. Barring background and individual erosion events, this scenario could be inferred

from the August 16 shoreline and the general spreading out of the fill.

The shoreline data can also be checked with the nourishment construction surveys

that were taken as the fill was placed. Figure 4.18 presents raw fill beach profiles that

were taken by the construction crew as the fill was placed. Surveys were taken on the

dates shown, at the indicated monument. Specific times were not given, but the date of

the profiles can be used to check the validity of the shoreline position extracted from

video data. Figure 4.19 shows the shoreline position according to the construction

surveys, and includes the raw fill and the graded fill. The jagged curves result from the

sparse data points, which exist only at the monuments.















Beach Fill Raw Profiles
4
R-117 6/14
o 116.56/15
+ R-116 6/17
x 115.56/19
2
z
* T-115 6/19

E
- 1


S 0



-1


--2
0 20 40 60 80 100 120 140

Range From Monument (m)

Figure 4.18 Hollywood Beach construction profiles (courtesy CP&E)









Shoreline Location
100

o Raw Fill
95
Graded Fill

E 90

E
: 85
0
E 80-
L.
a
75-


115 115.2 115.4 115.6 115.8 116 116.2 116.4 116.6 116.8 117

Longshore Coordinate/Monument

Figure 4.19 Shoreline location from construction surveys










The construction survey data show good agreement with the shoreline positions

found from the video data. For example, the survey transect off R-116 shows the water

line to be around 96 m from the monument. The water line is taken at 0 NGVD. A

corresponding image transect, measured from video data at R-116 on June 17, shows the

shoreline at 95 m from the monument. Image transects were taken with Global Lab

Image profile tool. The profile tool plots non-dimensional pixel intensity along any line

drawn on the image. Image water level is taken as the horizontal location where the

intensity abruptly changes value due to the intersection of water and sand. The presence

of a shorebreak eases this process, since it causes a bright band along the shoreline.

Frequently there will be a series of intensity peaks and valleys along the beach face (of

the rectified image), due to variations of sand moisture content and seaweed lines. The

computer operator can visually asses the image to resolve ambiguities in the intensity

profile and successfully select the shoreline.

The transect at R- 116 has the best correlation of the transects selected, due to the

convenience of the image baseline set parallel to the east-west ground coordinate of R-

116. The image baseline runs parallel to the west end of the broadwalk on Hollywood

Beach (about N 50 E) and passes through R-116. In order to check the shoreline position

at the other monuments, the survey data must be adjusted to the image baseline. Table

4.1 presents the results of the transects measured from the images. Monuments R-l 17

and T- 115 have reduced accuracy, since they lie just outside the effective video

window."

Table 4.1 Shoreline position comparison
Monument Survey Date Shoreline Pos. (m) Shoreline Pos. (m) Shoreline Pos. (m) Error (m)
From Survey Data Adjusted to Images From Image Data

T-115 6/19/91 86 80 -6
115.5 6/19/91 69 79 82 3
R-116 6/17/91 96 95 -1
116.5 6/15/91 75 83 79 -4
R-117 6/14/91 92 77 80 3










The RMS difference between the image based estimates and the survey based

estimates is 3.77 m horizontally. As stated previously, the image data profiles do not

have the influences of tides or swash action removed. Also, the problem of considering

differences between "mean water level" and "NGVD" can effectively reduce accuracy.

According to the NOAA Datum Division, Mean Lower Low Water for Haulover Pier

(nearest site to Hollywood Beach) is 0.28 m below NGVD. From tidal data in mid-June

1991, the mean water level was calculated to be 0.297 m above MLLW. Therefore, the

observed mean water level should be less than 2 cm above NGVD.

A more detailed study would need to be performed to get precise results; however,

the image measurement technique looks at the horizontal position of the shoreline, over a

large spatial area. This position can fluctuate depending on the beach slope, tide, and

wave conditions. As an example, the average beach fill slope is considered. From the

profile data, the average beach face slope of the raw profiles is around 0.097 and the

graded slope is around 0.076, averaging to 0.087. The average tidal range for mid-June

1991 was around 2.5 feet. As a worst case scenario, three feet vertical on a slope of 0.087

becomes 34 feet or roughly 10 m in the horizontal. With this information, the errors

resulting from image-based estimates are within the range of the tidal influence.

The maximum resolution of the video technique (in this case) is limited to 1 m in

any direction, since the images are rectified to 1 m/pixel. The dominant advantage to the

video technique is the large spatial scale and the flexible temporal scale at which such

information can be extracted. If a particular subject was under study, the VMS setup and

image analysis could be tailored for the specific problem. For example, the images could

be rectified at 0.5 m/pixel for a more detailed result.

4.3.2 Bars and Surf Zone Width

Sand bar morphology is important to coastal engineers and scientists because the

understanding of them is imperative to a total assessment of nearshore hydrodynamics

and sediment transport. The study of bars and surf zone width with video monitoring










techniques has been investigated by Dr. Robert Holman's group at Oregon State

University. The highlights of their intensive research were presented in Section 1.4. In

this paper, a similar technique is implemented for completeness of the video monitoring

subject matter.

The Miami Beach VMS deployment included the time averaging scheme as

outlined in Section 2.2.2. Selected scenes were digitized 10 times over 5 minutes (10

frames per scene) and digitally added together for a time average. The resulting images

reveal the areas of wave dissipation and the bar location, due to the preferential breaking

of waves over the sand bar (Lippmann and Holman, 1987). The shoreline is also

enhanced by the persistent shorebreak. This technique also removes the effect of the

swash, since we average over multiple wave periods. Figure 4.20 presents an image from

the Miami Beach deployment, taken on December 20, 1992. The shoreline is indicated

by the white band which corresponds to the shorebreak, and the offshore bar is clearly

visible with rhythmic characteristics. Some nearshore turbidity plumes are also present in

the upper left corer of the view. The heaviest wave dissipation areas clearly indicate the

bar location, but the areas of less frequent wave breaking or reforming shorewardd of the

bar) are apparent in the nearshore foreground. Use of a longer period of averaging or a

larger number of averaged images would eliminate the small amount of noise present, but

these "secondary" breaking zones may be of some interest and would not be visible in a

true time exposure.

The Miami Beach VMS data set is large and more difficult to review than

Hollywood Beach since the images are all digital. A sampling of the time averages were

examined and selected for the basic results presented. Figure 4.21 includes a series of

rectified images spanning the period between December 1992 and May 1993. Dates are

marked in the upper right corer of each image. All images are rectified at 1 m/pixel.

Qualitatively, the bar began with crescentic segments in late December. The bar retained








































a cross shore width of 30 to 45 m through the series. In early January, the bar seems to

have straightened considerably. Both of the previous images were collected during

relatively larger wave conditions, since the bar is well-defined. By January 22, the bar

has returned to its characteristic crescent shape. The bar definition is less due to smaller

incident waves and could be improved with a larger number of averaged frames. The

shape of the bar is also described in Section 4.4.1.2, rip currents, as rips are documented

on January 23 at the longshore locations which correspond to the offshore extents of the

crescentic bar segments.

The months of February and April were characterized by extended periods of little

wave activity. If the incident waves do not break over the bar (indicating a small incident








































































Figure 4.21 Miami Beach bar progression


1 04 93










wave height), then the time average technique provides no additional information about

the bar. This is a drawback of the technique, but little wave activity generally implies a

minimal amount of sediment transport and bar movement.

On March 15, the bar again has taken its characteristic shape, with a slight

shoreward dip in the middle of the image. May 30 provides an interesting scene, where a

persistent southeast wind had generated some good-sized waves. The curved pattern may

be due to rip currents or a wave reforming and breaking process. According to Lippmann

and Holman (1989), a combination of high waves and wind has a tendency to increase the

errors associated with the cross shore location of the bar. Under these conditions,

unexpected results may occur as foam may be blown shoreward of the dissipation zone,

effectively weighting the bar location landward.

In order to illustrate the general concept of the quantitative aspects which result

from the time averaging technique, a cross shore intensity profile is taken with the use of

an image processing system. Figure 4.22 presents the intensity plot and corresponding

surveyed beach profile which were taken at the same longshore coordinate on Miami

Beach, just south of 25th Street. The survey profile was performed on December 12,

1992, and the intensity plot was taken from an image on December 20. Beach profiles

were only taken on two occasions, December 12, 1992 and January 14, 1993, both during

periods of little wave activity. A general comparison can be made since little wave

activity occurred until December 17, so only small sandbar movement was expected.

Both transects were taken near mid-tide, at similar water levels.

The intensity plot in Figure 4.22 is non-dimensional, so only relative values on

the y-axis are used. The apparent correlation of the magnitude of the survey and intensity

profiles is pure coincidence. The intensity profile magnitude was scaled to fit the range

of the survey profile. Intensity peaks correspond to brighter areas of the image, including

wave dissipation zones. Moving from left to right, a series of peaks and troughs along











Miami Beach and Intensity Profiles
3
3 i------------------------------

Image Intensity 12/20/92
2
*Surveyed Beach Profile 12/12/92




SiWater Level
0 -----------------------------------------------------------tL--


-1


-2 -


-3
0 20 40 60 80 100 120 140 160 180

Range From Flag (m)

Figure 4.22 Comparison of image intensity and beach profile



the berm and foreshore due to onshore objects (like vegetation and high tide seaweed

lines) has been omitted for clarity. The intensity maxima around range 50 m corresponds

to the shorebreak, and correlates well with the location of sea level on the beach profile.

An area of relative darkness (range 55 m to 115 m) corresponds to the trough area and

little or no wave breaking. The offshore bar is apparent on both the intensity and beach

profiles. The landward skewing effect of the bar crest by around 10 m is evident,

probably due to the persistence of foam in the trough area (Lippmann and Holman, 1989).

The overall width of the bar is consistent for both the image intensity and the beach

profile, around 45 m. Similar profiles could be taken at various intervals along the

images to determine the bar location, width, and surf zone width spatially and temporally.

Overall, the time averaging technique is effective for the qualitative assessment of

sand bar morphology, as well as some general quantitative analysis. Particular attention

must be paid to the local weather conditions as they are not as obvious in the time










averages as in instantaneous images, and it has been shown that the conditions can affect

the results of the time averaging technique.


4.4 Turbidity Phenomena

According to Dompe (1993), turbidity is defined as a measure of the reduction in

the clarity of water due to the scattering and absorption of light by suspended particles.

Many factors account for differing levels of coastal turbidity, including particle

concentration and physical characteristics. The particles which make up turbid water may

include sand, silts, clays, and any variety of microscopic organisms or organic matter.

The great concern over turbidity levels in the coastal environment stems from the

fact that increased levels of turbidity, due to natural or man-induced events, are

detrimental to some benthic communities. During the Hollywood Beach renourishment

of summer 1991, in-situ instrumentation was deployed for a comprehensive turbidity

study, which included the causes and effects of fluctuating turbidity levels (Dompe,

1993). The instrument packages recorded data at two specific sites in 5 and 10 meter

depth. The video data from Hollywood Beach provides the ability to see the spatial

variations in turbidity patterns. The current goal is to map the turbidity patterns to

increase understanding of turbidity motion and its forcing mechanisms in the coastal

environment. Turbidity phenomena have been grouped into two categories for this study.

Readers should remember that although natural and man-induced turbidity phenomena

are categorized independently, they are not mutually exclusive events.

4.4.1 Natural Turbidity Structures

Natural turbidity structures, as pertinent to this paper, are defined as organized

and well described patterns of elevated turbidity which result from natural events. These

natural forcing events include, but are not limited to, wave activity, currents, wind, and

tides. Besides the normal background turbidity levels of any location and time, two

regularly occurring turbidity features are classified as natural: lobes and rip currents.









4.4.1.1 Lobes

Natural turbidity is always present as so called "background" turbidity,

qualitatively estimated by the "cloudiness" of the water and quantitatively measured in

Nephelometric Turbidity Units (NTU). A visual distinction is possible in the relatively

clear water of Hollywood Beach, which is labeled class 3 water (extending from Jupiter

Inlet to the Dry Tortugas). The video data from Hollywood Beach contains unique events

that document the propagation of turbidity fronts, made possible by the clarity of the

coastal waters. These events produce turbidity patterns of a finger- or lobe-like structure,

hereafter termed turbidity lobes.

Although many such events occur throughout the video data set, two of the most

dramatic are presented here. The first occurred on April 26, 1991, which was only two

days after the beach renourishment had begun (several miles south). Figure 4.23 presents

the rectified views of the lobe progressions which occurred between 3 PM and 6 PM.

The letters in the upper right of each image correspond to the plots in Figure 4.24. Figure

4.24 contains a graphical representation of the progression, including the mean shoreline

and dominant areas of wave breaking for April 26. The wave breaking zones were taken

as an average for the images used.

At 2:51 there is the persistent nearshore turbidity that extends to 188 m offshore.

This longshore band of turbidity is present under typical conditions at Hollywood Beach

and usually extends beyond the surf zone, to a depth around 6 m (20 ft). From profile

data at Hollywood Beach, this distance from shore corresponds to a somewhat steeper

area of the profile, offshore of the outside bar, where the depth drops from about 3 m to 6

m. The persistent band of turbidity remains during periods of relatively small wave

action, seemingly forced by offshore-directed circulation and diffusion until the energy

levels decrease to allow particle settling.

At 3:51 three distinct lobes form with longshore spacings of 320 m to the north

and 280 m to the south. A shorebreak is distinctly visible along the shoreline, and the
































































Figure 4.23 Turbidity lobes propagation, April 26, 1991












E 0 2:51 pm 3:51 pm
0 A
A E
Ui
0
2 500- 500
%c

o oo oo co oo oou o ooo no o a o
3 1000 i 1000
0 500 1000 0 500 1000



04:06 pm __4:21 pm

SF G
CO

e 500 500 -
o


[ 000 0 % -
uo oo a oooo
.' 1000 1000
0 500 1000 0 500 1000




0___ 4:51 pm 0lte 5:21 pm
SI K

a


wae oa oe c o n on a e o booopO oo 0n 0 a oro a toagoot
3 1000 -----------0-- 1000 -
0 500 1000 0 500 1000
Longshore coord. (m) Longshore coord. (m)


mean shoreline seaward turbidity limit a wave breaking zone

Figure 4.24 Plotted turbidity lobes progression, 4/26/91




waves tend to be concentrated on the "bumps" (shoreline deviations from a straight

beach) along the shoreline. The offshore turbidity lobe peaks occur offshore of the

concentrations of breaking waves. The turbidity lobes propagate eastward while the

central and south plumes merge by 4:06 PM. By this time the visible lobe tips are

separated by about 396 m. The northmost plume is visible but difficult to measure since

it extends north off the view. The surf zone appears wider during the middle frames of

the six shown. From 4:51 through 5:21 PM, the lobes continue offshore until they merge











become less distinct. The waves decrease and the northern lobes overtake the southern

lobe.

Table 4.2 presents the results from an analysis of the image data. The seaward

extent of each feature was identified and recorded by an operator with the aid of an image

processing system. Pertinent displacements alongshore (dy, + southward), cross shore

(dx, + offshore), and absolute (dz), elapsed time (dt), as well as corresponding velocities

are calculated and tabulated.

To be complete in the data presentation, Tables 4.3 and 4.4 present the in-situ data

from the instrument packages and the wind from Miami International Airport from April

26, 1991. The data in Table 4.3 were taken on the inner site at Hollywood Beach, which

is about 300 m offshore of Broward T- 15, and the outer site, which is about 0.5 miles

southeast of the inner site. Positive values of U and V, in-situ current velocities, are

onshore and longshore down the coast (southerly). Wave direction is denoted with

standard compass headings. NTU turbidity values were measured 0.77 m from the sea

bed for the inner site, and 0.45 m for the outer site. The tide values shown in Table 4.4

were taken from Tidemaster, a commercial tide prediction software package.





Table 4.2 Data summary of turbidity lobes, April 26, 1991
Images dt (s) dx (m) dy (m) dz (m) Vx (m/s) Vy (m/s) Vz (m/s)
North Plume
E-F 900 16 -22 27 0.018 -0.024 0.03
F-G 900 36 38 52 0.04 0.042 0.058
G-1 1800 68 86 110 0.038 0.048 0.061
I-K 1800 64 32 72 0.036 0.018 0.04
Averages 0.033 0.036 0.047

South Plume
E-F 900 58 54 79 0.064 0.06 0.088
F-G 900 40 42 58 0.044 0.047 0.064
G-1 1800 64 118 134 0.036 0.066 0.075
I-K 1800 62 58 85 0.034 0.032 0.047
Averages 0.045 0.051 0.069











Table 4.3 In-situ data summary, April 26, 1991
Site Inner (5 m depth) Outer (10 m depth)
Time 12:00 PM 4:00 PM 12:00 PM 4:00 PM
Hs (m) 0.2548 0.5596 0.2441 0.5599
( )- 83.571
Tp (s) 2.48 3.71 3.16 3.82
U (m/s) 0.0468 0.0541 0.03107 -0.02382
V (m/s) -0.0713 0.126 -0.03096 0.06348
NTU 0.88 0.914 1.02 1.07

Tides Time Elevation (ft)
High 8:11 AM 2.3
Low 2:32 PM -0.2


Table 4.4 Wind data from April 26, 1991
Time Direction Speed (kts)
11:00 AM 160 12
12:00 PM 180 12
1:00 PM 140 17
2:00 PM 140 16
3:00 PM 140 8
4:00 PM 90 5
5:00 PM 140 5
6:00 PM 140 7


The turbidity lobes of April 26 propagate offshore with an average velocity of

0.033 m/s and 0.045 m/s for the north and south plumes, respectively. The magnitude of

the cross shore lobe speed is similar to the in-situ cross shore current speed of about 0.05

m/s. However, the direction of the measured current is onshore. Because the sensor is in

the lower portion of the water column, this suggests a two-dimensional cross shore

circulation pattern with onshore flow near the seabed and offshore flow near the surface.

The wave direction is from the northeast, which corresponds to the southward

propagation of the lobes. The measured longshore current velocity (0.126 m/s) is in

agreement in direction but about three times the speed of the longshore lobe propagation.

The tide is incoming throughout the propagation and the afternoon sea breeze

phenomenon is in effect, as evidenced by the wind data. The east coast of Florida










undergoes typical sea breeze conditions during the summer months. This activity is

characterized by a clockwise cycle of winds that begin light and peak in the mid

afternoon. This wind forcing causes changes in the wave and current conditions that are

superimposed on the normal tidal conditions.

Turbidity readings from the in-situ instruments show no evidence of the lobes.

There are two likely explanations for this. First, the measurements are made close to the

bottom of the water column, and the observed turbidity is on (or near) the surface.

Second, turbidity levels increase between noon and 4 PM, but the level is low (0.9 NTU)

and no reading is made around 5 PM when the lobes reach the area of the inner

instrument site. Determination of the vertical distribution of turbidity is not directly

possible with video images and requires a more intensive study.

The other good example of turbidity lobes occurred on May 24, 1991. This set of

3 lobes is perhaps the most dramatic event recorded with the VMS. The lobes propagate

between 11:30 AM and 1:30 PM. After a period of inactivity, a second wave of plumes

forms. Figure 4.25 presents the progression images, and Figure 4.26 shows the plotted

plume boundaries with the mean shoreline location and areas of dominant wave breaking.

As in the previous example, the first image has relatively low wave activity

accompanied by a uniform longshore turbidity band. A perturbation in the turbidity band

at 10:57 leads to the formation of lobes by 11:28. By noon, the lobes are well-defined,

quite symmetric, and spaced about 400 m longshore. Here, the waves appear larger and

the width of the surf zone is around 85 m. The areas of concentrated wave breaking

provide suspension in the surf zone. The lobe feeder zones occur between areas of

concentrated wave breaking, around x = 500 and x = 800 in Figure 4.26.

At 12:16 PM the two lobes have moved offshore and south, with a velocity

around 0.1 m/s. Their spacing is still about 400 m, as they are moving together. A third

plume comes into view from the North. By 12:48 PM, the original two lobes have




































































lobe progression,









67




S10:57 am 11:28 am
SD F


2 500- 500-

o C I
C 1000 1000
0 500 1000 0 500 1000



O 12:00 pm 12:16 pm


0
G H

M
? 500 500 -

W a % aqb o ~ D h 0 P, o
S1000 1000
0 500 1000 0 500 1000




12:48 pm 1:20 pm
0 0
o I K


e 500 500
0

a opcP4 0 0 ooob o 0^
S1000 1000
0 500 1000 0 500 1000



0 ,2:08 pm 0 3:28 pm
M O

S500- 500-


| 'b 0a o o coo*00 o f a oooog coo % a
S1000 1000




00
S500 1000 0 500 1000q,






1000 1000
C 1000 --------- ------- l, --- 1000 ------- ------------
0 500 1000 0 500 1000
Longshore coord. (m) Longshore coord. (m)


mean shoreline seaward turbidity limit wave breaking zone



Figure 4.26 Plotted turbidity lobes progression, 5/24/91











merged into one (between x = 500 and 1000 in Figure 4.26) and the north lobe has moved

into the left of the view. The offshore tips of the turbidity lobes reach to around 4.5 times

the width of the surf zone.

From 1:20 PM through 3:28 PM, the lobes disperse and the uniform turbidity

band reforms out to around 240 m offshore. This occurs at low tide, when the surf zone

width decreases to 75 m. The turbidity band has various perturbations, but no well-

defined "bumps."

After 3:28 PM the lobes reform in similar locations to the earlier ones. Their

separation is about 360 m at 3:44 PM. These lobes, though not as dramatic as the earlier

set, propagate in a similar fashion toward the southeast. Their velocity is somewhat

lower, slowing from about 0.08 m/s to 0.04 m/s.

Table 4.5 presents a summary of the information gathered from image data on

May 24, 1991. Quantities shown in each column are as defined previously, with the

addition of the column on the far right, W, which is the change in width of the lobe at y =

330 pixels. This arbitrary location is about 217 m from the mean shoreline. The

corresponding in-situ instrument and tide data are listed in Table 4.6. Miami

International Airport wind readings are presented in Table 4.7.



Table 4.5 Data summary of turbidity lobes, May 24, 1991
Images dt (s) dx (m) dy (m) dz (m) Vx (m/s) Vy (m/s) Vz (m/s) W (m)
North Plume
D-F 1920
F-G 1920 126 70 144 0.066 0.036 0.075
G-H 960 32 100 105 0.033 0.104 0.109 64
H-I 1920 28 272 273 0.015 0.142 0.142 100
O-Q 960 8 90 90 0.008 0.094 0.094
O-S 1920 -78 -22 81 -0.041 -0.011 0.042
Averages 0.033 0.077 0.092

South Plume
D-F 1920 38 48 61 0.02 0.025 0.032
F-G 1920 78 98 125 0.041 0.051 0.065 126
G-H 960 30 84 89 0.031 0.088 0.093 60
H-1 1920 -
O-Q 960 64 30 71 0.067 0.031 0.074
Q-S 1920 62 34 71 0.032 0.018 0.042
Averages _____0.038 0.043 0.061











Table 4.6 In-situ data summary, May 24, 1991
Site Inner (5 m depth)
Time 12:00 PM 4:00 PM
Hs (m) 0.3858 0.5534
_U 115.714 115.714
Tp (s) 3.41 3.707
U (m/s) 0.0574 0.0445
V (m/s) 0.1653 0.0402
NTU 11.24 11.08

Tides Time Elevation (ft)
High 6:54 AM 2.1
Low 1:20 PM -0.1


Table 4.7 Wind data from May 24, 1991
Time Direction Speed (kts)
10:00 AM 110 9
11:00 AM 110 15
12:00 PM 100 14
1:00 PM 100 8
2:00 PM 100 12
3:00 PM 110 15
4:00 PM 80 13
5:00 PM 110 13
6:00 PM 110 14


Table 4.6 only contains data from the inner instrument site. The outer site was not

operational at this time.

An overview of the data presented leads to an inexplicable forcing mechanism.

The waves approach from the southeast (1150), residuals of a large storm which occurred

between May 19 and 22 (see Figure 4.3). The in-situ cross shore current reading and the

offshore average lobe velocity agree as in the previous example, on an order-of-

magnitude basis. Despite the wave direction, the in-situ current direction and longshore

lobe velocities are both toward the south, which is counter-intuitive. The wind data is

also inconclusive, with the speed varying throughout the day and the predominant










direction at 1000. In this case, the tide is dropping during the initial lobe progression

(low tide is at 1:20 PM) and rising during the secondary lobe progression.

In each case presented, the dominant forcing mechanisms appear to be the waves

and nearshore currents. The increased nearshore sediment suspension, due to wave

action, saturates the surf zone and eventually escapes through breaks in the surf zone or

rip currents. The offshore and longshore driving forces appear to be related to the

advective currents measured in-situ, but specific information about the nearshore

circulation (particularly convection) is not known. The effect of the seabreeze on the

lobes is somewhat ambiguous, since the wind direction is predominantly opposite to the

water surface flow. Initial expectations were that the tidal influences would be more

obvious (incoming or outgoing). However, the tidal height during the peaks of both lobe

progressions was around 0.1 ft. above MLLW. For the characteristics of this beach,

perhaps a certain combination of tidal height and waves triggers the jetting of the

turbidity offshore. The importance of the diffusion part of the lobe propagation is

difficult to estimate with the nearshore concentration (initial concentration for diffusion

analysis) unknown. An intensive study including 3D mapping of the nearshore sea bed,

video monitoring, and vertical concentration measurements within the lobes would help

to determine the forcing mechanisms of these turbidity lobes.


4.4.1.2 Rip currents

The second category of natural turbidity structures is rip currents. Rip currents

are responsible for the transport of potentially large amounts of water and sediment to a

region outside the surf zone. They are of importance to the coastal scientist because they

greatly influence (or are influenced) by the local sea bottom topography, currents, and

wave conditions.

Although the so-called "chicken or the egg" dilemma often arises when rip currents

are discussed, there are conditions required for rip current formation. A barred bottom










with quasi-periodic spacing is generally accepted as a rip current scenario, where incident

waves break preferentially over the bars. The longshore variation of wave breaking

forces a circulation which is shoreward over the bars and offshore at the rip currents. The

rip currents are fed by water flowing alongshore, toward them, from the areas of wave

breaking (Fredsoe and Deigaard, 1992). This is the simplest scenario which is

complicated considerably with complex bottom topography and irregular waves. At this

level the question is not how the rips form, but whether or not they can be recorded with

video.

Evidence of rip currents may be assumed from the results of Section 4.4.1.1 (as

mechanisms for forcing turbidity lobes), but the highest quality images of rip currents

were obtained in Miami Beach on January 23, 1993. Figures 4.27 and 4.28 present a

progression of a rip current which formed between 11 AM and 12 PM on January 23. At

the center of the images, the breaking waves are seen to be from the southeast. Figure

4.29 is an oblique time average of the scene, taken at 12:06 PM. The crescentic bar

pattern is revealed with the time averaged image. The rip forms at the horns of the

crescentic bar segments, where the bottom contours permit a return flow.

In Figure 4.27 at 12:00, the rip current has the traditional shape with a narrow

channel (about 20 m width) inside the surf zone and a mushroom-shaped head which

extends to 115 m offshore. The spacing to the next rip to the north is about 355 m. The

waves breaking over the bar near the center of the image are 75 m offshore, and the

waves which flank the rip head are 85 m offshore.

In Figure 4.28 at 1:00 PM, the rip current is somewhat more diffuse. It extends to

155 m offshore and is characterized by a northward hook outside the surf zone. In both

images, the sediment feed zone for the rip current is apparent just to the north of the

shoreward end of the rip. The offshore hook to the north and the rip current skew to the

north are in agreement with the incoming wave direction.





































Figure 4.27 Rip current at Miami Beach, 12:00 PM














































The further study of rip current geometry and frequency is ideally suited to video

monitoring. By implementing a higher frequency sampling scheme, the formation of

such rips can be documented in great detail. Rips can be identified from time lapse video

records in areas where swimmer safety is important, as in Florida, where rips often take

unsuspecting tourists by surprise. The video technique may also provide the ability to

assess the local waves and nearshore features which dictate rip current characteristics.










4.4.2 Man-Induced Turbidity Structures

The second category of turbidity structures results from man-induced events. For

this study, we consider dredge and fill operations during beach nourishment as the

primary source of man-induced turbidity. Two direct results from the dredge and fill

method are nearshore turbidity snakes, caused by the slurry discharge on the beach, and

offshore turbidity snakes that result from hopper dredge overflow.

3.4.2.1 Discharge snakes

The beach nourishment procedure inherently produces turbidity. As shown in

Figure 4.4, the dredge discharge is placed at the head of a dike and trough system which

runs parallel to the shoreline. In theory, the dike system routes the discharge slurry along

the beach, where the sand gets deposited and the water percolates into the beach face and

runs into the nearshore region. Unfortunately, the lower quality sediments (finer matter)

remain suspended in the water and escape to the nearshore region. This is the cause of

what are hereafter referred to as discharge snakes. These discharge snakes have been

observed on the Hollywood Beach video data when the nourishment operations were near

the camera field of view.

The most dramatic example of a discharge snake occurred on June 12, 1991.

According to the dredger's log, the discharge point was located 30.5 m south of Broward

County Monument R-118. This point is about 335 m south of the rectified camera view.

Figure 4.30 presents a progression of a discharge snake from June 12. Figure 4.31 shows

the same sequence, plotted, including a longshore and cross shore scale for reference.

Also included are the mean shoreline location for June 12 and areas of dominant wave

breaking, as identified in the images. All digitization was done by a computer operator,

by first enhancing the image contrast and manually selecting the locations from the

images. This technique may be considered primitive, but the non-uniformity of the

turbidity patterns makes computer identification almost as subjective as human

identification.






































































Figure 4.30 Discharge snake progression, 6/12/91














_0 3:45 pm
B





o a 00 0
0 0 a 0 o

01
0 I I
0 200 400 600

Longshore Coord. (m)


0 4:17 pm

D






o o a o oco a
0 0 0


0 200 400 600





0 4:49 pm
F





o00 00oo VR
00


0 200 400 600




n 5:37 pm


0 200


500 L
0


4:01 pm

C





00 00
0 0o


200 400 600

Longshore Coord. (m)

4:33 pm


E






o on db o
5 00 0'' '

500
0 200 400 600


5:05 pm


G









0 200 400 600


mean shoreline location

- plume intensity boundary

o dominant wave breaking zone


400 600


Figure 4.31 Plotted dredge discharge snake progression, 6/12/91


I


00 0
0 -- ^ --- 6 ^


c rnn










Similar to Holman et al. (1990), a computer routine could be implemented to

select an appropriate threshold of intensity (determined by the user on a best fit basis) and

search each column of pixels for that location. This would automate the identification

process, but would still take operator time and supervision. For our introductory analysis,

the manual identification is appropriate, since we are looking for general, large-scale

motion of the discharge plumes.

The snake advects generally northward throughout the progression at 0.08 m/s.

The cross shore motion of the snake oscillates, but has no significant offshore component.

At 3:45 and 4:01 PM the visible turbidity lies outside the surf zone. The shoreward edge

of the snake encounters the surf zone at 4:17 (around x = 400, Figure 4.31) and moves

toward shore, presumably under the influence of the waves. The offshore reach of the

snake hooks southward throughout the progression.

Between 4:33 and 4:49 PM a small lobe nearshore advects northward. During the

final two frames, the northern reach of the snake cannot be determined from the data, but

the offshore extent remains fairly constant around 265 m. The southern reach of the

snake spreads to cover the entire surf zone and extend seaward. Table 4.8 presents a

summary of data from an analysis of the images.



Table 4.8 Data summary of discharge snake, June 12, 1991
North Edge of Plume
Images dt (s) dx (m) dy (m) dz (m) Vy (m/s) Vz (m/s)
B-C 960 3 -84 84 -0.088 0.088
C-D 960 -25 -76 80 -0.079 0.084
D-E 960 17 -83 85 -0.087 0.088
E-F 960 -13 -58 59 -0.06 0.062
F-G 960 -
G-I 1920 -
Averages -0.079 0.081

Nearshore Tip
E-F 960 37 -123 128 -0.128 0.134










In Table 4.8 dx specifies cross shore displacement (positive offshore), dy specifies

longshore displacement (positive south), and dz is the absolute displacement. The

average velocity in a'northerly direction is about 0.08 m/s. The mean offshore reach of

the visible turbidity is 262 m offshore and the maximum extent of the snake offshore is

297 m. In images G and I the turbidity snake advects north of the rectified view, so no

estimation of position is made.

To help understand the forcing of the snake, reference is made to some of the in-

situ instruments and weather data from June 12. Table 4.9 presents a summary of the

pertinent in-situ data.


Table 4.9 In-situ data summary, June 12, 1991
Site Inner (5 m depth) Outer (10 m depth)
Time 12:00 PM 4:00 PM 12:00 PM 4:00 PM
Hs (m) 0.4616 0.4231 0.3368 0.3214
") 102.86
Tp (s) 3.16 3.6 3.71 3.71
U (m/s) 0.0161 0.0391 0.0361 0.0473
V (m/s) -0.0252 0.0917 0.0167 0.0153
NTU 16.72 16.59 4.98 5.5

Tides Time Elevation (ft)
High 9:25 AM 2.4
Low 3:52 PM -0.7



Wind data from the Miami International Airport is presented in Table 4.10.


Table 4.10 Wind data from June 12, 1991
Time Direction Speed (kts)
11:00 AM 70 12
12:00 PM 80 14
1:00 PM 90 12
2:00 PM 110 10
3:00 PM 80 15
4:00 PM 80 10
5:00 PM 130 11
6:00 PM 100 10










The wave direction, although indeterminate at 4 PM, is from the southeast at

noon. This explains the general northward motion of the turbidity snake nearshore.

According to in-situ instrumentation, a significant event occurred on June 10 when

significant wave height from the northeast exceeded 1 m. This was verified by the fact

that dredging was postponed on June 10 due to 4 to 7 ft seas. The waves diminished

throughout the following days until June 13. The 4 PM in-situ longshore current velocity

is 0.09 m/s, which agrees with the southerly "hook" of the plume offshore extent. It is

noteworthy that the plume average velocity is north at 0.08 m/s, which may indicate a

tidal gyre motion in the area. The small lobe which occurs between 4:33 and 4:49 PM

moves with a longshore velocity of 0.128 m/s. This lobe is within the surf zone.

The turbidity values measured by the in-situ instruments may not be characteristic

of the snake. According to the image data, the snake does not reach the inner site. Again,

we cannot form any definite conclusions about the settling and vertical distribution of the

turbidity from the images. To determine the actual forcing mechanisms of turbidity

snakes, an intensive field experiment as described in Section 4.4.1.2 would have to be

performed.

4.4.2.2. Dredge overflow

Turbidity can also occur offshore, as a result of the overflow from a hopper

dredge operation. As sediment is loaded onto the barge, the water and fine particles are

allowed to overflow the barge. This process causes the larger, higher quality sediment to

remain in the slurry that is deposited on the beach. The consequence of this operation is

the resultant turbidity plume that propagates away from the borrow site by advection and

diffusion (Dompe, 1993).

Resultant dredge overflow plumes are clearly visible in the offshore region of the

Hollywood Beach video data. Unfortunately, the borrow areas were south of the camera

views. Overflow plumes are only visible when they propagate northward. Viewing a

video time lapse allows a general idea of the propagation time scale, but detailed










measurements are difficult due to the distance from the camera. Figure 4.32 is an

example of an image from Hollywood Beach which includes a dredge overflow turbidity

plume near the horizon line. This phenomenon has been studied in substantial detail, but

could be visually monitored with the VMS. Goodwin and Michaelis (1984) present a

comprehensive summary of turbidity plumes generated by a variety of dredge and

discharge configurations used in Tampa Bay, Florida.


figure 4.32 Uredge overflow turbidity plume















CHAPTER 5
CONCLUSIONS AND RECOMMENDATIONS


5.1 General Conclusions

The primary objective of this study was to evaluate the feasibility of utilizing

video monitoring for coastal engineering purposes, as an alternative and addition to

traditional measurement techniques. This was accomplished with a brief literature review

of the published work to date, an introduction to digital image concepts and analysis, and

some examples of the applications of video monitoring for coastal engineering.

In a qualitative sense, basic video monitoring has proven to be an excellent

method of observing long-term processes (like shoreline position and coastal feature

change). Shorter-term observations including sea conditions, weather, and beach use are

monitored with a modification of the sampling scheme for higher frequency.

Measurement techniques which result from analysis of video images are in need

of some refinement, and even full development in some cases. The measurement of

swash motions (Holman et al., 1990) and sandbar morphology (Lippmann and Holman,

1989) has been intensively studied, the latter using a time averaging technique. Time

averaging allows for submerged feature and water level determination. The present

experiments have shown that shoreline position and general feature length scales are

within 5% of surveyed data. Bar crest determination is generally skewed shoreward due

to intensity weighting by foam in the profile trough, but the overall bar width is

preserved. The technique fails during periods of low wave activity (for submerged

feature identification), but this is not regarded as a significant limitation.

The mapping of turbidity phenomena is of current interest due to a growing

concern for benthic communities near beach nourishment projects. Turbidity lobes,










discharge plumes, and rip currents are all events which may be responsible for transport

of potentially large amounts of water and sediments in the nearshore (and offshore)

zones. Some new findings during this study have shown the turbidity plumes at

Hollywood Beach to propagate well outside the surf zone (exceeding 3 times the surf

zone width) and form under relatively low wave conditions. The longshore separation

between plumes averages around 5 times the surf zone width. The forcing of such events

is positively linked to waves and currents nearshore, but detailed process descriptions

could not be made with the in-situ data recorded. Propagation velocities of the turbidity

plumes agree with in-situ current data collected on an order of magnitude basis. The

influence of tides and wind is inconclusive from this study. The video data has been used

to supplement in-situ instrumentation readings (Dompe, 1993). The complexity of

nearshore hydrodynamics requires a more intensive study than performed during this

experiment. Quantification of turbidity levels from video data was not addressed due to

the complexity of required image analysis, which extends beyond our current capabilities.

Pertaining to the data and techniques used in this report, general benefits and

drawbacks of the video monitoring presented can be summarized. Drawbacks can be

categorized as logistics or applications. Application drawbacks are threefold. First is

determination of the vantage point location, which depends on accuracy of ground

surveys. Second, there is a limited ability to perform vertical measurements like wave

height. Vertical measurements can be done with advanced photogrammetry but are not

part of current research. The third application drawback is the limit on horizontal

accuracy. The photogrammetry relates image locations to world distances. Errors in

measurements increase with distance from the camera, due to the nature of the technique

and resolution of typical digital images. Regardless of this condition, errors in image-

based measurements normalized by the distance from the camera are consistently less

than 5%. The resolution of the camera, frame grabber, and computer display each affect

the measurement capabilities. An optimal scenario contains high resolution equipment










located as close to the desired study area as possible. The desired measurement

resolution should be known first, then the appropriate equipment and setup selected to

perform any particular study. Logistical drawbacks are also threefold. First, the VMS

requires power and a phone line to transmit images. Second, protection against the

environment is essential to outdoor deployments. The final disadvantage is the inability

to record data at night, unless artificial light is introduced or a low-light camera is

employed.

The value of video monitoring has been shown. Cost, versatility, and flexibility

are all additional benefits. The cost of a basic Erdman VMS is currently around $15,000.

Each study requires individual assessment, but the logistics of deployment and

measurement cost far less than a similar study with aerial photographs, survey crew, or

field instrumentation packages. This fact alone makes the development of video

measurement techniques an attractive option. Versatility is inherent in the many

applications of the VMS, which are not limited to coastal engineering. The pinnacle

feature of the VMS is the large spatial scale which can be covered with a single image.

The flexibility of the VMS allows it to be adapted to any scenario. Here, the benefit is

the temporal variability. Any sampling frequency can be set to the needs of the

experiment. Images can be sampled every second, hour, day, as dictated by the

application. Tape records (burst samples) can be used instead of instantaneous images,

and various image processing functions can be employed as necessary. Other benefits

include ease of digitization for use and analysis using computers, as well as electronic

transmission of data over telephone lines.


5.2 Future Developments

Current research includes the Longboat Key, FL, beach nourishment, where

independent measures of turbidity and sedimentation rates are being correlated with

beach nourishment. Not only is spatial mapping of turbidity plumes of interest, but also a










method of quantifying turbidity from video data. Problems result when intensity values

are extracted from monochrome images, since lighting and camera aperture settings

greatly influence the images. To quantify turbidity levels, a standardized method of data

collection and calibration (including ambient light levels) and color image analysis

techniques would need development.

Determination of the forcing mechanisms of coastal turbidity requires a much

more intensive study than that performed in Hollywood Beach, 1991. A spatial array of

in-situ instrumentation for measuring waves, currents, and turbidity, as well as three-

dimensional seabed plots would be required. The tidal influence on turbidity lobe

formation needs further investigation. The idea of a certain tidal elevation, coupled with

adequate wave activity, saturating the surf zone with turbidity and then jetting it offshore

may be tested with a more intensive study. Also, a sampling of turbidity levels on an

event-by-event basis of observed plumes (in a vertical array) would increase an

understanding of the vertical distribution of turbidity plumes.

The development of new processing and analysis techniques may reveal improved

measurement methods. Use of high resolution cameras and frame grabbers would

increase measurement accuracy. Improvements to the rectification program would

include a better method of remapping intensity (with interpolation methods, for example,

instead of the nearest neighbor technique) and an option to include image information in

the black-out area caused by the limit of the camera field of view. Pertinent information

may include image size, name, resolution, location, and date. Time could also be saved

by coding the programs in C language and running them on a workstation with high

speed and large active memory capabilities. A serious effort to improve video

measurement techniques is required to convince the scientific community of the

measurement capabilities within the video realm. The increased number of unique

observation opportunities, alone, may help to increase our understanding of the coastal

environment.

















APPENDIX A:
LISTING OF EQUIPMENT AND VENDORS


Hardware:


Model


Manufacturer


Address/Phone


Data Loggers


Frame Grabber


Hi8 Camera



Image Capture Board


Personal Computer


PUV/Turbidity


DT3851-8


CCD-V99



Computer Eyes/RT


486DX/50E


UF COE


Data Translation


Sony


Digital Vision, Inc.


Gateway 2000


336 Weil Hall, UF
Gainesville, FL 32611
904-392-1436

100 Locke Dr.
Marlboro, MA 01752
508-481-3700

Sony Drive
Park Ridge, NJ 07656
201-930-7669

270 Bridge St.
Dedham, MA 02026
617-329-5400

610 Gateway Dr.
N. Sioux City, SD 57049
605-232-2000


Omni-1 Model 3006


AG-1960


Video Monitoring System VM-2


Alpha Electronics


Panasonic



Erdman Video Systems


Littleton, CO 80161
303-795-8435


50 Meadowland Pkwy.
Secaucus, NJ 07094
201-348-7620

2301 Collins Ave, #A336
Miami Beach, FL 33139
305-531-8511


Commercial Software:

Product Description


Autocad


Manufacturer

Autodesk, Inc.


Address/Phone


2320 Marinship Way
Sausalito, CA 94965
415-331-8093


Product


Total Station


VHS











DOS 5.0


Microsoft Corp.


Global Lab Image

Image Alchemy 1.7


MATLAB 4.0


Tidemaster


Windows 3.1

Word for Windows 2.0c


Data Translation

Handmade Software


The Math Works, Inc.


Zephyr Services


Microsoft Corp.

Microsoft Corp.


1 Microsoft Way
Redmond, WA 98052
800-426-9400

Marlboro, MA 01752

15951 Los Gatos Blvd.
Suite 17
Los Gatos, CA 95032
800-358-3588

24 Prime Park Way
Natick, MA 01760
508-653-1415

1900 Murray Ave.
Pittsburgh, PA 15217

Redmond, WA 98052

Redmond, WA 98052

















APPENDIX B:
PROGRAM LISTINGS


%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
% m file to solve transformation equations for oblique image rectification
% solve equations in terms of ground targets for focal length, tilt, phi
% transform and map pixels to new locations
% by Tim Mason rev. 7/19/93
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

% input original coordinates and swing angle (if known), transfer to
% principal plane coordinates

np=input('Enter # of control points ');
x0=zeros(2,1);
x0(l)=input('initial guess focal length [m] ');
x0(2)=input('initial guess camera tilt [radians]');
zc=input('camera elevation above MSL [m] ');
xf=input('x image control array in column form fiduciall pixels] ');
yf=input('y image control array ');
xgp=input('Cross shore (x) ground control array [m] ');
ygp=input('Long shore (y) ground control array ');
zg=input('z ground control array ');
p=input('initial guess for phi [degrees clockwise from Xg to E] ');

% input ground/camera coordinate offsets(origin of N-E in X-Y)
% also change in control.m if used

disp('Enter ground coordinate offsets relative to camera coordinates:')
xo=input('Xg offset:');
yo=input('Yg offset:');

%phi=control(p); % use if you want to solve for swing angle phi
phi=p*pi/180;

% convert image pixels to screen mm (1 pixel=.325 mm)
% alter as necessary for various monitors


theta=input('enter swing theta [degrees clockwise from vertical]:');
theta=theta*pi/180;
if theta-=0,
xf=.000325*(xf.*cos(theta)-yf.*sin(theta));
yf=.000325*(xf.*sin(theta)+yf.*cos(theta));
else
xf=xf*.000325;










yf=yf*.000325;
end

% convert xgp,ygp to correct coordinate system

disp('initializing...')
xg=xgp*cos(phi)+ygp*sin(phi)+xo;
yg=ygp*cos(phi)-xgp*sin(phi)+yo;

% now solve equations for focal length and tilt by least squares

detail=zeros(16,1);
detail(4,1)= 1;
c=zeros(6,np);
for i=l:np,
c(l,i)=xf(i);
c(2,i)=yf(i);
c(3,i)=xg(i);
c(4,i)=yg(i);
c(5,i)=zg(i);
c(6,i)=zc;
end

soln=leastsq('focal',x0,[],[],c,np);
phi
fc=soln(1)
t=soln(2)
disp('hit any key to continue...')
pause
clear xgp ygp xO zg c p np xf yfxg yg detail soln

a=input('enter image height ');
b=input('enter image width ');
res=input('enter desired output resolution [m/pixel] ');
save inp a b zc fc t phi xo yo res

% Initialize ground and image arrays

Ng=zeros(l,b);
Eg=zeros(a, );
Xg=zeros(a,b);
Yg=Xg;

disp('initializing arrays')
forj=l:b
Ng(lj)=(502-j)*(res); % res is rectified grid m/pixel
end
for i=l:a
Eg(i,l)=(460-i)*(res); % (a,b) =location of final grid origin
end % on rectified map (default = (0,0))










% transform Eg,Ng axes to Xg,Yg camera axes
disp('Transforming ground coordinates')

for i=l:a
forj=l:b
Xg(ij)=Eg(i,1)*cos(phi)+Ng(lj)*sin(phi)+xo;
Yg(ij)=Ng(l j)*cos(phi)-Eg(i, 1)*sin(phi)+yo;
end
end
clear Eg Ng res

% save camera/ground grid for use with other similar images
save inp2 Xg Yg

num=input('Enter number of images to rectify')

for s=l:num,
file(s,:)=input('filename (pad w/ spaces to 5 characters): ','s');
id(s)=input('# of significant characters in filename: ');
end

for s=l:num,

disp('loading image')
nam=file(s,l:id(s));
eval(['load' nam '.asc'])
gi=zeros(a,b);
gi=eval(nam);
gg=zeros(a,b); % initialize ground-rectified grid


% find (xi,yi) which correspond to (Xg,Yg) and map gi value to gg

disp('remapping values')
for i=1:a

forj=l:b
yq=fc*tan(atan(Yg(ij)/zc)-t);
xq=Xg(ij)*sqrt((yqA2+fcA2)/(zcA2+Yg(ij)^2));

%%%%%%% compensate back for swing angle

if theta 0,
xq=xq*cos(theta)+yq*sin(theta);
yq=yq*cos(theta)-xq*sin(theta);
end
% convert to pixels and round to nearest integer
yq=round(yq/3.25e-4);
xq=round(xq/3.25e-4);
m=(b/2)+xq;
k=((a/2)+ )-yq;
if(m>=l & m<=b)&(k>=l & k<=a),
gg(ij)=gi(k,m);




University of Florida Home Page
© 2004 - 2010 University of Florida George A. Smathers Libraries.
All rights reserved.

Acceptable Use, Copyright, and Disclaimer Statement
Last updated October 10, 2010 - - mvs