Marine Geodesy, 25:27– 36, 2002 C 2002 Taylor & Francis Copyright ° 0149-0419/02 $12.00 + .00 Digital Tide-Coordinated Shoreline RONGXING LI RUIJIN MA KAICHANG DI Department of Civil and Environmental Engineering and Geodetic Science Ohio State University, Columbus, Ohio, USA The shoreline is one of the most important features on earth’s surface. It is valuable to a diverse user community. But the dynamic nature of the shoreline makes it dif cult to be represented in a naturally dynamic style and to be utilized in applications. The of cially used shoreline,for example in nauticalcharts,is the so-calledtide-coordinatedshoreline. It is also the shoreline that makes the computation of shoreline changes and associated environmentalchanges meaningful.Mapping of the tide-coordinatedshoreline has been very costly. On the other hand, instantaneous shorelines extracted from different data sources may be available. Also, high-resolutionsatellite and airborne imagery have the capacity of stereo imaging and can be used to extract instantaneousshorelines at a high accuracy and low cost. This article proposes an approach to derivation of digital tidecoordinated shorelines from (a) those instantaneous shorelines and (b) digital coastal surface models and a digital water surface model. Some preliminary study results, analysis, and the potential of the approach are discussed. Keywords shoreline, tide-coordinatedshoreline, coastal terrain model, water surface model Shorelines are among the most important terrain features on earth’s surface. They are recognized by the International Geographic Data Committee (IGDC) as one of the 27 most important features. The location and attributes of a shoreline are highly valued by a diverse user community (Lockwood 1997). Shorelines have never been stable in either their long-term or short-term positions. The changes are caused by natural processes, human activities, or both. Regardless of the causes, the shoreline changes impact their immediate environments either positively or negatively. Thus shoreline mapping and shoreline change detection become critical to safe navigation, coastal resource management, coastal environmental protection, sustainable coastal development, and planning. By de nition, a shoreline is a linear intersection of coastal land and the surface of a water body (Figure 1). Because of the dynamic nature of the water body and the coastal land, the shoreline changes all the time. Thus, this shoreline is usually called an instantaneous shoreline. In a geographic information system (GIS) it is currently impossible to depict the dynamic characteristics of the shoreline. In practice, the instantaneous shoreline cannot be directly used for shoreline mapping and navigation, nor can it be employed for quantifying Received 18 July 2001; accepted 12 October 2001. We acknowledge funding from NSF Digital Government Program and Sea Grant—NOAA National Partnership Program and the Lake Erie Protection Fund (LEPF), and matching funding from the Coastal Service Center, Of ce of Coastal Survey, and National Geodetic Survey of NOAA. Address correspondence to Rongxing Li, Department of Civil and Environmental Engineering and Geodetic Science, The Ohio State University, Columbus, OH 43210-1275. E-mail: [email protected] 27 28 R. Li et al. FIGURE 1 A shoreline de ned as a linear intersection between coastal land and a water body. shoreline changes. A shoreline that is de ned based on a stable vertical datum can be treated as a reference shoreline and used to differentiate shorelinechanges. Such a shoreline is called a tide-coordinated shoreline, that is, the linear intersection between the coastal land and a desired water level. In the United States, internal shoreline mapping is the responsibility of the National Oceanic and Atmospheric Administration (NOAA). The National Geodetic Survey (NGS) of NOAA uses MHW (Mean High Water) and MLLW (Mean Lower Low Water) to de ne the tide-coordinated shorelines (Figure 2). MHW and MLLW are the averages of high and lower low water levels, respectively, over a period of 19.2 lunar years. All high water heights are included in the average of MHW, where the type of tide is either semidiurnal or mixed. Where the type of tide is predominantly diurnal, only the higher high water heights are FIGURE 2 The pro le of MHW and MLLW. Digital Tide-Coordinated Shoreline 29 included in the average on those days when the tide is semidiurnal. The lower low water is the lower of the two low water levels of any tidal day where the tide is of the semidiurnal or mixed type. The single low water occurring daily during periods when the tide is diurnal is considered to be lower low water (Shalowitz 1962). On NOAA nautical charts both MHW and MLLW coordinated shorelines are shown on tidal areas. Figure 2 depicts the pro le of MHW and MLLW. A Review of Shoreline Mapping Techniques The rst shoreline mapping endeavor in the United States was in 1807, and the rst shoreline survey was completed in 1834. Around the time of World War I, the entire U.S. coast, except for Alaska, had been surveyed at least once. In the early years, the main device used was the plane table, which can obtain a high accuracy. As control for shoreline mapping, the surveying system was de ned by a series of points with known latitudes and longitudes, and this established the rst geodetic surveying system. This geodetic system was called the “line of sight” observation method in which each surveying point must be visible by at least one other surveying point. This geodetic system evolved into more advanced surveying methods like space-oriented observations, Bibly towers, electronic distance measurement (EDM), and the Global Positioning System (GPS) (CSC 2001). In the early era of coastal surveying, the mean high water line was delineated. The line was determined more from physical appearance than precisely running spirit levels along the coast. What the topographer actually delineated were the markings left on the beach by the last proceeding high water, barring the drift cast up by the storm tides (Shalowitz 1962). Plane table mapping was replaced by photogrammetry during the onset of World War I. In 1919, an investigation was started to evaluate the feasibility of aerial photography in compiling shoreline maps, and by 1927 the full potential of photogrammetry to complement the production of charts and maps was recognized. But until 1927, practically all the topographic surveys were made by plane tables. Since 1927, aerial photographs and photogrammetric methods have been utilized increasingly to provide the required topographic information along the coast (Shalowitz 1962). Analytical photogrammetry has been the primary technology for shoreline mapping. With recent advances in digital photogrammetry, GPS, and other all-weather sensors, researchers have been exploring the potential of more ef cient and economic shoreline mapping techniques.Land vehicle based mobile mapping technologyin local shoreline mapping uses GPS receivers and a beach vehicle to trace watermarks along the shorelines (Shaw and Allen 1995; Li 1997). In contrast, regional and national shoreline mapping has been conducted by aerial photogrammetry and LIDAR (Light Detection and Ranging) depth data (Slama et al. 1980; Ingham 1992). GPS technology has been applied to provide orientation information and to enhance aerial photogrammetric triangulation (Lapine 1991; Merchant 1994; Bossler 1996). Recently, satellite-imaging systems have increasingly improved image resolution. A new generation of high-resolution (one-meter) satellite-imaging systems has been or will be launched (Fritz 1996; Li 1998), including the IKONOS imaging system with the capability of stereo imaging. Since the in-track stereo mode is provided, stereo pairs that are necessary for deriving elevation information of objects can be formed in quasi real time; the cross-track stereo requires additional time allowing the satellite to revisit the same area from a neighboring track. An investigation of shoreline mapping using such high-resolution satellite images demonstrated a promising shoreline mapping accuracy of 2 m and a great reduction of required ground control points (Zhou and Li 2000; Li et al. 2001). 30 R. Li et al. Digital Tide-Coordinated Shoreline The objective of this article is to give the preceding review of shoreline mapping techniques and to discuss new methods of tide-coordinated shoreline mapping. In current practice, aerial photographs for shoreline mapping are taken when the water level reaches the desired value (MLLW). This requires coordination between the water gauge reading and aerial photographing to make sure that the shoreline that appears in the images is the tide-coordinated shoreline. We call this tide-coordinated shoreline physical tide-coordinated shoreline. In the case of satellite imaging, the imaging technology has improved so much that the image resolution is comparable to that of aerial photographs, and it also has stereo mapping capability. In principle, the images can be taken repeatedly within a short period. In addition, it provides multispectral signals that are not available or are limited in the aerial imaging case. However, it not realistic to arrange satellite imaging at the desired water levels. The shorelines delineated from the satellite images are instantaneousshorelines. We believe that there are relationships between the instantaneous shoreline and the tide-coordinated shoreline. We call the tide-coordinated shoreline thus derived digital tide-coordinated shoreline (DTS). Such a digital tide-coordinatedshoreline does not require the eld coordinationbetween the gauge reading and the aerial photography and can eliminate the associated costs. It improves the shoreline mapping ef ciency by integrating instantaneous observations that are more widely available and less costly. The tide-coordinated shoreline and shoreline changes can be accurately computed, and future shorelines can also be generated through scenarios. This shoreline mapping technology may mark the start of a new era of digital shoreline mapping and coastal change detection and monitoring. Two approaches toward generation of the digital tide-coordinated shoreline are discussed below. Approach I: DTS from Instantaneous Shorelines Figure 3 depicts the situation where instantaneous shorelines f (X ; Y ; Z )t 1 ; f (X; Y; Z )t2 , and f (X; Y; Z )t 3 at times t1 ; t2 , and t3 are derived from nontide-coordinated airborne or satellite images. The objective is to compute the tide-coordinated shoreline F(X, Y, Z) at the desired water level of MLLW. The lake shoreline in Figure 3 can be extended to an open sea environment and should not affect the discussion in the remaining part of the article. A simpli ed model such as EPR (End-Point Rate) method may calculate a recession/ advancing rate on each transect of the shoreline. The desired shoreline position can then be estimated by a temporal interpolation or extrapolation. But the EPR method assumes that the shoreline position changes in one direction and linearly, which does not FIGURE 3 Instantaneous shorelines (t1 ; t2 , and t3 ) and the tide-coordinated shoreline. Digital Tide-Coordinated Shoreline 31 match the situation in the real world. In Figure 3, the shoreline at t3 is between those at t1 and t2 , possibly because of the tidal effect. This demonstrates that an improved approach should consider both the shoreline positions and water levels. In principle, the general DTS function F(X, Y, Z) can be decomposed into X D FX (Ä; 8; t ); Y D FY (Ä; 8; t ); and (1 ) Z D FZ (Ä; 8; t ); which are functions of the coastland geometry Ä, water surface 8, and time t. The interaction between Ä and 8 at a certain time t results in the DTS. To simplify the shoreline geometry, we propose piece-wise polynomials to describe the shoreline in Figure 3. The particular piece can be parameterized by a one-dimensional parameter s that starts from the beginning and is measured along the shoreline. Equation (1) then becomes X D FX (ao ; a1 ; a2 ; : : : ; an ; s; t ); Y D FY (bo ; b1 ; b2 ; : : : ; bn ; s; t ); and (2 ) Z D FZ (co ; c1 ; c2 ; : : : ; cn ; s; t ); where ao (h); a1 (h); : : : ; cn (h) are temporal polynomial coef cients that characterize the geometric shape of the shoreline. They may themselves be represented as polynomials of the water level h: ao D ®oo C ®1o h C ®2o h 2 C : : : : : : ::: ::: (3 ) cn D ·on C ·1n h C ·2n h 2 C : : : : : : The function F may take different forms according to characteristics of various coastal areas. For example, we may adjust orders of a0 ; a1 ; : : : ; cn based on shoreline topography ( at, steep slope, bluff, wetland, etc.). We can change the order of s considering the long shore topography; the detailed terms of the parameter h should represent the effect of the tide and coastland change on the shoreline. Further, for each coef cient in Equations (2) and (3) a signi cance coef cient may be used to measure if the term is needed for this particular piece of shoreline. As observations, vertices of the instantaneous shorelines derived from satellite and airborne images provide (X, Y, Z ) measurements in Equation (2) with given s and t. Gauge station water level observationsare time series observations with locations. A preprocessing of the gauge station data is needed to associate the locations of gauge stations with the instantaneous shorelines, as well as times. These measurements contribute mostly to the determination of the temporal relationships in Equation (3). The entire shoreline is described by a collection of the polynomial pieces. There is a need to investigate the ways and criteria for breaking the shoreline to pieces. We may consider lengths, curvatures, topography, geological material types, hydrographic nature, shoreline erosion rate, and other information from an existing coastal GIS. The overall shoreline model requires that the shoreline pieces be continuous at the connections of adjacent pieces, which should be enforced in an integrated adjustment of all observations. 32 R. Li et al. The coef cients of Equations (2) and (3) are estimated using the Least Squares principle in the global integrated adjustment. Finally, given the time and water level information, we are able to determine the digital tide-coordinated shoreline. In this way, the shoreline pieces are estimated globally in the least squares adjustment. In addition to the temporal modeling, the overall shoreline shape, the detailed connection between the neighboring pieces, and the internal shoreline topography within the pieces are incorporated implicitly. This makes the new approach theoretically more comprehensive and accurate than the existing point and change-rate-based models. Approach II: DTS from Digital Models The second approach toward generatinga DTS is to simulate the intersectionof the coastland and the water surface. The coastland is represented by a coastal terrain model (CTM) that contains topographic information in a narrow zone of the coast and near-shore bathymetry. The topographic information can be derived from the one-meter resolution satellite stereo imagery (for example, IKONOS) and airborne images. There is often a gap in the shallow water area between the topographicmodel and the bathymetric data that are usually acquired by a multibeam system and do not cover shallow water areas. LIDAR mapping seems to ll this gap very ef ciently. The CTM is then built by georeferencing and integrating the topographic, data, LIDAR data, and bathymetric data in the same planimetric and vertical datum. Given the date and time, the CTM at that time can be derived from the periodically acquired CTMs. The water surface is depicted by a water surface model (WSM) that can be produced by a hydrologicalmodeling system using meteorological data and coastal physical environmental data as boundary conditions. The DTS is created digitally by an intersection of the CTM and WSM. Theoretically, the shoreline can be derived by a subtraction of the WSM from the CTM, where the grid points with differential value of 0 represent the shoreline. Technically, a number of issues need to be addressed before a quality digital shoreline can nally be obtained. A test run of our existing data set demonstrated great potential success if the processing steps in Figure 4 are followed. The subtraction result of the CTM and WSM is further smoothed so that the shorelines are continuous lines in the grid represented by grid points with value 0. Also, spikes and ‘shoreline segments’ whose lengths are smaller than a threshold are eliminated. Furthermore, a classi cation based on the elevation/bathymetry differential values is performed to delineate grid points into land, water, and land-water FIGURE 4 Generation of a tide-coordinated shoreline from CTM and WSM. Digital Tide-Coordinated Shoreline 33 interactionpointsand to create a thematicimage. Subsequently,a clump operation groups the same kinds of grid points together to form clumps of land, water, and land-water interaction areas. After a noise detection and deletion process, the re ned clump image is used to nd the shoreline which is de ned as one of the boundaries of the clump areas. In shoreline detection, topological information indicating that a shoreline separates water from land is also checked. The raster or grid shoreline is then converted to the vector shoreline. Finally, after a visual inspection and editing process, the digital shoreline becomes available for various applications. If the WSM represents the water surface at the desired MLLW time, the derived shoreline is the required digital tide-coordinated shoreline. Experimental Result and Discussion In order to test the concept of the second approach, we performed an experiment in a Lake Erie study area that covers a shoreline of 11 km from Sheldon Marsh to Oberlin Beach, Ohio (Figure 5). A set of NOAA tide-coordinated aerial photographs, taken when the water level reached the MLLW, was processed to produce a digital terrain model and an orthophoto of the area. The tide-coordinated shoreline can be digitized directly from the orthophoto. This shoreline is treated as a master shoreline because it is tide-coordinated and has a very high accuracy. The terrain model derived from the aerial photographs was integrated with bathymetric data acquired by ODNR (Ohio Department of Natural Resources) to form a CTM. A WSM of the same area was generated by the Great Lakes Forecasting System developed by OSU (Bedford and Schwab 1991). A second WSM that has a water level difference of 80 cm from the rst one was generated by the same system. Figure 6 shows the two digital shorelines in a marshland subarea located westmost in Figure 5. Since the CTM is very at in the marshland, the 80 cm water difference created a signi cant shoreline change. Since the WSM does not correspond to MLLW the shorelines generated are not tide-coordinated. A comparison between the two digital shorelines and FIGURE 5 DTS experiment in Lake Erie area. 34 R. Li et al. FIGURE 6 Two digital shorelines created from CTM and WSM. the master shoreline indicated very small differences (Li et al. 2001). We will continue our investigation on the integration of tide information in the hydrological modeling and subsequent shoreline modeling, so that the digital shoreline produced in this way becomes a DTS. The accuracy of the above shorelines is affected by the accuracies of the CTM and WSM. The CTM consists of a DTM derived from aerial photographs and a bathymetric data set. The DTM has an accuracy of 2.1 m, while the bathymetric data have an estimated accuracy of 40 m. When we merged these two data sets, the DTM grid points were chosen in an overlapping area. If there is a gap between the two data sets, we interpolate at the grid points in such a way that a normalized weight of 2/3 is used for DTM points, and 1/3 for bathymetric points. Thus, the interpolated points have a propagated standard deviation of 13.4 m. Overall, the accuracy of the CTM ranges from 2.1 m to 13.4 m for the area of land-water interaction where the digital shoreline was generated. The water surface data is estimated to be accurate to several centimeters. Assuming a ve-degree coastal slope, a water surface error of 5 cm would theoretically produce a horizontal error of 0.6 m, which can affect the accuracy of the generated digital shoreline. Therefore, the major error source of the nal digital shoreline comes from the CTM. The quality of the CTM can be signi cantly improved by better aerial image processing and LIDAR shallow water mapping so that the land-water intersection area will have far better terrain information for improved shoreline extraction operations.We expect that the digital shoreline thus produced will have an accuracy of around 1 m. To compare various techniques of shoreline mapping, we reviewed existing shoreline maps in the study area, and the data and techniques employed to produce them. The master shoreline used for comparison was extracted from orthophotos that are one of the products of a bundle adjustment of the NOAA tide-coordinated aerial photographs. The DTM used to produce the orthophotos has a horizontal error of 2.1 m. The dif culty in delineation Digital Tide-Coordinated Shoreline 35 TABLE 1 Estimated Accuracy of the Shorelines in the Study Area Derived from Various Sources Shoreline T-sheet USGS DLG Aerial orthophoto Digital shoreline from CTM and WSM IKONOS 1 m simulated image IKONOS 4 m image Estimated standard deviation 2.5– 20 m depending on scale 12 m (1:24000) 2.6 m (considering DEM error) 2 – 13 m dep. on CTM / WSM quality 2– 4 m 8.5 m of the shoreline from the orthophotos may introduce an error of 1.5 pixels in the images. Therefore, the master shoreline used in this study has an estimated error of 2.6 m. An attractive data source is the IKONOS 1 m resolution satellite imagery. The accuracy of the shoreline derived from the 1 m simulated IKONOS images is 2 – 4 m, considering the fact that the accuracy of 3D ground points reaches 2 – 3 m and the identi cation error of conjugate shoreline points in a stereo image pair is about 1.5 pixels, about 1– 2 m (Zhou and Li 2000; Li et al. 2001). The 4 m IKONOS multispectral images in the same area came as Geo-product with an accuracy around 24 m. After a polynomial georecti cation using eight GPS ground control points, the improved images have an accuracy of 6 m estimated from the differences between the GPS surveyed coordinates and the recti ed coordinates at the ground control points. Taking the shoreline delineation dif culty into account, an optimistic estimation of the shoreline accuracy derived from the 4 m IKONOS images in this speci c case is about 8.5 m. NOAA T-sheets have large and medium scales from 1:5000 to 1:40000. Taking 0.5 mm on the map as the error source, the accuracy of the digitized shoreline from the T-sheet is about 2.5 m to 20 m. Similarly, the shoreline extracted from the USGS DLG (1:24000) data should have an error of 12 m. The ODNR map in the area has a scale of 1:12000 and the estimated error of the shoreline is about 6 m (Table 1). Conclusions A high-accuracy, tide-coordinatedshoreline is required in many applicationssuch as coastal management, mapping, environmental monitoring and protection, and the insurance industry. In order to achieve ef cient and cost effective tide-coordinated shoreline mapping, more research should be conducted in the generation of digital tide-coordinated shoreline modeling. Based on the above results, the following conclusions can be drawn: ² High quality, digital instantaneous shorelines can be extracted from high-resolution images, such as IKONOS satellite images and airborne images. The instantaneousshorelines derived from the CTM and WSM are capable of performing simulation and prediction of shorelines. ² Tide-coordinatedshorelines should be used to calculate shoreline erosion rates and shoreline changes in order to take the dynamic nature of the shoreline into consideration. ² Further research efforts should be made in establishing and implementing mathematical models for derivation of tide-coordinatedshorelines based on (a) instantaneousshorelines observed from high-resolution remote sensing imagery, and (b) CTM and WSM. 36 R. Li et al. References Bedford, K. W., and D. Schwab. 1991. The Great Lakes Forecasting System—Lake Erie Nowcasts/Forecasts. Proc. of Marine Technology Society Annual Conference (MTS’91). Washington, DC: Marine Technology Society, pp. 260– 264. Bossler, J. D. 1996. Airborne Integrated Mapping System. GIM, July, pp. 32– 35. CSC 2001. CSC/NOAA web page: http://www.csc.noaa.gov/shoreline/history.html; status: June 29, 2001. Fritz, L. W. 1996. Commercial earth observation satellites, Intn. Archives of Photogrammetry and Remote Sensing. ISPRS Com. IV: 273– 282. Ingham, A. E. 1992. Hydrography for the surveyor and engineer. London: Blackwell Scienti c Publications. Lapine, L. A. 1991. Analytical calibration of the airborne photogrammetric system using a priori knowledge of the exposure station obtained by kinematic GPS techniques. Report No. 411, 1991, Department of Geodetic Science and Surveying, The Ohio State University. Li, R. 1997. Mobile mapping: An emerging technology for spatial data acquisition. J. of Photogrammetric Eng. and Remote Sensing 63(9):1085– 1092. Li, R. 1998. Potential of high-resolution satellite imagery for national mapping products. J. of Photogrammetric Eng. and Remote Sensing 64(12):1165– 1169. Li, R., K. Di, and R. Ma. 2001.A comparativestudy on shorelinemappingtechniques,4th International Symposium on Coastal GIS, Halifax, NS, Canada, June 18– 20, 2001. Lockwood, M. 1997. NSDI Shoreline Brie ng to the FGDC Coordination group, NOAA/NOS. Merchant, D. C. 1994. Airborne GPS-photogrammetry for transportation systems. Proc. of ASPRS/ ACSM, pp. 392– 395. Shalowitz, A. L. 1962. Shore and sea boundaries—with special reference to the interpretationand use of coast and geodetic survey data. U.S. Department of Commerce Publication 10-1, Two Vols. Washington, DC: U.S. GPO. Shaw, B., and J. R. Allen. 1995. Analysis of a dynamic shoreline at Sandy Hook, New Jersey, using a Geographic information system. Proc. of ASPRS/ACSM, pp. 382– 391. Slama, C. C., C. Theurer, and S. W. Henriksen. 1980. Manual of Photogrammetry. American society of photogrammetry, Falls Church, VA. Zhou, G., and R. Li 2000. Accuracy evaluation of ground points from IKONOS high-resolution satellite imagery. J. of Photogrammetric Eng. and Remote Sensing 66(9):1103– 1112.
© Copyright 2026 Paperzz