Analysis of conventional aerial photography to determine shoreline position

Material Information

Analysis of conventional aerial photography to determine shoreline position
Series Title:
Kreuzkamp, August Joseph, 1970-
University of Florida -- Coastal and Oceanographic Engineering Dept
Place of Publication:
Gainesville Fla
Coastal & Oceanographic Engineering Dept., University of Florida
Publication Date:
Physical Description:
x, 60 leaves : ill., maps ; 28 cm.


Subjects / Keywords:
Aerial photography in hydraulic engineering -- Florida -- Lee County ( lcsh )
Coastal engineering -- Florida -- Lee County ( lcsh )
Beach nourishment -- Florida -- Lee County ( lcsh )
Coastal zone management -- Florida -- Lee County ( lcsh )
government publication (state, provincial, terriorial, dependent) ( marcgt )
bibliography ( marcgt )
theses ( marcgt )
non-fiction ( marcgt )


Thesis (M.E.)--University of Florida, 1997.
Includes bibliographical references (leaves 58-59).
Statement of Responsibility:
by August Joseph Kreuzkamp III.

Record Information

Source Institution:
University of Florida
Holding Location:
University of Florida
Rights Management:
All applicable rights reserved by the source institution and holding location.
Resource Identifier:
37857027 ( OCLC )

Full Text

August Joseph Kreuzkamp III Thesis






I would like to express my gratitude to Dr. Robert G. Dean who acted as both my advisor and supervisory committee chairman. I would also like to thank the members of my committee, Dr. Ashish J. Mehta and Dr. D. Max Sheppard for helping me follow through with my research.
I have really enjoyed my past two years at the University of Florida with much due credit to the staff and students of the Coastal and Oceanographic Engineering Department. The department has been much like a family in the classroom, outside at the beach doing fieldwork and at the stadium cheering on the national champion Florida Gators. My brothers and sisters include Marshall, Paul, Wally, Jie, Rob, Cindy, Pete, Mike, Bill, Mark, Hugo, Eric, Chris, Kerry Anne, Greg, Al, Craig, Helen, Becky and Terry. I wish all of them well in there future endeavors and hope at least some of them finish their graduate work before my children go to college.
Friends and family have always been important to me and are to whom I attribute my success. The boys who grew up down in the Knight's basement including Ed, Ullman, Trev, Bri, Rich and Nick are the ones responsible for tearing my knee, giving me a concussion and being my oldest and dearest friends. My college sailing friends including Eric M., Barr, Wilson, Erick S., Chad, Paul, John and Heather are the ones who showed me that nothing is as important as sailing, friendship and coubelibres. My life-long friend,

Marc Fermanian, who has shown me that the right answer is all that really counts, in every aspect of life. My grandmother, who has been a true inspiration to me over the past many years, has proved that perseverance and internal strength can overcome any obstacle. My parents, who have always been there for me, deserve more praise and accolades than I could ever give. They are truly the most giving people I have ever known.
The final thank you goes to my wife, Kristen, who will always be my best friend and soul mate. I did this for you.

ACKNOWLEDGEMENTS ................................................................ ii
LIST OF TABLES ........................................................................... vi
LIST OF FIGURES ......................................................................... vii
1 INTRODUCTION .................................................................... 1
TO DECIPHER THE SHORELINE POSITION .................................. 5
Introduction ........................................................................... 5
Factors Limiting the Accuracy of Aerial Photography ........................... 6
Point Measurements .................................................................. 9
Photo Enlargements (Orthogonal Grid Matrix System) ......................... 10
Zoom Transfer Scope ................................................................ 10
Analytical Stereoplotters ............................................................ 11
Metric Mapping ...................................................................... 13
Airborne Laser Swath Mapping ..................................................... 13
3 METHODOLOGY .................................................................... 15
G round T ruth ......................................................................... 15
Aerial Photographs .................................................................... 19
Shoreline Location .................................................................... 22
Digitizing Photographs ............................................................... 26
4 ANALYSIS OF DATA .............................................................. 31
C alculations ........................................................................... 31
Method of Analysis ................................................................... 33

5 RESULTS AND DISCUSSION .................................................... 38
Shoreline Position Error ............................................................. 38
Shoreline Change Error ............................................................. 44
Sources of Error ...................................................................... 48
Error Analysis ........................................................................ 51
Improvements ........................................................................ 52
ST U D Y : ................................................................................ 55
Sum m ary ............................................................................. 55
Recommendations for Further Study ............................................. 57
REFERENCES ............................................................................. 58
BIOGRAPHICAL SKETCH ............................................................ 60


Table F=
3-1: Dates ground and aerial surveys were performed ............................. 19
3-2: Flight and equipment information ................................................ 20
4-1: Sample data set for Routine #1 .................................................. 35
4-2: Sample data set for Routine #2 .................................................. 36
4-3: Sample data set for Routine #3 .................................................. 37
5-1: Summary error results for each aerial survey by island ....................... 40
5-2: Potential errors due to camera tilt ............................................... 50


Figure Pae
1-1: Vicinity map of Lee County, Florida.................................5
2-1: Comparison of orthogonal vs. tilted Images........................... 7
2-2: Geometry of a tilted photograph showing the principal line.............. 8
2-3: Tilt displacement in the principal plane of a tilted photograph ........... 9
2-4: Correcting for tilt distortion........................................... 11
3-1: Profile locations on Captiva Island ................................... 17
3-2: Sample profiles taken at R-088........................................ 18
3-3: Tidal prediction near Estero Island, Florida 2/04/96 2/06/96 .......... 21
3-4: Tidal prediction near Estero Island, Florida 8/25/96..................... 22
3-5a: Aerial photograph taken August 25, 1996 at the south end of Gasparilla Is. with evidence of runup ...........................................25
3-5b: Aerial photograph taken August 25, 1996 at the north end of Lacosta Is.
with no sign of runup................................................ 25
3-6: AutoCAD display window showing a portion of the base map of Captiva Island ...................................................... 28
3-7: AutoCAD display window showing the composite map of Captiva, Island being assembled .............................................. 29
4-1: Normal distribution curve............................................ 33
4-2: Sample histogram....................................................34
5-1: Comparison of uncorrected data sets................................. 39

5-2: Comparison of uncorrected data sets excluding problem areas .............. 41
5-3: Comparison of results determined by applying various computations
to all February data (including problem areas) .................................. 42
5-4: Comparison of results determined by applying various computations
to all August data (including problem areas) ................................... 42
5-5: Comparison of accuracy obtained between the two aerial surveys ......... 43 5-6: Comparison of shoreline changes for the entire county .......................
5-7: Comparison of shoreline changes for the entire county excluding
nourished areas ..................................................................... 45
5-8: Comparison of shoreline changes at profiled locations excluding
nourished areas .................................................................... 46
5-9: Comparison of shoreline changes at profiled locations within nourished
areas ................................................................................ 47
5-10: Comparison of shoreline changes determined from sequential ground
surveys .............................................................................. 5 1

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
August Joseph Kreuzkamp III
August 1997
Chairman: Dr. Robert G. Dean
Major Department: Coastal and Oceanographic Engineering
The outer coastline of Lee County, Florida, consists of a series of eight barrier islands all of which experience episodic erosion due to storms and portions of which are subject to long term erosional trends. With such a dynamic and valuable shoreline, the county has participated in actions to maintain portions of the shoreline by nourishment. In the process of maintaining the coastline, monitoring the shoreline position would be quite expensive and time consuming by ground survey alone.
In an attempt to devise an efficient and relatively inexpensive monitoring method, a technique was developed based on aerial photography. This technique incorporated digitizing current aerial photography and referencing State of Florida Coastal Construction Control Line photos to create a composite map providing a continuous shoreline for each aerial survey performed. From this composite map, the change in shoreline position was obtained with reference to the coordinates of each DNR monument located throughout

the county. Ground truth surveys were performed as soon before and after the aerial photographs as possible and used later to adjust the data and evaluate the method.
Throughout the process of refining this method, several developments were made which were found to produce more reliable data. The improvements included specifying larger scale aerial photographs, setting visual targets over known coordinates and coordinating the aerial survey with high tide. The best accuracy for this method was found to be approximately 11 feet for specific areas. The results for this six month study indicate that the average shoreline had accreted approximately 9 feet ( 3 feet) due primarily to nourishment projects that occurred in the interim of this study. After the nourished areas were excluded from the sampling, the average natural shoreline was found to have eroded approximately 7 feet (3 feet). In considering these results, it is necessary to note that the period over which these results were obtained reflects the effects of individual storms rather than an erosional trend.
The best application for this method would be to evaluate overall annual shoreline changes, monitor the performance of beach nourishment projects and evaluate coastal damage due to large storms. The effectiveness of this program and the improved reliability of its results would require the refinements already developed to be continued and further progress to be made to minimize the remaining distortion in the photographs which could not be avoided with the equipment used.

The nearshore area is very dynamic due to the large amounts of wave and current energy that must be dissipated in a relatively small region. Relevant factors affecting nearshore dynamics include sand grain size, incident wave direction and height, nearshore bathymetry, and impacts of inlets and man-made structures. Particularly in regions with unstable shorelines, it is important to document accurately the position of the shoreline at strategic time intervals in order to identify overall trends in shoreline change and anomalies caused by significant storms. Ground survey work has always been found to be the most reliable in monitoring shoreline position while also being the most time consuming and costly (Dolan et at., 1978). This has forced groups including the project sponsor (Lee County, Florida) who work with a limited budget to consider other techniques including aerial photography to track migrating shorelines.
Aerial photography was first utilized to determine shoreline change in coastal regions in the late 1960s (Stafford and Langfelder, 1971). Aerial photography provides the needed coverage and detail required for long shoreline segments at a fraction of the time and expense associated with maps created from ground survey work. Aerial surveys can be employed in several applications including developing sediment budgets, monitoring beach nourishment projects, and observing the effects of coastal structures or

inlets on an immediate area. Unfortunately, the location of an object found in an aerial photographs is somewhat misrepresented due to various distortions that exist in aerial photography. Through the years technology has played a major role in improving the accuracy of these shoreline positions by utilizing various equipment and techniques in an attempt to minimize the amount of error associated with aerial photographs.
Developments in lens manufacturing has greatly reduced photographic distortions as rectification equipment including zoom transfer scopes, space resection programs and analytical stereoplotters. Each piece of equipment and/or software has had varied levels of success in correcting the distortion in photographs caused by plane pitch and roll, change in altitude, relief displacement and radial scale variations. The most effective rectification equipment, analytical stereoplotters, can correct near-vertical or even terrestrial (horizontal) photographs; however, the relatively high costs may be beyond the means of various entities with shoreline monitoring responsibilities.
The project area lies in Lee County, in Southwest Florida, and consists of approximately 40 miles of active coastline on the Gulf of Mexico. This area is comprised of eight barrier islands which require periodic monitoring in order to document shoreline changes associated with past and present nourishment projects as well as natural shoreline evolution. The method of obtaining the shoreline positions for this region involved digitizing features from uncorrected aerial photographs including the identified shoreline based on local indicators seen in the photographs. The digitized images were next combined into a composite map from which the shoreline position could be recalled from any chosen reference. The acquired shoreline positions to better conform with the shoreline positions was then adjusted determined from the ground surveys. Due to the

areal extent of the project area and the amount of data collected, Captiva Island was chosen as the region of focus in the main body of this paper to show meaningful examples with a reasonable amount of graphics and tables.


o ~ 1~




0 4 8 MILES
po!o o !m


0 160 320 MILES

FIGURE 1-1 Lee County,




2.1 Introduction
Aerial photography first originated in the mid-1800s using cameras mounted on balloons and kites. With the invention of the airplane in 1902, it was not until 1913 that the airplane was first used for aerial photography for mapping purposes. The use of aerial photography was expanded during World War I (19 14-1918) when its application for reconnaissance was recognized and later during World War 11 (1939-1945) when reasonably high quality stereo vertical (aerial) photographs were developed. Since that time, a multitude of technological advancements in equipment and technique have been made to advance the accuracy of aerial photography (Wolf, 1983).
While aerial photographs contain large quantities of information, they also include a variety of distortions, which leads to their inaccuracy if left uncorrected. The distortions include those caused by: changes in the camera's altitude, changes in tilt and pitch of the aircraft, radial scale variations, and relief variations of the surface photographed. Researchers attempting to determine historic erosion rates found that vertical photographs dating as far back as the 1930's were a more reliable source for shoreline position when compared with shoreline maps established during the same era (Crowell et at., 1991). These researchers developed different techniques to extract the shoreline position from

vertical photographs with varying levels of success in terms of their accuracy. Ironically, only a few authors actually discuss the errors associated with their work which complicates matters when attempting to compare the accuracy of alternate methods (Dolan et al., 197 8; Anders and Byrnes, 199 1).
2.2 Factors Limiting the Accuracy of Aerial Photography
There are several factors which limit the precision of locating points which are photographed from a skewed or moving platform. The following are factors which cause changes in scale between sequential photographs or within the same photograph.
Changes in camera altitude will cause an aerial photograph to have a unique scale compared to others taken previously or subsequently after any photograph. This is very common in smaller aircraft which are more susceptible to wind gusts. The scale of a photograph, S is calculated by dividing the focal length of a camera, f by the altitude flown, H.
S f (2-1)
Tilt of an aircraft will create a scale variation within each photograph that is not easy to correct due to its non-linear properties shown in Figure 2- 1. Two important terms which must be defined are the principal line and isocenter of a photograph. Referring to Figure 2-2, vertical plane Lno is defined as the "principal plane". The intersection of the principal plane and the plane of the tilted photograph is the principal line labeled no. The isocenter of a tilted photograph, I, is located at the bisector of where the plane of the tilted




FIGURE 2-1 Comparison of orthogonal vs. tilted images (Leatherman, 1983)
photo intersects the plane of the theoretical vertical photo shown in Figure 2-3. Based on the geometry of Figure 2-3, Equation 2-2 represents the displacement error, dt, caused by airplane tilt, t.


dt = r sin tcos2 P
f rsin tcos P

Defining the radial distance from the isocenter of the photograph to the point of interest represented as r, camera focal length (f) tilt angle (t) and the angle measured from the principal line to the radial line between the isocenter and the point of interest (P). Once dt is calculated, the scale of the photograph is taken into account to convert the error to ground dimensions (Wolf, 1983).
Objects away from the isocenter of a frame will cause an increase in radial scale variations. Refinements in lens manufacturing have helped to minimize this effect (Clow and Leatherman, 1984).

FIGURE 2-2 Geometry of a tilted photograph showing the principal line (Wolf, 1983)
Different elevations in the surface photographed will cause a relief displacement. In essence, objects rising above the average land elevation will be displaced outward from the photo' s isocenter while objects below the average grade will be displaced inward of the photo's isocenter. Due to the fact that most of the coastal features in the project area had low relief, radial distortion due to elevation differences is considered minimal.


FIGURE 2-3 Tilt displacement in the principal plane of a tilted photograph (Wolf, 1983)
2.3 Point Measurements
Utilizing historical aerial photographs, Stafford and Langfelder (197 1) developed a method to determine the historical coastal erosion along the Outer Banks of North Carolina. Shoreline positions were found by taking point measurements perpendicular to the shore from stable reference points. Distortion due to plane tilt was minimized by utilizing only the central portion of each photograph and by applying a unique scale to each photograph. Stafford and Langfelder were also able to minimize the inherent change in scale between photographs caused by altitude changes by grouping beach widths taken


from the same photographs and then, after determining the scale for each photograph, the beach widths could be modified. The final product became a data base of beach widths which were compared with beach width data taken from other surveys in the same area.
2.4 Photo Enlargements (Orthogonal Grid Matrix System)
Based on the work of Stafford and Langfelder, Dolan et al. (1978) were able to document and compare continuous shoreline positions on drafted maps to determine rates of erosion. Their method, called the orthogonal grid matrix system (OGMS), involved using a projecting light table to superimpose photo images to the approximate scale of a base map and manually traces the shoreline onto a USGS 'T-Sheet" (topographic map). Changing the scale of the projected images once again helped eliminate errors associated with the change in camera altitude. Once all of the shorelines were traced onto one common base map, beach widths were measured at 100 meter intervals. Maps were then recreated from this data but could not reflect the same completeness as the original base map. Although a portion of distortion was eliminated by maintaining a constant scale throughout entire sets of photos with this method, no attempt was made to eliminate the distortion due to plane tilt. The documented accuracy of the OGMS method was 6.3 meters (Dolan et al, 1978).
2.5 Zoom Transfer Scop~e
An important piece of equipment used to further reduce the amount of error in vertical photographs is the zoom transfer scope. The zoom transfer scope (ZTS) is very efficient in eliminating scale differences between sequential photographs by using a base

map for control The ZTS has the ability to stretch or shrink images in one direction about the axis of tilt to reduce a portion of the distortion caused by camera tilt as demonstrated in Figure 2-4. The reason why only a portion is corrected is due to the fact that the ZTS is a linear adjustment device while distortion due to tilt is nonlinear as mentioned earlier. Since zoom transfer scopes can not make the necessary scale corrections needed to eliminate all tilt distortion, it is necessary to utilize one side of an image at a time and superimpose the corrected points of interest onto a base map. This process is considered to produce reasonable results but found to be time-consuming and tedious and should be reserved for smaller-sized projects (Leatherman, 1983).

FIGURE 2-4 Correcting for tilt distortion (Leatherman, 1983)


2.6 Analytical Stereoplotters
An analytical stereoplotter is a very precise device which places an aerial photograph back into its tilted position and projects a rectified image, or corrected image, which is used to produce an orthophotograph. An orthophotograph is a corrected aerial photograph and has the advantages of having pictorial qualities which have planimetric correctness. In other words, information including the dimensions of images may be extracted from such a photo as if referencing a map.
The procedure to create an orthophotograph is based on determining three orientations, in particular, interior, relative and absolute orientations. Interior orientation involves determining the photographic coordinate system with respect to the instrument's image coordinate measurement system for each diapositive (a transparent positive) by applying a least squares procedure to compute a coordinate transformation. The relative orientation involves selecting six or more common points between two photographs and again applying a least squares procedure to determine the orientation parameters needed to solve for the relative coordinates for any point in the 3-D model. Finally, the absolute orientation is determined by applying a least squares procedure to two horizontal and three vertical control points in order to determine a mathematical relationship between model and ground coordinates. The final product of this process can be in the form of orthophotographs, corrected maps or corrected digital images. An experienced operator can be expected to accurately locate the shoreline position to within 5 feet (Fisher and Overton, 1994). Although accuracy is relatively high, many factors including high cost, size of equipment and operator time may outweigh the benefits of this technology.

2.7 Metric Mapping
Instead of utilizing expensive technical equipment, metric mapping employs the use of more conventional computer equipment including a personal computer with moderate storage and processing capacity and a digital scanner or digitizer. The technique first involves scanning or digitizing an aerial photograph to convert each photo into digital format. Each image is then rectified by applying the least squares method within a written Fortran program. The image produced is in its best fit position whose accuracy is solely based on the number of control points included. The methodology and theory being quite similar to that of the stereoplotter without being as expensive and time consuming (Leatherman, 1983).
2.8 Airborne Laser Swath Mapping
Recently, a new system has been developed which is able to obtain the shoreline position as well as the topography of the nearby coastline. The process uses a solid state laser ranging system mounted on the bottom of an airplane which simultaneously references a differential GPS navigation system. During the time of operation, the laser ranging system time-tags (with GPS time) each laser pulse in order to measure the relative distance to a ground point below the aircraft based on the amount of time elapsed between the time the pulse is sent and the time the return signal is received. After the flight, the relative distances are compiled with the flight GPS coordinates to determine the ground surface position and elevation for every point. Presently, the manufacturer is reporting the newest equipment is able to emit 25,000 pulses per second. At this rate, it would be possible to document the topography for hundreds of miles of coastline in only a few

hours. The accuracy for this revolutionary system is in the process of being tested at this time; however, preliminary findings indicate a vertical error of 10 centimeters. Once this system has been fully-developed, the amount of error in the shoreline position would be completely due to the measuring technique alone since all of the environmental factors that plague the accuracy of the shoreline position in photographs would be eliminated. Since this procedure can only be used for present and future aerial surveys, aerial photography will still be the best source for the most accurate shoreline positions of the past. Thus, even in the future, the procedures and techniques mentioned earlier will remain valuable.

The method employed in this study involved analyzing the calibrated shoreline positions determined from aerial photography and can be described as several phases. The details of each phase with their associated procedures are included below to ensure that all of the necessary steps can be duplicated or improved upon in future studies. The particular phases to be discussed in this chapter include: ground truth, aerial photography, determining shoreline locations, and digitizing photographs.
3.1 Ground Truth
In order to be able to check the geometric accuracy of aerial photography, the data obtained from aerial photography required comparison with other reliable data that could only be obtained from ground surveys. From the ground surveys, beach widths could be determined and compared with the beach widths measured from aerial photography. Beach widths were defined as the distance from a known position to a defined shoreline elevation.
The ground truth surveys were performed as soon as practical before and after the aerial photographs were taken. The availability of two ground surveys for each set of photographs had two advantages. One benefit for having several sets of survey data was

stonn altered topography could be accounted for immediately before or after the photography was taken. With ground truth data available both before and after the time of the aerial survey, it was possible to select the ground truth data which best represented the shoreline at the time of the aerial survey or to interpolate or average results from the two surveys. The second benefit for having a series of ground surveys gave the ability to compare data between ground surveys since each set of ground surveys was within a fairly short time period. By cross-referencing the data, irregularities due to human error in field measurements would be evident.
Each ground survey consisted of a series of 46 profiles along Lee County's coastline on the Gulf of Mexico at approximate one mile intervals. A three or four person crew conducted the surveys with areas accessed by either boat or car depending on the particular island. The equipment used to survey the profiles at the selected monuments included that for standard ground surveys: level, level rod and tape. Each profile incorporated one of the 239 Department of Natural Resources monuments which exist along the approximately 40 mile length of shoreline within Lee County. The elevation and location of each monument reference the National Geodetic Vertical Datum (NGVD) and state plane coordinate system respectively. This information is provided by the Florida Department of Natural Resources Bureau of Beaches and Coastal Systems referred to as Monument Descriptions. Locations of the profiles taken on Captiva Island are shown in Figure 3-1.
Beach widths were then obtained from the ground survey data by plotting crosssections for each of the 46 profiles and measuring the distance from the known monument

0 5000 10.o000 FT.



FIGURE 3-1 Profile locations on Captiva Island
location to the elevation of the Mean High Water Datum (MHWD). The MHWD relies upon statistical data based on the average height of the High Water elevation over a nineteen year period (Shalowitz, 1964). This datum, which changes slightly along a long shoreline, is documented in "Transformation of Historical Shorelines to Current NGVD Position For the Florida Lower Gulf Coast" by Balsillie, Carlen and Watters (June 1987). The range in elevation of the MHWD varies along the Lee County shoreline from +1.17 feet NGVD to +1.43 feet NGVD with the average value equal to +1.28 feet NGVD. With

the elevation of the MHWD known at each monument throughout the county (based on Balsillie et al) and having the profile information from the surveys, the beach width could be determined as shown in Figure 3-2 below. The beach widths were then later used to calibrate and evaluate the amount of error in the computer composite map by comparing beach widths obtained from the composite map with these ground truth beach widths.



O io 20 Fr.
O 40 so FT.

- -- SUNWY 04 (/27/96)


I '
FIGURE 3-2 Sample profiles taken at R-088
Dates for the ground and aerial surveys are shown in Table 3-1. Survey number 2 was chosen to represent the ground truth for aerial survey #1 due to its timing being much closer to the date of the photography than survey #1. Resulting beach widths for survey

1 .. I

numbers 3 and 4 were averaged to approximate the most accurate beach width at the time aerial survey number 2 was performed. Certain exceptions were made in cases where the survey crew was denied access or previously chosen monuments were destroyed or altered by construction or erosion. Refer to Figure 3-2 for an example of how the profiles were documented. Notice in the figure that the vertical scale is exaggerated compared to the horizontal scale to emphasize the profile relief.
TABLE 3-1 Dates ground and aerial surveys were performed
Survey #1 11/18/95 11/19/95 & 12/02/95 12/03/95
Aerial Survey #1 2/05/96 (10:54am 1 l:36am)
Survey #2 2/10/96 2/12/96
Survey #3 8/16/96 8/18/96
Aerial Survey #2 8/25/96 (9:49am 10:36am)
Survey #4 8/27/96 8/28/96
3.2 Aerial Photographs
This study included collecting two sets of aerial photographs of the Gulf of Mexico coastline of Lee County on February 5, 1996, and August 25, 1996. The specifications of the equipment and information for the two flights are presented in Table 3-1. Under "Camera Equipment" in Table 3-1, the "15" signifies the focal length of the lens (actually 152.28 mm) and the "23" signifies the width of the negative (actually 230.0mm). The filter mentioned, the Minus B filter, was used while taking the photographs to reduce the amount of blue light through the lens. This light causes a haze to develop on the film reducing the contrast required to see detail.

TABLE 3-2 Flight and equipment information
7 Contractor Kucera South Kucera South
Plane Cessna 206 Cessna 206
Altitude Flown 3,000 feet 1,500 feet
Camera Equipment Ziess RMKA 15/23 Ziess RMKA 15/23
Filters Minus B, yellow filter Minus B, yellow filter
Film Type Kodak Type 2405 B/W Kodak Type 2405 B/W
Approximate Scale 1"=500' 1"=250'
Width of Negative 230.0 mm 9 inches 230.0 mm 9 inches
Time of Flight 10:54am 11:36am 9:49am 10:36am
The six month period between photographs was chosen to provide the greatest amount of beach width contrast in a limited time frame. This variation in beach width is due to the variation of the beach profile caused by the change in seasonal climate. In a six month time period, a change in the cross-shore profile is caused by different types of wave activity between the winter and summer months. In the winter months, storm systems tend to be more intense and frequent which results in greater and more frequent wave activity compared to that in the summer. The general result of this increased wave activity is offshore sand transport which tends to form what is commonly referred to as a winter bar.
Scheduling the time and day for the aerial surveys was initially considered to be inconsequential. With no special instructions provided to the sub-contractor for the first aerial survey as to the phase of the tidal cycle during which the project area was to be photographed, the aerial survey was performed near noon-time. This time is typically preferred by aerial photographers during the winter months to avoid unnecessary shadows. The semidiurnal tide effect in the vicinity of Estero Island during the first aerial survey is

provided in Figure 3-3 (taken from the program Xtide (Version 1.5beta)). From this figure it can be seen that the significant high tide before the first aerial survey occurred at 12:14 am, approximately 11 hours before the time the region was photographed. At the time the region was photographed, the approximate tidal elevation for the area was 1.7 feet Mean Low Low Water Datum (MLLWD).
Pt~anzft 6" st Itl and Forld6
02-,04 02-04 02-05 02-05 02-05 02-05 02-06
134*5 18.-41 00~4 7:0~ 13;31 194* 7 0:49
4 Ft
High tide priac to agiieI sume --0ime of first aeriml survey
FIGURE 3-3 Tidal prediction near Estero Island, Florida 2/04/96 2/06/96
More planning and coordination with the aerial photographer preceded the second aerial survey was involved. Based on the detail shown in the first set of aerial photographs (1"=500'), a larger scale was specified for the second set of aerial photographs (1"=250'). For clarification, small scale maps or photos tend to show larger portions of the earth's surface while large scale maps or photos approach the actual size of the entity being documented. For scheduling purposes, it was learned that it is necessary to acquire aerial photographs in the early morning during the summer in South Florida to avoid the daily cloud cover. This cloud cover which obstructs the detail in the photos typically appears around mid-morning and continues throughout the rest of the day. Since the success of an

aerial survey is also weather dependent, the subcontractor was provided with a tide table including high tides for the month and requested to start taking photographs within a time frame of an hour (approximately) after the estimated time for high tide. The second set of aerial photographs was taken on August 25, 1996 between 9:49am and 10:36am with the high water tide at 9:08am at a tidal elevation of 2.9 feet MLLWD (Xtide ). Refer to Figure 3-4 for details.
FIGUR 3-4 idal reditnz neas Estero Island, Flori8/59
3.32 Shorlin Locatio
change,~ ~ ~ ~~~~ ~Tm seimn supply, cosaaeloyadm rpholgryadhmnitreio
procedure ~ ~ ~ 3. waSeiin h horeline osi tion o h htgah.We eiwn
each set of photographs taken from the two aerial surveys, it was important to choose an indicator which was identifiable throughout the entire 40 miles of shoreline for each set of

photographs while attempting to maintain a consistent elevation. Regardless of the indicator's elevation, the measurements taken from the digitized image models would later be corrected using the ground truth surveys to adjust the results.
Three choices of indicators typically exist on the beach profile which can be used in determining the position and migration of the shoreline. The indicators are the dune line (which is non-existent in South Florida), the High Water Line and the Water Line. Researchers base their decision for which indicator to use on the local climate variables at the site of interest. Both the HWL and WL can be found at the place of tonal (color) change between wet and dry sand based on the amount of moisture content found in the sand. McCurdy (1947) identified the High Water Line (HWL) as a perceptible feature of the subaerial beach face which represents the highest extent of the most significant high tide over the past day which would also include wave run-up and setup. The location of the Water Line (WL) is where there is a contrast in tonal color in the sand dependent on the existing wave runup and tidal elevation at the time of photography.
There are a few factors which complicate the ability of the HWL to represent a consistent elevation along a lengthy shoreline. Variable wave runup and setup caused by the presence of nearshore bars and shoals interfere with the amount of wave energy that reaches the shore which in turn displaces the wet sand line above the elevation of the still water line. The short life span of the HWL is the other factor. The identification of this line is dependent on the percolation rates of the sand material on the beach face combined with intense sun exposure will affect the amount of time the sand in an area will remain wet (dark). The only advantage to using the WL over the HWL, is the WL can be identified throughout an entire day without having any concern of the indicator fading

since it is always actively part of the swash zone at the swash terminus. Should the WL be used as the indicator, it is important that the shoreline position be converted to a known elevation datum.
In choosing the shoreline identifier, all the previously mentioned factors were taken into consideration. For the first aerial survey, too much time had elapsed between the time of the daily significant high tide and the time the photographs were taken to be able to clearly see the High Water Line at the scale of the photographs. The only reasonable choice was to choose the Water Line. For the second set of aerial photographs, the HWL was easily identified with the larger scale photographs however a large variation in wave mun-up along the shoreline was noticed. Since the location of the High Water Line is dependent on tide, wave setup and wave run-up; near-shore bathymetry will affect the amount of wave energy that reaches the shore. The presence of ebb tidal shoals, for example, can cause discrepancies in the High Water Line elevation as shown in Figures 3-5a and 3-5b taken at the south end of Gasparilla Island and the north end of Lacosta Island respectively. With Boca Grande Pass separating the two photographs, just over a mile in width, it can be seen how such changes in bathymnetry can affect the wave energy that reaches the shore. Based on this finding, the base of the swash zone, was found to be well defined throughout and chosen as the indicator for the second set of photographs in an attempt to maintain a constant elevation throughout the 40 miles of shoreline. Once the shoreline positions were recorded and compared, it would be possible to calculate the amount of shoreline advancement or recession.

Figure 3-5a Aerial photograph taken August 25, 1996 at the south end of Gasparilla Is.
with evidence of runup

Figure 3-5b Aerial photograph taken August 25, 1996 at the north end of Lacosta Is. with
no sign of runup

3.4 Digitizing Photographs
Once the photographs were taken and the ground survey data were recorded and processed, the next step was to determine the best method of digitizing the aerial photographs taking into account time required and best results. Equipment used throughout all different approaches included using a Cal Comp Digitizer tablet which was connected to a 486 DX computer with AutoCAD Release 12 software as the user interface. The paragraphs below describe the sequential development procedure for photograph digitization.
The first technique was to align a series of corresponding photographs and then digitize each set of photographs. Much attention was paid to common objects on sequential photographs which aided in both the manual assembling of five photographs and later the assembling of those sets of five with other sets forming entire islands using AutoCAD. For control points, it was desired to use common objects, that had minimal relief variations (changes in elevation which cause distortion of the object). Parking lots, pools, tennis courts and roads were found to provide the best references. This worked well in developed areas however in undeveloped areas some difficult choices had to be made by selecting items including large debris on the beach or shadows of large structures near the beach. Digitizing each set of photographs involved calibrating the digitizer tablet in order to preserve the scale of the photographs and then recording the location of reference points and the identified shoreline on each set of photographs to form a computer drawing image. The sets of images were then joined together using the drawing program with every attempt made to obtain the best fit of all the control points included in the images. Once an entire island was assembled, visual targets that were set over DNR

monuments were used to transform the islands' shorelines from an arbitrary coordinate system to the State plane coordinate system. After the method was applied throughout a substantial amount of shoreline, significant errors became apparent. It seemed likely that joining the photos together by hand into the sets was responsible for a substantial amount of the error.
. The next technique included digitizing photographs individually and then assembling the digitized images of individual aerial photographs together using AutoCAD. Control points on sequential photos were again used to guide in the assembly of the various islands. The benefit of this method was that the images could be overlaid in the computer program to form an image mosaic similar to using a light table. This enabled more control points to be seen and be used in the assembling process. Once assembled and moved into the State plane coordinate system, using the same visual targets mentioned earlier, it became apparent that there was still a considerable amount of error in the series of photographs, most likely due to the inherent distortions in the photographs.
The final technique minimized some of the distortion in the aerial photographs by using a method similar to Dolan's OGMS Method. It was concluded that the State of Florida Department of Natural Resources Coastal Construction Control Line (CCCL) aerial photographs for Lee County (taken on 10/14/88) could be considered to be rectified since the amount of distortion appeared minimal and had state plane coordinates inscribed on the photos. The first step was to chose control points found in all three reference sets (CCCL photos, Feb. '96 photos and Aug. '96 photos) having similar properties as those mentioned earlier. After calibrating the digitizing tablet based on the state plane coordinates displayed, the selected control points in the CCCL photographs were then

digitized creating a map of control points with known state coordinates. A map of control points on Captiva Island can be seen in Figures 3-6 and 3-7.

RM MM4O 30495.1401,807904.4-301 M PROFILES






FIGURE 3-6 AutoCAD display window showing a portion of the base map of
Captiva Island
Once the control points were determined, both sets of recent photographs were digitized including the depicted shoreline position, reference points and visible targets. To achieve greater accuracy, before the second set of aerial photographs was taken, visual targets were placed over 16 DNR monuments throughout the county. In order to minimize fieldwork, the monuments selected to be targeted were part of the group used in the profile surveys. Targeting the monuments was accomplished by the combined efforts

r15H Pj&55 7ARCET

430352.2862.807904.4901 PROFILES




FIGURE 3-7 AutoCAD display window showing the composite map of Captiva Island
being assembled
of the University of Florida Coastal and Oceanographic Engineering Department and the Division of Natural Resources Management of Lee County. The individual digitized images of the 9" x 9" photos for both sets of recent photographs were next imported into the base map and placed in its best-fit location based on the control points mentioned above. Here, a CAD application best served our needs by having the ability to change the scale, change the alignment and shift images all with ease and minimal computer time. Each set of images was then placed on a drawing layer inside the CAD application. This was done to keep the two sets of recent photo images separated in order to reduce confusion in editing later on. Figure 3-7 shows image blocks (created by digitizing aerial

photographs) being positioned in their best fit location over the base map utilizing the control points labeled to create a composite drawing.
Once the aerial photos had been digitized and used to assemble the composite map, reference lines were then drawn originating at the coordinates for every DNR monument (information provided in the "Monument Descriptions") and extended past and perpendicular to the representative shoreline. The distances from the monument locations to the corresponding shorelines were then recorded to be compared with the beach widths determined earlier from the ground truth profiles.
After making these comparisons, some isolated areas were found to still have errors ranging from 40 to over 100 feet when compared with other nearby profiles. These particular locations were then re-digitized and imported again into the composite map utilizing additional reference points. In some cases, this provided a better fit and produced better results. Discrepancies found in the CCCL photos and severe plane tilt could account for the significant error which remains.

Once beach widths were obtained from the composite map, they required adjustment with the ground truth corresponding beach widths to conform to a known datum. With the shoreline indicators chosen being found to represent inconsistent elevations, a correction had to be applied to account for difficulties including variable tides, wave runup and setup as discussed in Chapter 3. After correcting the data using various routines, the corrected data were compared to the ground survey data to judge the accuracy of the method established.
4.1 Calculations
There are several ways to attempt to account for the changing elevations associated with the shoreline position indicator. Setting up the proper instrumentation along the entire county to measure tide, wind strength and wave characteristics at key locations simultaneously with the photography would be both expensive; time consuming and impractical. It was more advantageous and practical to adjust the high water line position data based on ground truth data. Three different routines of adjusting the data included:
1. The data were left unchanged in order to provide a control to measure
the benefits obtained by adjusting the data (refer to Table 4-1).

2. For each island, all beach widths obtained from the composite map
were shifted by a quantity X so that the average error for each island's data reduces to zero (refer to Table 4-2). With k being the number of profiled points on each island, Laerial being a specific beach width determined from the aerial photographs, Lgt being the associated beach width attained from ground truth and Lc being the
associated corrected beach width.
1 k(41
X --1 (Laeriat-L (4-1)
k =1
Lc = L,.,,eria X (4-2)
3. For each island, the change in water level, Y (due to tide and run-up)
required to reduce the average error of the sample error data to zero was calculated using equation 4-3 then used in equation 4-4 to obtain Lc (refer to Table 4-3). This differs from procedure 2, by accounting for a varying beach slope, S along the county shoreline and is more
consistent with the factors which vary along the shoreline.
-1 L -,rial S L8] = 0 (4
4 = Lara, - (4-4)
Once the data sets were adjusted by the above methods, all three sets of data were
then calibrated to minimize remaining errors. Calibration includes finding errors
associated with any two particular monuments and then linearly interpolating those
amounts and applying these as a correction factor to all those locations between the two.
The errors found at monuments other than those selected for calibration, would then be
used to help determine the accuracy accomplished by each of the procedures.

The two methods of data calibration included:
1. Utilizing the first and last monuments on each barrier island as the
ground truth profiles. (Calibration #1)
2. Utilizing every other monument on each island as ground truth
profiles. (Calibration #2)
4.2 Method of Analysis
In photography, most distributions of error are well represented by a normal or Gaussian distribution (Wolf, 1983). This type of distribution is typically described as a bell-shaped curve as shown below in Figure 4-1.

FIGURE 4-1 Normal Distribution Curve (Wolf, 1983)
Histograms are a statistical tools capable of showing the symmetry of a sample of data, the scatter of the data and the precision of the measured values based on the overall

Size of residuak

0. 19 0 0 0 W
Error between Aerial Photograph* and Ground Truth (Feet)

dimensions of a plot. Relative frequency is usually chosen for such a plot to ensure fair comparisons are being made between different size samplings. The standard deviation, S of a sample, which is contained in the hatched area in Figure 4-1, indicates to what degree 68 % of the data fluctuates from the mean. Also labeled in Figure 4-1 is the class interval, Av, which is the range of error size chosen for a particular sampling. For each size residual, the corresponding relative frequency can be determined graphically.
For each comparison made in the study, a relative frequency distribution of error was plotted in the form of a histogram in order to highlight the merits of one procedure over another as demonstrated in Figure 4-2. Throughout the results of this study, the standard deviation and mean have been provided for each set of data. Using the "raw data" in Figure 4-2 as an example, 68 % of the total sample hes within 11.8 feet from the mean or -8.5 < (68% of the Raw Data) < +15.1 feet.


FIGURE 4-2 Sample Histogram

TABLE 4-1 Calibration of Raw Data from February 1996 Survey
Raw Data Calibration Utilizing Calibration Utilizing
DNR Ground High Water Line Error Without The First and Last Monument of the Island Ever Other Monument of the Island
Monument Truth Position Without Correction Interpolation Calibration High Water Line Error Interpolation Calibration High Water Line Error
(Feet) Correction (Feet) Nunmber Coefficient Position (Feet) (Feet) Nurmber Coefficient Position (Feet) (Feet)
A-46-2 126.1 6.2 119.9 6.2 119.9
R-85 833.3 6.2 827.1 6.2 827.1
R-86 520.0 6.2 513.8 6.2 513.8
R-87 383.4 6.2 377.2 6.2 377.2
R-68 251.3 257.5 6.2 6.2 251.3 0.0 6.2 251.3 0.0
R-89 453.9 1 4.9 449.0 1 6.6 447.3
R-90 215.1 2 3.6 211.5 2 6.9 208.2
R-91 157.5 3 2.3 155.2 3 7.3 150.2
R-92 192.2 4 1.0 191.2 4 7.7 184.5
R-93 224.8 5 -0.3 225.1 5 8.1 216.7
R-94 148.3 153.3 .0 6 -1.6 154.9 6.6 6 M4 144.9 -3.4
R-95 148.3 7 -2.9 151.2 7 8.8 139.5
R-96 117.7 8 -4.2 121.9 8 9.2 108.5
R-97 108.7 9 -5.5 114.2 9 9.6 99.1
R-98 130.1 10 -6.8 136.9 10 0.9 120.2
R-e9 221.8 232.1 10.3 11 -8.1 2402 1.4 11 10.3 221.8 0.0
R-100 194.0 12 -9.4 203.4 1 7.0 187.0
R-101 277.0 13 -10.7 287.7 2 3.6 273.4
R-102 219.0 14 -12.0 231.0 3 0.3 218.7
R-103 275.0 15 -13.3 288.3 4 -3.1 278.1
R-104 256.4 268.1 11.7 16 -14.6 282.7 26.3 5 -6.4 274.5 18.1
R-105 156.8 17 -15.9 172.7 6 -9.8 166.6
R-106 425.3 18 -17.2 442.5 7 -13.1 438.4
R-107 429.4 19 -18.5 447.9 8 -16.5 445.9
R-106 176.9 157.1 -19.8 20 -19.8 176.9 0.0 9 -19.8 176.9 0.0
R-109 374.2 -19.8 394.0 -19.8 394.0
Average Error: 2.7 Avg. 17.1 Avg. 7.3
Standard Deviation: 12.9 S.D. 9.9 S.D. 15.2
Note: The monuments at which ground truth profiles are located are shown in bold type.

iz 0 m


The data produced from the various computations discussed in Chapter 4 were published in the annual report produced for Lee County titled, "Pilot Program To Quantify Shoreline Changes in Lee County" by Kreuzkamp and Dean (1997). This report includes all of the ground truth and photographic data obtained from the study. In an attempt to consolidate all of the data from the report, the following chapter will include only the most consequential information. The following chapter includes histograms used to illustrate the effectiveness of the various methods, the sources of error considered to affect the results, the favorable improvements made during the study and recommendations for further study in this area.
5. 1 Shoreline Position Error
The initial beach widths recorded from the composite map were first recorded and plotted in Figure 5-1 before being adjusted by the ground truth information.. Based on the information contained in this figure and those that follow, the success of the various calculations and methodology could be evaluated. The most distinguishable features found in Figure 5-1 are the difference in scatter of the data and the similarities of the mean values between the two surveys. The standard deviation of each set of data represents the

Lee County Raw Survey Data

Error between Aerial Photographs and Ground Truth (Feet)J
FIGURE 5-1 Comparison of uncorrected data sets
accuracy for any point within each sampling. The smaller standard deviation value for the August survey is most likely due to the improvements to be mentioned in section 5.3 which helped produce more accurate data as Figure 5-1 presents. The mean values given depict the amount each digitized shoreline would need to be shifted to conform on the average to the true High Water Line position. Although it is curious that the mean values resemble one another, it seems possible that the discrepancies in the tidal elevations and shoreline indicators for the two aerial surveys could nullify the impacts of each other on the data.
A summary table of the uncorrected data is provided in Table 5-i to show the size of the residuals associated with each island. Examining the standard deviations, the larger contributors of error are: Lacosta Island, Estero Island and Lover's Key. Lacosta Island and Lover's Key are less developed islands which results in having less available control


points (i.e. roads, pools, tennis courts, etc.) With less control points, it becomes much more difficult to create an accurate composite map. Adding visual targets to these areas will help remedy the problem in future aerial surveys. The error associated with Estero Island is probably due to a discrepancy found while digitizing the CCCL photos for that region. After several attempts of digitizing the CCCL photos, the main road near monument R-203 was consistently disjointed by 20+ feet on the created base map. It is therefore possible that there may be an error in the CCCL photos. This local area of the base map should be corrected by referencing a set of photos that were prepared at an earlier time or another source should be employed.
TABLE 5-1 Summary error results for each aerial survey by island
. . ........ .*. *. . . *. ............ .. ..*..*.*.. . .
Capotva Island
Sanibel Island 10.8 185337.5
Estero Island 20.0 2. .614.1
Lover's Key/Bonita Beach 1.0 2931913.2
In an attempt to focus on the better accuracy of the method, these "problem areas" were excluded from the data with results shown in Figure 5-2. Removing the problem areas from the group sample slightly enhanced the difference between the mean error values for the two surveys while it reduced the standard deviations. The change in standard deviation appears to be the influence of having access to only a limited number of control points within the less developed islands. The change in mean error seems to

Lee County Raw Survey Data Excluding Problem Areas

Error between Aerial Photographs and Ground Truth (Fee) FIGURE 5-2 Comparison of uncorrected data sets excluding problem areas
The next question was if any of the correction routines performed on the data resulted in significant improvements. Figures 5-3 and 5-4 focus on the various routines applied to each complete set of data. For both surveys, the scatter and average error of the data were reduced by comparable amounts by either shifting the data with or without taking into account the slope of the beach face. Although the results for either method were not dramatic, it is still felt that these techniques have merit by accounting for the varying shoreline indicators or discrepancies caused by the shoreline being inconsistently identified between sets of aerial photographs. After applying the two correction routines, each set of processed data was calibrated two different ways as discussed earlier in

indicate that the shoreline positions obtained within the problem areas did influence the data, however minimal.

9 C;

FIGURE 5-3 Comparison of results determined by applying various computations to the
February data (Including problem areas)

FIGURE 5-4 Comparison of results determined by applying various computations to the
August data (Including problem areas)

Chapter 4. Resulting errors indicated that both methods of the latter calibration produced unreliable results shown by the inconsistent changes in the standard deviations of the data.
Based on this information, the best accuracy achieved from the aerial photography for any individual measurement within the entire county was found to be 11 feet (obtained by shifting the data). This was improved further by shifting the data and then excluding the problem areas. This is shown in Figure 5-5 with the standard deviation for the August data being 7.4 feet. This level of accuracy should be compared with the level of accuracy achieved by other researchers using similar means. As mentioned earlier, Dolan et al. (1978) attained an accuracy of 21 feet with the OGMS method while Fisher and Overton (1994) documented that the accuracy of digitizing directly from aerial photographs and using USGS Topographic maps for control will have an error of 50 feet.

Lee County Shifted Survey Data Excluding Problem Areas


o W~ 0 $ W00W 0 U9 01W0 W W880W0
Error between Aerial Photographs and Ground Truth (Feet)

Figure 5-5 Comparison of accuracy obtained between the two aerial surveys

5.2 Shoreline Change Error
In order to examine the reliability of using aerial photography in determining overall shoreline change, the change in beach widths between February and August obtained from the composite maps were compared to the changes detected by ground truth. Figure 5-5 shows the changes that occurred for the entire county over the six month time span based on aerial photography and ground truth. Results based on aerial photography documented shoreline changes at every DNR monument (approximately 239 points spaced at 1/5 mile intervals) while the ground truth profiles determined the shoreline change at 40 points spaced at approximately 1 mile intervals. The positive mean values for both methods in Figure 5-5 are the influence of nourishment projects that occurred between the two aerial surveys. In an attempt to investigate the natural shoreline processes for the region, Figure 5-6 excluded known nourished areas from the

FIGURE 5-6 Comparison of shoreline changes for the entire county

Relative Frequency Distribution of Adusted Seasonal Shoreline Changes Throughout the Entire County
M Aerial Photos
0.25- Avg.=8.6 ft; S.D.-48.9 ft
E Ground Truth
0.o Avg.-22.7 ft; S.D.-48.9 ft
"7 .7 17.
Shoreline Change (Feet)

Relative Frequency Distribution of Adjusted Seasonal Shoreline Changes Throughout the County Excluding Nourishment Sites
0.35 M Aerial Photos
Avg.=-7.0 ft; S.D.-27.9 ft
. 0.30
0.25 Ground Truth
3.. 0.25.. Avg.,2.5 ft; S.D.-15.8 ft
0 010
v Shoreline Change (Feet) ^
FIGURE 5-7 Comparison of shoreline changes for the entire county excluding
nourished areas
calculations. These nourished areas included the southern tip of Gasparilla Island, Captiva Island, parts of Sanibel Island, Ft. Meyers Beach and Bonita Beach. By eliminating these nourished regions in the county, the histogram clearly shows that the remaining unnourished areas experienced a net shoreline recession. This recession is most likely the influence of storm activity and not part of a seasonal trend.
To evaluate the validity of the results of Figures 5-6 and 5-7, Figure 5-8 eliminated any composite map data that did not have corresponding ground truth data. This was done since the ground surveying provided a different amount of coverage (one mile spacing) compared with the aerial survey data which was analyzed at 1000 foot increments. The standard deviation values of the two samples represented in Figure 5-8 continues to support the fact that the accuracy of the aerial photo data is still less compared to the accuracy determined for the ground truth. Based on the average values

of the data included, it is possible to determine the accuracy of the method being applied to analyze the average shoreline change to be 3 feet. Similarly, Stafford and Langfelder (1971) found that although they had a large magnitude of error in some local instances, the mean of the composite differences was very small. It was concluded by Stafford and Langfelder that,
the composite error was concluded to be sufficiently small so as to not have a detrimental effect on the study results expressed by mean beach location changes as long as adequate care was taken in the measurement process and provided that the most accurate type of aerial photograph
available in a particular area was used. (page 574)

0.35 0.30
0.25 U. 0.20 S0.15
- 0.10
0.05 0.00

Relative Frequency Distribution of Adousted Seasonal Shoreline
Changes at Profile Locations Excluding Nourishment Sites
UAerial Photos
Av g.--0.2 f t; S.D.=28.5 f t
..... I Ground Truth
Avg.=2.5 ft; S.D.=15.8 ft

v Go I-- M W l C
vShoreline Change (Feet) A

FIGURE 5-8 Comparison of shoreline changes at profiled locations excluding nourished areas
The last significant finding was the success in tracking nourishment projects as can be seen in the right portion of Figure 5-6. After removing all other areas, Figure 5-9

shows the impact those projects had on the area beaches and the level of accuracy of the data obtained from the composite map. It was detected by means of ground truth survey that the average beach width within the designated nourished areas was increased by more than one hundred feet in a six month span. The results from the composite map (with data taken from profiled locations only) reflect changes of the same order with an error of 3.1 feet. This error agrees perfectly with the error determined earlier as 3 feet. It is also important to understand that this method not only has the ability to detect shoreline change at sites of beach construction but can easily include areas adjacent to construction to include the natural spreading of material. These transitional areas can extend for miles beyond the limits of a nourishment project and are important to include to insure that all of the material within a project is accounted for during the monitoring process.

Relative Frequency Distribution of Adjusted. Seasonal
Shoreline Changes at Nourished Profile Locations
flAerial Photos 0.50 -Avg.=103. ft M Ground Truth Avg.=l00O0ft
LL 0.30
0.004 M
80 0 0 0 0
8 Op C.J W O
VShoreline Change (Feet) A
FIGURE 5-9 Comparison of shoreline changes at profiled locations within nourished areas

5.3 Sources of Error
There are several sources of error in determined shoreline positions from aerial photography. These sources include inconsistent water level elevations along any chosen shoreline indicator, misinterpreting or inaccurately digitizing the shoreline indicator, distorted images within the aerial photographs and small scale photographs. The approximate magnitude of total error, UTO,., for the February and August flights were found to be 14.7 and 7.4 feet respectively in the previous section. This section will account for each of the contributing sources of error and attempt to quantify each one.
Although difficult to quantify, any shoreline indicator's elevation will have some degree of fluctuation which will later be referred to as CT JNDIcATOR. This variation along the shoreline is caused by inconsistent tide and other changing environmental factors. The presence of inlets and large bodies of water near a region can affect the tidal elevations along a coastline by causing the water surface elevation near an inlet to be higher or lower than the water surface elevation a few miles alongshore from the same inlet. The degree of wave setup and runup in the nearshore and swash regions can fluctuate depending upon nearshore bathymetry, wave activity and wind variability. Any of these factors can alter the elevation of a shoreline indicator from one to several feet depending upon on the location of the study area and local climate. Based on a 1: 10 slope of the beach face, the range in horizontal discrepancies can reach over 30 feet.
The effectiveness of this entire process is dependent on the person(s) identifying and digitizing the shoreline position. The method is fairly labor intensive and requires attention to detail. Each step of the methodology requires total consistency throughout each set of photographs which contributes to the difficulty. A trained and careful

draftsperson or technician would be the best candidate for this type of work. Error created by inaccurately determining the shoreline position from photographs is presently unavoidable and would increase should a smaller scale be used. A limited study was conducted to determine the variability of results obtained by different persons identifying and digitizing the shoreline from aerial photographs. Based on photographs produced at a scale of linch=250 feet, it was found that the range in error in identifying the shoreline was 3 feet. During the same study, the range in error in digitizing established points along the exact same image was I foot. The Calcomp digitizer tablet used to create the composite map for this study has limitations in precision of +-0.005 inches. At the scale of P=250 feet, the precision of this tablet conforms wen with the study results having a value of 1.25 feet. Should a smaller scale be used, lesser detail in the photographs would cause a substantial increase in error due to the limitations of personnel and equipment. Identifying and digitizing the shoreline position at a scale of 1"=500' could have a potential error of t 6 feet and 2.5 feet respectively. Since only one person digitized the photographs for the entire study, the only relevant error is the limitation of the digitizer with a standard deviation designated for later purposes as (T Digitizer
Aerial photographs will always have an inherent error due to distortion which has to either be removed or circumvented, as this project attempted to accomplish. The sources of error that cause variations in scale and distortion include: changes in the camera's altitude, changes in tilt of the aircraft, radial scale variations and relief variations of the surface photographed. The largest of these sources of error for this project was felt to be the distortion due to camera tilt. This 'lilt" is usually due to aircraft roll. Approximately one half of aerial photographs taken for mapping purposes are tilted

between I to 3 degrees (Anders and Byrnes, 1991). Even a one degree tilt can cause non-linear scale variations throughout a photo which are difficult to correct without the proper equipment/technology. The amount of error specific to this project that could be expected due to camera tilt, Cy Roll, is shown in Table 5-2. These results apply for points located a considerable distance off the principal line and therefore should be considered as the maxima.
TABLE 5-2 Potential distortion in aerial photographs due to camera tilt
I degree 26 ft 13 ft
3 degrees 80 ft 40 ft
The ground truth is the last potential source of error since a ground survey could not be performed within a one day time frame as the aerial photography was able to accomplish. The temporal spacing between the first two ground surveys allowed too much natural change of the beach to occur in order to reasonably approximate the shoreline position based on the ground truth at the time of the first aerial survey the however the latter two surveys had only 2 weeks of time between them. Even within such a short period of time, a sizable storm could alter the topography enough to affect the ground survey results as shown in Figure 5-10 below. Considering the possible differences in shoreline position at the time of flight as one-half the difference observed in the ground surveys, the standard deviation of error, (TGT, was found to be 9.0 and 5.7

feet for the February and August surveys respectively. This was based on the relationship
=GT 1 (mean)2 + SD2

FIGURE 5-10 Comparison of shoreline changes determined from sequential ground surveys
5.4 Error Analysis
With only a few contributions of error capable of quantification in section 5.3, based on the relationship shown in Equation 5-1, Roll and U Indicator could be evaluated to assess the significance of each of these factors in the study.
aTo 2 = 2GT + 2Digr +(a2R + indicator ) (5-1)

Relative Frequency Distribution of Shoreline Change Between Ground Surveys
0.450 ......
0.400 Changes between surveys 1 & 2
(Avg-3.7 ft,SD,17.7 ft) 0.350 C MEChanges between surveys 3 & 4
3 0.300 (Avg-2.7 ft, SD-11.0 ft)
i. 0.250 S0.200
0 0.150
0.100 0.050 0.000

v Shoreline Change (Feet) A

Based on the known values of error determined in section 5.3, the cumulative error due to roll and the indicator was found to be 11.4 and 4.6 feet for the February and V T- 2 2
August data respectively (Le. U +a2 I.&W., = Pa ToW a GT a Digitizer ). After
comparing with the values of error for roll stated in Table 5-2, it would seem that the method applied reduced to some degree the error due to plane roll. Based on the information given and assuming the error associated with the shoreline indicator was insignificant, it could be estimated that the resultant error would require approximately 1/2 degree tilt for both altitudes (scales) flown.
5.5 Improvements
Significant changes were implemented between the first and second sets of aerial photographs which are believed to have improved the accuracy of locating the shoreline position. It is strongly recommended that the improvements discussed in the following paragraphs be included in subsequent aerial surveys.
Increasing the scale of the aerial photos seemed to have had a significant beneficial influence on the accuracy of the data obtained. In this particular project, the first set of photographs was taken at a scale of I"=500'. This scale made it difficult to see detail including control points, shoreline indicators and visual targets while digitizing photographs. The second set of photographs had a larger scale of I"=250' which helped improve detail in the photographs and reduce distortion. Displacements of objects due to camera tilt is inversely proportional to the scale of a photograph. Theoretically, by increasing the scale of the photographs by a factor of two, the displacement of the distorted object will be reduced by one half. It is believed that the larger scale is at least

partially responsible for the difference in values of standard deviation shown in Figure 5- 1. Only additional comparisons will confirm this hypothesis.
Increasing the number of visual targets set over the DNR monuments located throughout the county helped in several ways. Visual targets were felt to be more reliable compared to other objects that were selected as primary control points since their state plane coordinates were already known compared to the coordinates of a tennis court that were identified in the CCCL photos. These visual targets also aided in the importing of images into the composite map, especially in areas where there were a limited number of common features found in the CCCL photos and the aerial photographs recently taken. These undeveloped areas include Lacosta Island, Lovers Key and parts of North Captiva and Sanibel Islands. It is important that for any subsequent aerial photographs, efforts should be made to have just as many, but preferably more, visual targets than were set for the August 25th set.
Coordinating the time of the aerial survey with the forecasted time of high tide for the region was believed to help photographically document a much more concise location of the High Water Line. By providing a tide table for a one-month time frame, the photographer was able to conform to the time frame specified while maintaining his constraints for time of day to photograph with clear skies.
To further improve the degree of accuracy, it has been recommended by researchers such as Anders and Byrnes (1991) among others that the error within erosion rate studies can be minimized by extending the time between surveys. If a shoreline change trend exists, increasing the temporal spacing results in shoreline changes which will eventually be larger than the inherent errors of the method used. Thus, the temporal

spacing allows the error to be distributed throughout the associated time frame. For example, a 5 foot error determined for a 5 year erosion study would result in one foot of error per year.

Since initiation of the project, many different techniques have been employed to create a composite map with the purpose of extracting the most accurate shoreline measurements. The most significant improvements in map accuracy are believed to be due to small changes to the flight specifications. Preliminary planning for the August survey specified a lower flight altitude (which produces a larger scale), an increase in the number of visual targets and flight time coordinated with the predicted high tide; all of which are believed to have contributed to improving the results.
The method developed here would also be suitable for documenting larger shoreline changes which occur over a shorter time span such as the result of beach nourishment projects and the impacts of major storms. For example, the average shoreline change for all of the nourished areas throughout the county was approximately 103 feet determined with aerial photography while ground truth determined the change to be 100 feet. When monitoring nourishment projects, it is important to document the topography of the beach in order to keep track of the material being deposited within the design template and naturally spread alongshore. With the cost and time required for ground surveying, areas adjacent to the nourishment projects are often not monitored resulting in no possibility of evaluating volumetric conservation. Aerial surveys provide the wide coverage needed in order to help account for all of the material placed in a region. By combining the results from a composite map with strategically placed ground truth profiles, a rapid, inexpensive and complete method can be used to track entire volumes of placed material.

6.1 Summaa and Conclusions
The best accuracy for the method developed in determining shoreline position for any one point utilizing unrectified aerial photography was 1 1 feet. The method developed by Dolan et al (1980) determined an error of 20 feet on a more energetic shoreline while Fisher and Overton (1994) documented that digitizing directly from aerial photographs and using USGS Topographic maps for control resulted in an error of 50 feet. When evaluating average shoreline change for an entire region, the accuracy of the composite error improved to 3 feet.
After evaluating all of the data, it was determined that the average natural shoreline receded approximately 7 feet ( 3 feet) during the six month time frame. When including the nourished areas, the average shoreline accreted approximately 8.6 feet ( 3 feet). This shows that the nourishment projects appear to control the net amount of beach recession throughout the entire 40 miles of shoreline within the county. It is also important to realize that these results are influenced by the seasonal profile changes and storms as mentioned earlier. The best way to filter out this uncontrollable factor and other inaccuracies is by using aerial surveys with sizable temporal spacing with special care that both aerial surveys are performed during the same season.

Since initiation of the project, many different techniques have been employed to create a composite map with the purpose of extracting the most accurate shoreline measurements. The most significant improvements in map accuracy are believed to be due to small changes to the flight specifications. Preliminary planning for the August survey specified a lower flight, altitude (which produces a larger scale), an increase in the number of visual targets and flight time coordinated with the predicted high tide; all of which are believed to have contributed to improving the results.
The method developed here would also be suitable for documenting larger shoreline changes which occur over a shorter time span such as the result of beach nourishment projects and the impacts of major storms. For example, the average shoreline change for all of the nourished areas throughout the county was approximately 103 feet determined with aerial photography while ground truth determined the change to be 100 feet. When monitoring nourishment projects, it is important to document the topography of the beach in order to keep track of the material being deposited within the design template and naturally spread alongshore. With the cost and time required for ground surveying, areas adjacent to the nourishment projects are often not monitored resulting in no possibility of evaluating volumetric conservation. Aerial surveys provide the wide coverage needed in order to help account for all of the material placed in a region. By combining the results from a composite map with strategically placed ground truth profiles, a rapid, inexpensive and complete method can be used to track entire volumes of placed material.

6.2 Recommendations For Further Study
A second phase of this study is recommended to further develop and automate the process. It is recommended that the second phase concentrate on: (1) rectifying the aerial photographs to remove distortions due to roll of the aircraft, and (2) automate digitization of the indicator representing the Mean High Water shoreline. The use of ground truth surveys would be continued as a means of evaluating the accuracy of the methodology implemented. Finally, it is suggested that the second phase effort would be most effective if focused on two areas, one developed and one in a more natural condition. This focus on two areas of reasonable longshore extent would allow methods to be developed and evaluated more rapidly and effectively and efforts to be concentrated on development of a procedure for application in future.


Anders, F.J. and M.R. Byrnes, "Accuracy of Shoreline Change Rates as Determined from
Maps and Aerial Photographs." Shore and Beach, Vol. 57, pp. 17-26, 1991.
Balsillie, J.H., J.G. Carlen and T.M. Watters, "Transformation of Historical Shorelines to
Current NGVD Position For the Florida Lower Gulf Coast." Florida Department of
Natural Resources; Tallahassee, Florida, 1987
Carter, W.E. and R.L. Shrestha, "Airborne Laser Swath Mapping: Instant Snapshots of
our Changing Beaches." Fourth International Conference, Remote Sensing for Marine
and Coastal Environments; Orlando, Florida, 1997.
Clow, J.B. and S.P. Leatherman, "Metric Mapping: An Automated Technique of
Shoreline Mapping." 44th American Congress of Surveying and Mapping, Falls
Church, VA, pp 309-318, 1984.
Crowell, M. and S.P. Leatherman, M.K. Buckley, "Historical Shoreline Change: Error
Analysis and Mapping Accuracy." Journal of Coastal Research, Vol. 7,
pp. 839-852, 1991.
Dolan, R., B.P. Hayden and J. Heywood, "New photogrammetric method for determining
shoreline erosion." Journal of Coastal Engineering, Vol. 2, pp. 21-39, 1978.
Dolan, R., B.P. Hayden, P. May, and S. May, "The reliability of shoreline change
measurements from aerial photographs." Shore and Beach, Vol. 48, pp. 22-29, 1980.
Fisher, J.S. and M.F. Overton, "Interpretation of Shoreline Position from Aerial
Photographs." 24th International Conference on Coastal Engineering, Vol. 2,
pp. 1998-2003, 1994.
Kreuzkamp, A.J. and R.G. Dean, "Pilot Program to Quantify Shoreline Changes in Lee
County." University of Florida, Gainesville, Florida, 1997.
Leatherman, S.P., "Shoreline mapping: a comparison of techniques." Shore and Beach,
Vol. 51, pp. 28-33, 1983.

McBeth, F.H., "A method of shoreline delineation." Photogrammetric Engineering,
Vol. 22, pp. 400-405, 1956.
Moffitt, F.H., "History of Shore Growth from Aerial Photographs." Shore and Beach,
Vol. 4, pp. 23-27, 1969.
McCurdy, P.G., Manual of Coastal Delineation from Aerial Photographs, U.S. Navy
Hydro Off. Pub. No. 592, U.S. Navy Hydrographic Office, Washington, D.C., 1947.
Stafford, D.B., J. Langfelder, "Air Photo Survey of Coastal Erosion." Photogrammetric
Engineering, Vol. 37, pp. 565-575, 1971.
Weber, J.D.,"Photographic Monitoring of Shoreline Movement." Shore and Beach,
pp.36-38, April 1970.
Wolf, P.R., Elements ofPhotogrammetry, 2nd Edition, McGraw-Hill, New York, 628 pages, 1983.


August J. Kreuzkamp mII was born in Mineola, NY on 03 November, 1970. After moving to Connecticut in 1984, he attended and graduated from Amity Regional High School, Woodbridge, CT, in 1988. In 1992 he graduated from Roger Williams College, Bristol, RI, with a Bachelor of Science in Civil Engineering. After graduation, he was employed by a coastal engineering firm, Ocean & Coastal Consultants Inc., where he first became exposed to the coastal engineering environment. In August 1995, he and his wife Kristen moved to Gainesville, FL, to allow August the opportunity to begin his graduate studies at the University of Florida within the Coastal and Oceanographic Engineering Department. In August 1997 he graduated with a Master of Engineering degree. Currently, he is again employed at Ocean & Coastal Consultants Inc. Trumbull, CT. with family, friends and an education that will last him a lifetime.