Below is the uncorrected machine-read text of this chapter, intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text of each book. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.
6 Predicting Future Shoreline Changes INTRODUCTION In order for the Federal Emergency Management Agency (FEMA) to administer the Upton-3ones Amendment, reliable erosion rate data for the U.S. coastlines are required. There are essentially two ap- proaches for the acquisition of such data: analysis of historical shore- line changes to forecast future evolution and a statistical method (Monte CarIo simulations) based on synoptic oceanographic data (Table 6-1~. The first method is based on an analysis of the long-term data base of shoreline location, which must also take into account the time history of human interferences (e.g., beach nourishment, navigation entrances, dredging projects, sea wails, and groin emplacement). This analysis provides an average rate of evolution as well as a distribution of the fluctuations around the trend caused by seasonal variations and episodic storm events. The existing data base of shoreline locations generally is long term (many decades to over a century) and site specific. The statistical approach is based on a knowledge of the deep and shallow water oceanographic environment (e.g., waves, wind cur- rent, storm surge) and sand sources and sinks that affect shoreline position. This information can be compiled as time series or statis- tical summaries. Sediment transport corresponding to a succession 120
PREDICTING FUTURE SHORELINE CHANGES TABLE 6-1 Methods for Determining the Ram of Shoreline Change 121 Methods Shoreline Change Analysis Monte Carlo Simulations Data base Phenomenological Not needed relationships Calibration from shoreline change data Adv antages Disadvantages Shoreline data are site specific and discrete Interpretation only Relatively simple Risk of bias on Me effects of extreme events End product Average yearly rate of erosion plus fluctuations around the average Implementation Short term Oceanographic data are synoptic arid of coarse resolution Alongshore transport plus cross-shore transport combined Calibration of model for all categories of events Physics-based approach Lack of accuracy of functional relationships Determination of probability distribution of shoreline locations with confidence bands Long term of oceanographic events then is calculated based on physics-based equations relating the magnitude of forces causing the change to observed shoreline evolution. The data base of deep-water oceanographic events is broad (i.e., valid for very large areas) and can be summarized statistically. Deep- water wave information for an ocean basin can theoretically be trans- formed into site-specific, shallow-water wave data that, in theory, can be used to determine long-term shoreline changes. The pre- dicted shoreline fluctuations are then the result of the vagaries of the forcing functions (e.g., storm occurrence, E! Nina conditions). An additional consideration in evaluating these two approaches is the need of any insurance-based program to be able to assess the relative distribution of risks in order to establish appropriate insurance premiums. This task of establishing a comrnonaTity of risk is not easy, considering the following points: 1. A single common method of predicting shoreline changes, valid for all situations, does not appear to be possible, owing to
122 MANAGING COASTAL EROSION the extreme variations in coastal morphology and oceanographic climatology. 2. Risk is based on a combination of time-dependent effects from both Tong-term trend averages, which can be established from past observations of shoreline locations, and large fluctuations of stochas- tic processes, caused by the random nature of storms. On an eroding beach of established evolution, the average risk increases with time. Seasonal fluctuations (winter-summer profiles in some regions) can be added (based on previous field data) to long-term profile change. However, damage often occurs during an unpredictable short-term episodic event or perturbation that takes place in addition to (and contributes to) general shore erosion. This chapter reviews the validity and Irritations of the two major approaches for predicting future shoreline changes. The rec- ommended approach for FEMA in determining shoreline change is the Emote Cario" method, but its utility presently is limited by the lack of sufficient correlated oceanographic and shoreline change data. Therefore, an interim methodology (historical shoreline analysis), which is based on good-quality shoreline change data and an appro- priate computer-based processing system, should be implemented. It is proposed that this interim methodology should be improved by incorporating existing data on oceanographic forces with correlated observations of shoreline change. HISTORICAL SHORELINE CHANGE METHOD Available Data Base A wide variety of data and information on beach erosion exist for the Atlantic, Gulf of Mexico, Pacific, and Great Lakes coasts. The information, however, ranges from highly accurate engineering surveys to fairly general comparisons of historical photographs and maps at various scales. The primary federal agencies engaged in the systematic collection of coastal information are the U.S. Army Corps of Engineers (COE), the National Oceanic and Atmospheric Administration (NOAA), and the U.S. Geological Survey (USGS). In 1971 the COE published an inventory of the nation's coastlines in HA Report on the National Shoreline Study." Unfortunately, the report was broad and had limited usefulness, but it is the only comprehensive, nationwide assessment of America's coasts.
PREDICTING FUTURE SHORELINE CHANGES 123 Additional information ~ available from the 30 coastal states as well as local governmental departments, colleges, and universities carrying out coastal research and from private engineering and en- vironmental consulting firms. For example, New Jersey, Delaware, North Carolina, Florida, Louisiana, California, Michigan, and Illi- nois all have active coastal data collection programs for management purposes. Changes in shore position have been delineated using a wide range of methodologies, including field measurements of beach pro- files (e.g., Dewall and Richter, 1977), visual comparison of historic changes from hand-held photography (Johnson, 1961; Kuhn and Shepard, 1980), and quantitative analysis of historical maps and vertical aerial photography through various photogramrnetric proce- dures (Leatherman, 1983a). Although field measurements have the potential to yield the most accurate data to determine beach changes, the utility of such information is severely constrained because of its temporal and spatial nonhomogeneity. Florida has the best statewide program for the collection of such information, where 10 to 15 years of data are available. For the other coastal states, such data are not even available for most beaches, and the record often extends for a decade or less for even the best-monitored recreational beaches. The data needed for shoreline mapping can be obtained from maps and charts and aerial photographs. Within NOAA the Na- tional Ocean Service (NOS) performs coastal surveys of the United States and its territories. For many coastal areas NOS topographic surveys are available dating from the m~-1800s. When historical data are compared to modern high-quality maps, long-term rates of coastal erosion and accretion can be computed. These maps can be augmented and updated with historical aerial photographs (late 1930s to present) available from many public agencies; the USGS's National Cartographic Information Center (NCIC) serves as a cen- tral repository of such data. The NOS IT" (topographic) sheets are generally the most accurate maps available for the coastal zone. Stable points located on these maps are accurate to within 0.3 mm of their actual positions at the scale of the map (often I:10,000~. The smallest field distance measurable is between 7 and 16 feet. This high accuracy makes them quite useful in delineating the land-water boundary and particularly for determining net changes over the long term. Along the Great Lakes, seasonal and long-term changes in water levels make the bluff top edge a better measure of erosional trends than the wetted bound.
124 MANAGING COASTAL EROSION The USGS topographic maps can provide additional detail in comparison to NOS T sheets, but these maps are updated at infre- quent intervals. Also, their accuracy is a problem since the USGS topographic maps are produced just within the guidelines of National Map Accuracy Standards, which allows for no more than 10 percent of the stable points tested to be in error by more than one-fiftieth of an inch at the scale of the map. On the standard 7.5-minute quad- rangles (1:24,000), one-fiftieth of an inch would mean an error of 40 feet in the actual location of a stable point. Other points, such as shoreline positions, are located with an even larger potential error. Aerial photographs can be used to provide the necessary detail and short-time interval required to detect and evaluate the processes shaping the coastline. The use of vertical air photographs to de- termine the rates of shoreline change at selected points was well documented by Stafford (1971~. Since then a number of coastal sci- entists have used air photos to monitor shoreline recession (Dolan et al., 1978; Leatherman, 1979) and to quantify changes in barrier envi- ronments (Leatherman and Zaremba, 19863. It must be remembered that air photos are not maps, and corrections for a range of distor- tions must be made to rectify this imagery for usage in quantitative analysis. In most cases final maps are produced by transferring the air photo data to an appropriate base map, and the readily available USGS topographic maps often have been selected for this purpose. As previously discussed, these maps have a 40-foot potential error to which any errors associated with air photo mapping would be added. The product is a map of relatively low accuracy, and only major changes in the shoreline can be measured meaningfully. Shoreline Indicator for Mapping Purposes The shoreline is defined as the interface between the land and water. However, the position of the shoreline on the beach face is highly variable because of changes in water level caused by tides waves, and wind and on the Great Lakes by hydrologic factors. The mean high water line is depicted on NOS T sheets as the shoreline indicator. It is preferred to any other tidal boundary for the shoreline indicator because this wetted bound can be recognized in the field and approximated from air photos. This allows for the direct comparison of data obtained from the NOS T sheets and ver- tical aerial photographs. While vegetation lines are easily recognized
PREDICTING FUTURE SHORELINE CHANGES 125 on photographs, such information was not always recorded on the historic NOS T sheets. Because the T sheets are referenced to mean high water (MHW) and this is virtually the only long-term data on historical shoreline changes available, this shoreline indicator has been adopted by most coastal investigators for mapping purposes. It should be kept in mind that the longer the period of record, the greater confidence and reliability can be placed on the trend mea- surement of shoreline change, provided there are a sufficient number of data sets. Error Analysis of Map Compilation Beach erosion measurements are subject to a variety of error sources, depending on the exact methodology used in the histori- cal shoreline change analysis. A discussion and comparison of the various mapping procedures appear elsewhere (Leatherman, 1983a). A conservative (worst-case) estimate of the maximum possible error in high-quality techniques of analyzing historical shoreline data am preaches 40 feet, which is within National Map Accuracy Standards. In practice, errors often have been shown to be less than 20 feet, the average being 11 feet for the Delaware coastal erosion mapping project (GaIgano and Leatherman, 19893. Projection of Shoreline Positions Extrapolation of trends bayed on historical shoreline change anal- ysis takes into consideration the inherent variability in shoreline response based on differing coastal processes, sedimentary environ- ments, and coastal exposures. The following discussion concerns the validity of determining long-term shoreline position changes from limited observations (e.g., snapshot views of the beach through time via time-series air photos). For the sake of simplicity, consider a hypothetical case where the shoreline location Is defined as a function of time. The trends in shoreline position shown on the two curves (Figure ~1) correspond to differences in the length of record available. The average rate of shoreline position change with respect to time is different for the record extending from 1920 to 1930 as compared to the 1920 to 1940 record. This difference is caused by the occurrence of extreme events in 1927-1930, for example, a hurricane in 1930 and the fact that accretion took place between 1930 and 1937.
126 MANAGING COASTAL EROSION The important point to consider in this hypothetical illustration (Figure ~~) is that the estimation of shoreline trend depends on both the long-term trend ( ~signal" ) and short-term "noises caused by seasonal and storm (hurricane-induced) effects. To quantify this point, suppose the shoreline position trend rate is 1 foot per year and the noise range ~ 100 feet (an extreme example). If it is desired to limit the error in defining the trend to +30 percent, the required time period of measurement to meet these conditions is 333 years. As a comparison, suppose the shoreline position trend rate is 20 feet per year, with the noise range and error limit the same. In this case the required time period of measurement is only 16.7 years. The signifi- cance of these two examples is that if it is necessary to establish, with confidence, the shoreline change trend in locations where the trend is small to moderate, the required time periods of measurement can be longer than the data record avaflable. Realizing that a portion of the noise at a point is time dependent for example, caused by seasonal and storm effectsand a portion is caused by spatial (alongshore) effects, it is possible to decrease the necessary measurement time interval by averaging the shoreline changes alongshore. In addition, poststorm and wintertime photos should not be compared to imagery acquired in the summer. Rates of beach recession can be calculated from the change in shoreline or bluff position over time. If a number of historic shore- line positions are available, then it is possible to determine rates of change in beach recession both temporally and spatially. Of course, this straightforward projection of new shoreline position based on his- torical change assumes that all the oceanographic forces (e.g., waves and sea level change) remain essentially constant. If the greenhouse- induced climate warming increases, then the rate of sea level rise likely will accelerate in the future, so adjustments must be made to project further sea level rise/shoreline movement relationships (weatherman, 1983b). In summary, erosional trend rates can only be established accu- rately in those areas where long-term shoreline positions are available or where the trend rates are large. Where beach erosion rates are calculated to be in the low range (1 foot or less per year), it must be realizecl that the reliability of this measurement is probably low owing to natural fluctuations in beach width. Therefore, prudence demands that a certain minimum setback distance be added to the average annual erosion rate to compensate for the larger possible error in trend measurements.
PREDICTING FUTURE; SHORELINE CHANGES Shoreline Location in Feet ~ 90 O 0 - ~ G 70 O Z _ ~' ~ 50 O ~ I Oh Oh 30 Benchmark 127 10 Year Base Erosion Rate I'm Event ~ ~ 1 ~ . ~ 1 20 Year Base Erosion Rate 4 1920 1 930 YEARS 90-30 1940 Rate of Erosion 1920 to 1930: = 6 ft/yr 10 90-50 Rate of Erosion 1920 to 1950: = 2 ft/yr 20 FIGURE 6-1 Example of the variation of the rate of erosion as a function of the duration of the period of Aberration, 1920 to 1940. E~tmg Nationwide Data on Beach Erosion Maps depicting shoreline changes along portions of the U.S. coasts have been made for a host of reasons and by literally hun- dreds of researchers using a range of methodologies. For instance, the New Jersey coast has been mapped to determine beach erosion rates by at least three separate groups during the past few decades (Dolan et al., 1978; Farrell and Leatherman, 1989; Nordstrom, 1977~. In addition to these statewide mapping efforts, other agencies (e.g., U.S. Army COE) have mapped specific portions of the New Jersey coast. Mapping methodologies have ranged from photocopy reduc- tion/eniargement of historical maps and photos for direct overlay to sophisticated computer-based mapping methodologies that permit correction of the inherent errors and distortions of the raw data.
128 MANAGING COASTAL EROSION The USGS has compiled these widely disparate historical data on erosion rates as part of the National Atlas series (Dolan et al., 1985~. The National Atlas is plotted at a scale of 1:6,000,000, and recession rates are displayed at intervals of ~ meter per year of shore change. Although this map provides an overview of the national coastal recession situation, it has limited utility because of its coarse resolution. The data used for the USGS National Atlas on shore change exist at a better resolution for most locations and are on an IBM- compatible PC. This system, called the Coastal Erosion Information System (CElS), also contains a bibliography of reference sources and is capable of computing standard statistical parameters (erosion rates, standard deviation, and running means) at selected distances along the shoreline. The CElS approach has limited utility for FEMA because there is essentially no quality assessment of the input data. Further limi- tations exist because most of the data records are temporally short (decades) and a variety of mapping methodologies have been used to produce the data, which therefore vary widely in resolution and reli- ability. The CElS data assembly does provide a general qualitative index of erosion-prone areas along U.S. coasts, but the data are not sufficient to help implement an interim methodology for FEMA. An example of the unplementation of the historic shoreline change method is illustrated by the New Jersey coastal erosion project. The raw data used included all available NOS T sheets, New Jersey Department of Environmental Protection orthophotos, and a specially acquired set of large-scale vertical aerial photographs for the Atlantic coast beaches. Five sets of T sheets were available from the NOS archives, ranging in age from the mid-1800s to the 1960s. The state orthophotos (based on air photos corrected for dis- torsion by stereoplotters) were of 1970s vintage. A professional aeria] survey company was commissioned to fly the coast to obtain the most recent data on shoreline position. Engineering surveys of particular areas were available for many towns but were not used because of their limited coverage on a statewide basis. Also, the USGS T maps (7.5-minute quadrangles) were not used because the historical shore- line change maps and hence calculation of beach erosion rates would have been considerably less accurate. The Metric Mapping computerized technique of data entry, dis- tortion correction, and map plotting was used to generate the his- torical shoreline change information (Leatherman and Clow, 1983~.
PREDICTING FUTURE SHORELINE CHANGES 129 Approximately eight shorelines covering a period of the past 130 years now are available to determine the long-term trend in beach erosion. Only good, high-quality raw data were used; the orthophotos and air photos acquired during summertime for seasonal consistency were selected to complement and update the NOS T sheets. Other aerial photographic data were available, but the number of sets used was limited by the funds available. The total cost for erosion map- ping was approximately $2,000 per mile of shoreline. New Jersey's data requirements for the continued implementation of its building setback regulations are for an erosion update every 5 to 10 years. PREDICTIVE MODELS Batkgronnd Within the past several decades there has been substantial in- terest in the development of calculation procedures (herein termed "models") for quantitative prediction of future shoreline changes as a result of natural or human-induced effects. These models, which include both analytical and numerical types, are well beyond the infancy stage and provide a sound foundation for the recommended longer-term methodology, yet they are not presently at a level where they can be applied and interpreted without substantial effort and skill. Therefore, these models cannot be applied easily in their present form to the type of predictions needed to implement a FEMA erosion program. The paragraphs below provide a brief overview of the status of modeling of beach systems. Appendix C presents a more detailed review, including a discussion of the features of individual models. Shoreline retreat can occur as a result of longshore sediment transport, offshore sediment transport, or a combination. Offshore sediment transport primarily is responsible for shoreline retreat dur- ing storms, whereas long-term retreat can be caused by either or by a combination of these transport components. Individual models have tended to concentrate on shore response to either longshore or cross-shore transport. Models are generally site specific for erosion and require validation against the history of a particular site. A process-based mode! requires two types of equations: (1) a transport equation relating the volumetric movement of sediment to the causative forces (e.g., waves, tides, etc.), and (2) an equation that carries out the bookkeeping of changes as a result of the sediment
130 MANAGING COASTAL EROSION movement. Some of the earliest modeling efforts simplified the above equations for the case of longshore sediment transport, thus allow- ing analytical solutions to be developed that provide considerable insight into the effects of individual parameters, such as wave height and direction. Larson et al. (1987) have summarized a number of such solutions, including the effect of constructing a groin along the shoreline, evolution of a beach nourishment project, and shoreline changes from delivery of sedunent to the coast by a river. In addition to analytical solutions, numerical solutions have been developed that allow specification of time-varying waves and tides. Cross-shore transport models have received little attention until the last few decades. These models generally are based on the concept that if the prevailing waves and tides are of sufficient duration the profile will evolve to an equilibrium shape. The complexity of these models ranges from simple ones based on correlation with field and laboratory data to those that simulate profile evolution based on time-varying wave heights and storm surges as input. The state of Florida uses a simplified version of a cross-shore mode! in defining the zone of impact owing to a Goodyear storm event. Within the last decade models have been extended to represent both longshore and cross-shore transport. This capability is espe- cially important in situations involving structures where the inter- ruption of longshore transport (e.g., by a groin) causes increases and decreases in the profile slope on the uplift and downdrift sides of the structure and corresponding cross-shore transport components. At present, longshore transport models probably are accurate to within +30 percent and cros~shore transport models to within +60 percent, providing there ~ no storm-induced interruption to the assumed prevailing condition. They are, however, the type of operational too] that would be highly useful eventually to a FEMA erosion program. Perlin and Dean (1985) have developed an e-line mode] in which both longshore sediment transport and cross-shore sediment trans- port are represented. As an example of its application, consider transport in the vicinity of one or more groins. UpUrift and downdrift of a groin, the shoreline would be displaced seaward and landward, causing an increase and decrease in the profile slopes, respectively, and thereby inducing offshore and onshore components, respectively, of cross-shore sediment transport. At each time step, the governing equation is solved by matrix inversion.
PREDICTING FUTURE; SHORELINE CHANGES 131 Limitation of Existing Modeling Methodology Both episodic (storm) events and long-term prevailing wave cli- matology influence the alongshore and cross-shore transport of sedi- ments and, consequently, the resulting displacement of the shoreline. If one considers the cross-shore transport alone, then generally the relative balance between waves (and storm surge) causing erosion and swell waves causing accretion determines the long-term evolu- tion of the shoreline. On many shorelines the seasonal changes in sea and swell cause cross-shore erosion/accretion fluctuations of the shoreline. Theoretically, given functional relationships relating the oceano- graphic wave conditions to longshore and cross-shore transport, pre- dictive methodology can be devised through mathematical modeling and statistics. However, even though much progress has been made in the understanding of surf zone hydrodynamics, there remain a num- ber of gaps, particularly relating the hydrodynamics functionally to the distribution of sediment transport. Furthermore, these processes are highly nonlinear; therefore, cumulative effects of longshore and cross-shore transports cannot be obtained deterministically from a wave rose such as found in climatological atlases (Le Mehaute et al., 1981 and 1983~. Even if one considers a single mode of transport, either longshore or cross-shore, the effects of a time series of wave forcing on shoreline dynamics cannot be added linearly. Because of this nonlinearity, the results over a number of cli- matologic events vary with the sequential order of the wave events (i.e., the same wave climatology does not necessarily yield the same shoreline evolution but depends upon the order of events). For exam- ple, 1 month of prevailing waves followed by a 2-day storm does not yield the same results as a 2-day storm followed by 1 month of pre- vailing waves, everything else being equal. Finally, littoral transport is a nonlinear, sensitive function of wave approach direction, and wave rises are reported too coarsely for an accurate determination of littoral drift. Given reliable functional relationships between wave-surge-cur- rent and sediment transport, only short-term episodic events can in theory be analyzed deterministically based on the short time history of the incident waves and surge during these individual storms. Under these conditions, short-term evolution of shorelines can be predicted deterministically in the two cases where (1) shoreline changes are the result of variations of longshore drift because of wave refraction and curvature of the bottom contours (Le Mehaute and Soldate,
132 MANAGING COASTAL EROSION 1980; Perlin and Dean, 1986) and (2) shoreline change is the result of cross-shore movement from waves and storm surge combined (Kriebel and Dean, 1985~. This is described in Appendix C of this report. However, it is reemphasized that these two methods are based on functional relationships that must be improved by further research. A COMPREHENSIVE METHOD Ok PREDICTING SlIORE[INE CHANGES Backgrolmd Shoreline displacement is influenced both by the long-term pre- vailing wave climate of frequent occurrence (i.e., of high probability) and by rare, extreme episodic events of low probability. It is influ- enced also by the quantity and quality of sand sources and sinks and by human-induced factors. Wave climate summaries provide a large population of oceano- graphic events over a relatively short period of time (years to decades). The sequential time history of these numerous events is seldom avail- able. Most likely they are grouped statistically in the form of prom ability of exceedance curves and wave roses. Because these data are based on a large population, they generally are reliable. However, because of cumulative errors over a large number of events (and their sequential and nonlinear effects), shoreline changes cannot be determined by the phenomenological deterministic approach at this time. Their average effects can be obtained theoretically from the site-specific data base of shoreline change over a relatively short pe- riod of time, excluding the effect of extreme events. Unfortunately, adequate data for this purpose are seldom available. Episodic storms present a relatively smaller population of events, requiring a much longer period of observation than may be available from the site-specific and duration-lim~ted shoreline data base. How- ever, a data base of the causative oceanographic events valid for a broad area over generally longer periods of time may be available, from which site-specific forcing events can be determined. Then, formulation of the functional relationships relating sediment trans- port to these forcing events can allow for determination of shoreline changes. These functional relationships can be calibrated from the site-specific shoreline data under extreme conditions and the post- storm recovery time. Then, by adding their effects linearly, the cu- mulative shoreline changes from a series of extreme events of known
PREDICTING FUTURE SHORELINE CHANGES 133 probability can be calculated determ~nistically. For example, this approach can cover a distribution of extreme events occurring over a century. The data base can be extended synthetically and the method can be refined by application of a method currently in use by FEMA to establish storm surge statistics. Method Based on Monte CarIo Simulations The method described in the preceding paragraphs requires that it be valid to add linearly the effects of all oceanographic events. Actually, beach changes are caused by every uncorrelated event, depending upon the initial conditions resulting from the previous change. In particular, beach accretion following a storm (i.e., during recovery time) depends upon the plan and profile, the topography and bathymetry left by the storm, and the movement of nearby es- tuaries, etc. Owing to the lack of a complete time history of sea state and the sequence dependence and nonlinear nature of the re- lated beach processes, a stochastic approach to shoreline evolution is necessary for scientific exactness. The problem is indeed inher- ently chaotic (i.e., deterministic but unpredictable). A Monte CarIo simulation of incident waves provides a method consistent with the natural processes. This method translates the random nature of the sea state into deterministic events, the sum of which gives the same wave energy rose as provided by summary atlases. Shoreline evolu- tion then is deterrn~ned statistically by a succession of Monte CarIo simulations of wave climatology, including both prevailing wave con- ditions and storms (Le Mehaute et al., 1983~. The forcing functions are randomly defined by season and by a multiplicity of time series of varying events (e.g., direction, intensity, etc.) that can all be grouped in the same statistical summaries. The random nature of oceanographic events is accounted for by a large number of Monte Cario simulations for the same location. The Monte Cario sunulations of wave approach direction at fine resolution, in addition to the multiplicity of wave energy levels, also alleviates the discrepancy resulting from the coarse discretization of the wave angle from the atlases. Multiple Monte Cario simulations for the same location allow the determination of a multiplicity of shoreline distances from a reference point, from which a probability distribution of shoreline locations can be obtained as a function of time. Standard deviation and confidence bands that increase with time and number of simulations also can be obtained.
134 MANAGING COASTAL EROSION The Monte CarIo sunulation also could be used for determin- ing storm surge statistics ~ well as the Joint Probability Method presently used by FEMA. The Monte CarIo simulation has not been used due to the relative complexity of storm surge calculations and the large number of simulations that this method requires. (Typi- cally, 300 storm surge calculations for each study area are needed for the Joint Probability Method.) However, the Monte Cario simulation for storm surge Is not needed, since storm surges are independent events and their effect on flooding generally is not cumulative. Even though it represents a physically more complex phenomenon than storm surges, the mathematical modeling of shoreline evolution is relatively straightforward. This allows the processing of a great number of simulations as required by the Monte CarIo method. In principle, the Monte CarIo simulations provide as many shoreline positions as needed] for determining confidence bands and standard deviations. The most significant limitation of Monte Cario simulation is our present poor understanding of beach recovery time scales that would be required to provide an initial condition for each succeeding event to be modeled. Even in the case of well-documented, long-term statistics of prevailing climatology and episodic events, the state of the art is such that this approach, even though desirable, is difficult at this time. This is due primarily to the lack of reliable functional relationships relating the physics of sediment transport to the forcing events. In particular, more research and understanding are needed of the processes relating to the long-term erosion trend and the recovery of beaches following storms. Nevertheless, because it is practically the only theoretically rational approach to the problem, every effort should be made for its development and application. A summary and comparison of the Historical Shoreline Analysis and Monte Cario simulation methods are presented in Table ~1. FEMA's Present Methodology DEFINITION OF IMMINENT COLLAPSE A house is defined as in danger of imminent collapse (1) if its distance from shore is less than a critical distance defined below and (2) if its structural integrity and, in particular, its foundation do not satisfy construction code criteria such as described in the FEMA report (January 22, 1988) on "structure subject to imminent
PREDICTING FUTURE SHORELINE CHANGES 135 collapse." The reference line from which the requisite distance is to be measured ~ defined as follows: 1. bluff edge; 2. top edge of escarpment on an eroding dune (normal high tide should be near the toe of the dune and there should be indications that the dune is actively eroding); 3. normal high tide line, which is indicated by the following: a. vegetation line (must be in an area of low tidal range where permanent vegetation exists just above high tide), b. beach scarp (a 4- to Chinch cut at the upper limit of high tide), debris line (deposited by the normal high tide), and d. upper limit of wet sand; and 4. vegetation line (when none of the above can be located, use the seaward-most edge of permanent vegetation this is intended to be seldom used). This list is presented in priority order; if the first recommended feature is not present or suitable for use, the next feature should be used, and so forth. However, only the first two reference features are appropriate for use with the interim methodology recommended herein. In most state programs the setback line refers to a fixed baseline determined at a given time and rarely changed. In the present context the reference line is continuously moving shoreward and must be reviewed annually at the time of the construction permit process. Accordingly, the reference line may vary for two adjacent structures built at two different periods of time. The distance from that line defining the zone of imminent col- lapse is obtained by the sum of five times the annual rate of erosion and an additional distance defined by a 50 percent probability that the distance will be exceeded within the next 3 years. For lack of accurate determination of the confidence bands allowing an accurate definition of the 50 percent probability deviation from the mean, a [(}foot distance will be added to the five times annual rate of erosion. RECOMMENDED METHODOLOGIES Introduction The following two methodologies are recommended by the com- mittee for use in determining shoreline change rates. The committee
136 MANAGING COASTAL EROSION recognizes that the use of available shoreline recession data such as aerial photos and profiles would be the least costly to implement shoreline change mapping. However, it would be preferable, although more costly, to utilize oceanographic data in the determination of shoreline change rates. Historical Shorelme Change Method Historical shoreline mapping can immediately provide the req- uisite data on erosional trends for implementation of the FEMA program. While there is already a plethora of such data available for the U.S. coasts, standards must be set and met for inclusion of such information into the national computerized data bank. There are three basic requirements for the acquisition of reli- able, accurate, and readily usable information on erosion rates as determined from historical shoreline analysis: 1. Use of only good-quality raw data (from high-quality maps, large-scale air photographs, and survey profiles). 2. Utilization of a mapping procedure that allows for the com- pilation of both map and air photo data and that permits the rec- tification of these raw data such that they meet or exceed National Map Accuracy Standards. 3. Output of PC-based digital data in the State Plane Coordi- nate System that readily permits the calculation of erosion rates at a predetermined basis along the shoreline or at specified locations and includes a map-plotting capability on a PC-based system. Determination of the erosion trend should be based on the longest period of record available, but provisions must be made for any human-inducec] effects (e.g., groins, jetties) on shoreline position during recent times. The earliest reliable maps commonly available for the U.S. coasts are the NOS T sheets. There are approximately 6,000 T sheets available from NOS archives in Rockville, Maryland. The earliest information dates from the m~-1800s and extends to the 1970s; for many coastal areas, approximately four sets of histor- ical maps are available at 30~year intervals. All rectifiable NOS T sheets should be utilized in a long-term analysis of historical shoreline changes. Vertical aerial photography should be used to complement and update the NOS map data. Literally ganglions of historical air photos exist of the U.S. coasts. The earliest imagery dates from the late 1930s and early 1940s to the present. From the 1960s to date, most
PREDICTING FUTURE SHORELINE CHANGES 137 coasts have been photographed at 5-year intervals by various agencies (e.g., USDA, SCS, COE, DOl). Urbanized coasts, such as Ocean City, Maryland, are now photographed several times a year by professional aerial survey companies. Therefore, lack of data generally is not a problem, and its selection for mapping typically is governed by the coverage and scale. Good-quality raw maps and air photo data still must be corrected for distortion or otherwise rectified to make them usable for the determination of reliable, accurate rates of beach or bluff erosion. A host of methodologies have been utilized in the past, ranging from photocopy reduction/enlargement of air photos and maps for direct overlay comparisons to sophisticated computer-based mapping systems (Leatherman, 1983a). It must be clearly understood that the best raw data cannot be expected to yield high-quality erosion rate information without proper corrections/rectifications to remove inherent errors and distortions. The compiled data on historical shoreline changes should in any case meet or exceed National Map Accuracy Standards. Shoreline change maps and the derived rates of beach erosion must be interpreted by professionals; otherwise, misleading or even wrong conclusions can be drawn from a causal inspection of the data. For example, a variety of patterns of shoreline behavior (e.g., lin- ear recession, cyclic changes near inlets, and engineering structural- induced trends) exhibited along the New Jersey coast are determined from the historical data (Farrell and Leatherman, 1989~. Therefore, professional judgment ~ required for proper interpretation and ap- plication of erosion data for the FEMA program. There are two possible outputs for the erosion data: tabular format or map products. Both products are desirable as each has special advantages to FEMA in terms of information display and an- alytic calculations. The historical shoreline change data should be in digital format on a State Plane Coordinate System for computational ease. A PC-based, user-friendly, menu-driven system should be uti- lized to facilitate calculation of erosion rates on a predeterminated basis (e.g., 50 meters) along the shoreline or at specified locations of particular interest. In addition, FEMA personnel should have the option of viewing the data in map format, wherein all or a specified portion of the historical shorelines appear in their spatial context. Oftentimes, this spatial representation of the shoreline data can be invaluable in understanding apparent anomalies in the beach erosion data as it appears in tabular form.
138 MANAGING COASTAL EROSION [ong-Term Methodology In a previous section the recommended ~10, ~30, and ~60 zones are defined by the distance between the reference line and 10 times, 30 times, or 60 times the annual rate of erosion, respectively. This is based on the assumption that the erosion rate is constant over time and is determined by the historical shoreline change method. In actuality, erosion rates can vary through time, depending upon the geological features of the substrata. For example, a rock formation may be uncovered by erosion, which would not respond to waves as readily as loose sand. Conversely, dune overtopping and massive overwashing could increase the amount of storm-induced beach erosion. The long-term methodology provides a better definition of the ~ zones by taking into account the probability distribution of shoreline location from short-term climate variation (such as storms) about a slowly changing average location. Once this long-term methodol- ogy is implemented, a redefinition of the Zones is recommended, commensurate with the new knowledge obtained. Beach erosion trends can be determined by the historical shore- line mapping methodology until this preferred long-term approach of using oceanographic data and statistical treatments can be un- dertaken. In actuality, the preferred methodology involves utilizing available records of shoreline recession for analysis of the time history of oceanographic forces (e.g., wind waves, storm surges, etc.~. For lack of data, established functional relationships (relating longshore transport to the incident alongshore wave energy on the one hand and cross-shore transport to the onshore wave energy and storm surge on the other hand) will be used, as described in a previous section. Following the availability of long-term oceanographic data, sta- tistical methods can be introduced to improve the accuracy of shore prediction. Standard variation and confidence bands about the av- erage or most probable shoreline location also can be defined. As previously stated, implementation has to remain flexible, consider- ing the geographic variability of coasts, because a set relationship valid at one place may not be valid elsewhere. Time and availability of resources are also a factor in implemen- tation of a statistical approach. Initially, a ranking based on storm intensity may be used, but results from Monte Cario simulations are much desired. The corresponding methodologies should be de- veloped in parallel in order to assess the errors and discrepancies between various approaches by sensitivity analysis.
PREDICTING FUTURE SHORELINE CHANGES 139 It is anticipated that this program will be implemented county by county. At a later date, the matching of the prediction at the boundaries of each county anti be required so as to make the result consistent for sake of fairness in the insurance rate. ESTABLISHMENT OF A COMPUTERIZED NATIONAL DATA BASE A tremendous amount of shoreline change data must be as- sembled, analyzed, and interpreted properly for implementation of erosion-based setbacks by FEMA. High-quality data obtained from historical shoreline change analysm should be acquired at close in- tervals along the coast. Predicted shoreline change from a statistical treatment of the long-term oceanographic information will be less site specific, but interval measurements can be extracted from the data set. In essence, this voluminous information available as output from both methodologies must be entered, manipulated, and out- put on a personal computer system for ease of operation by FEMA personnel and other users. Advent of the personal computer with high-capacity hard-disc storage and computerized mapping capabilities makes this technol- ogy most attractive to agencies involved in the analysis and re- trieval of geographical-based data. These new technologies, which include off-the-shelf Geographic Information Systems (GISs) and user-friendly computerized plotting routines, can be readily utilized by FEMA for their national data base on erosion rates. A GIS data base management system is based on true geograph- ical locations (latitude and longitude) but also utilizes town/city boundaries, postage zones, and locality names in terms of search and retrieval. This aspect greatly facilitates the utility of such a system, making it readily understandable to the casual user. Another major advantage of the GIS approach as compared to just ~inventory"-type system is that it can be used to display erosion information in a map format. Computerized mapping virtually has replaced manual cartography at all major companies and government agencies involved in mapping efforts of geographical-based data. Erosion data in numerical strings (e.g., Metric Mapping; I.eather- man, 1983a) or in discrete point format (e.g., state of Florida monu- ment system) can be easily entered and utilized in a GIS computer- ized format. Therefore, existing high-quality data sets can be entered directly into an aIready-available data base management system. As
140 MANAGING COASTAL EROSION computer-based, long-term erosion data for the other coastal states are generated by historical shoreline change analysis, all this infor- mation can be incorporated to form a true national data base. Analysm of the historical shoreline data wall be useful to FEMA for comparative purposes once the data from the modeling efforts become available. Fortunately, the GIS data management system can accornrnodate both data sets and, in fact, serve as the host for ancillary information, such as the location of buildings. This system offers the capability for the overlay of erosion data (historical and/or mode} derived) and insured beach-front buildings on a map at any scale desired. This ability To view the situation" or gain a quick spatial overview provides FEMA administrators with perhaps as much utility as the actual sophisticated computer-based processing and data base management utilities. RESEARCH AND DATA NEEDS The problem of shore erosion is not new, and unplementation of the Upton-Jones program can capitalize on a large amount of information from past investigations. Nevertheless, coastal processes are complex, such that the state of the art in predicting coastal erosion remains relatively poor in regard to applying it to the FEMA program. In order to improve the methodology for assessing beach erosion and the risk of collapse of structures, much more research needs to be undertaken by FEMA and other appropriate agencies, such as NOAA and the COE on the following: 1. determination of the long-term wave climatology through field data collection programs; 2. monitoring of beach response to wave climate variations and episodic events; and 3. more research on predictive mathematical and probabilistic models of probability distribution of shoreline locations by a Monte CarIo simulation of the wave climate, taking into account both the longshore and cros~shore transport. It is clear that better and longer-term data on shoreline changes should be collected. Specific efforts should be directed toward quan- tifying storm-generated erosion through prestorm and poststorm sur- veys as well as the period of beach recovery. The use of remote sensing should be considered for monitoring of beach and dune erosion.
PREDICTING FUTURE SHORELINE CHANGES REFERENCES 141 Dewall, A. E., and J. J. Richter. 1977. Beach and Nearshore Procession South- eastern Florida. Pp. 425-443 in Proceedings, ASCE Specialty Conference on Coastal Sediments 377. Dolan, R., B. Hayden, and J. Heywood. 1978. Analysis of coastal erosion and storm surge hazards. Coastal Eng. 2:41-53. Dolan, R., F. Anders, and S. Kimball. 1985. Coastal Erosion and Accretion. National Atlas. Reston, Va.: U.S. Geological Survey (map). Farrell, S., and S. P. Leatherman. 1989 (in press). Erosion Rate Analysis of the New Jersey Coast. New Jersey Department of Environmental Protection report, Trenton, N.J. Galgano, F., and S. P. Leatherman. 1989. Coastal Erosion Mapping of the Delaware Atlantic Coast. Delaware Department of Natural Resources report, Dover, Del. Johnson, J. W. 1961. Historical photographs and the coastal engineer. Shore Beach 29~1~:18-24. Kriebel, D. L., and R. G. Dean. 1985. Numerical simulation of time-dependent beach and dune erosion. Coastal Eng. 9:221-245. Kuhn, G., and E`. P. Shepard. 1980. Coastal Erosion in San Diego County, California. Coastal Zone 80: Proceedings of the Second Symposium on Coastal and Ocean Management, November 17-20, 1980, Hollywood, Florida, ASCE, Vol. III, pp. 1899-1918. Larson, M., H. Hanson, and N. C. Kraus. 1987. Analytical Solutions of the One-Line Model of Shoreline Change. Technical Report CERC-87-15. U.S. Army Engineer Waterways Experiment Station, Coastal Engineering Research Center. Le Mehaute, B., and M. Soldate. 1980. A Numerical Modelling for Predicting Shoreline Change. U.S. Army Corps of Engineers, Coastal Engineering Research Center, No. 80-6. Le Mehaute, B., J. D. Wang, C. C. Lu. 1981. Monte Carlo Simulation of Wave Climatology for Shoreline Processes. Proceedings of Conference on Directional Wave Spectra Applications, Berkeley, California, ASCE. Le Mehaute, B., S. Wang, and C. C. Lu. 1983. Wave data discretization for shoreline processes. J. Waterways, Port, Coastal Ocean Eng., ASCE 109(2)63-78. Leatherman, S. P. 1979. Migration of Assateague Island, Maryland by inlet and overwash processes. Geology 7:104-107. Leatherman, S. P. 1983a. Shoreline mapping: A comparison of techniques. Shore Beach 51:28-33. Leatherman, S. P. 1983b. Geomorphic effects of projected sea level rise: A case study of Galveston Bay, Texas. Proceedings of Coastal Zone 83, ASCE, pp. 2890-2901. Leatherman, S. P., and B. Claw. 1983. UND Shoreline Mapping Project. Geoscience and Remote Sensing Society Newsletter. IEEE, Vol. 22, pp. 5-8. Leatherman, S. P., and R. E. Zaremba. 1986. Dynamics of a northern barrier beach, Nauset Spit, Cape Cod, Massachusetts. Bull. Geological Soc. Am. 97:116-124. Nordstrom, K. F. 1977. The Coastal Geomorphology of New Jersey. Rutgers University Technical Report, New Brunswick, N.J. 39 pp.
142 A~4NAGING COASTAL EROSION Perlin, M., and R. G. Dean. 1985. 3D models of bathymetric response to structures. J. Waterways Port, Coastal Ocean Eng., ASCE, 111~2~:153- 170. Perlin, M., and R. G. Dean. 1986. Prediction of Beach Planforms with Lit- toral Controls. Proceedings, Sixteenth Conference on Coastal Engineering, ASCE, New York, pp. 1818-1838. Stafford, A. L. 1971. Shore and Sea Boundaries, Volume 2, U.S. Department of Commerce, Publication 10-1. Washington, D.C.: U.S. Government Printing Once. 749 pp.
