Proceedings of the Sixth International Conference on the Mediterranean Coastal Environment MEDCOAST 03, E. Ozhan (Editor), 7-11 October 2003, Ravenna, Italy The CoastView Project Carlo Albertazzi (1), Renata Archetti (2,3), Clara Armaroli (5,6), Mauro Ceroni (7), Paolo Ciavola (5), Alberto Lamberti (2), Silvia Medri (2,4) (1) Regione Emilia-Romagna, Servizio Pianificazione di Bacino e della Costa,Via dei Mille 21, 40121 Bologna, Italy E-mail: calbertazzi@regione.emilia-romagna.it (2) Università di Bologna, Facoltà di Ingegneria, DISTART, Viale Risorgimento 2, 40136 Bologna, Italy E-mail: alberto.lamberti@mail.ing.unibo.it (3) E-mail: renata.archetti@mail.ing.unibo.it (4) E-mail: smedri@ambra.unibo.it (5) Università di Ferrara, Dipartimento di Scienze Della Terra, C.so E. I d'este 32, 44100 Ferrara, Italy E-mail: cvp@unife.it (6) E-mail: clara.armaroli@unife.it (7) Regione Emilia-Romagna, Servizio Tecnico Bacino Fiumi Romagnoli Sede di Ravenna, P.zza Caduti per la Libertà 9, 48100 Ravenna, Italy E-mail: mceroni@regione.emilia-romagna.it Abstract CoastView is a three years UE project, involving as Italian partners the Emilia- Romagna Regional Government and the Universities of Bologna and Ferrara. The project aims at developing video remote sensing tools in aid and in the context of Integrated Coastal Zone Management (ICZM). These video systems are capable of monitoring waves, surface currents and the evolving coastal morphology, providing useful information on coastal erosion and recession rates, the location of shipping channels and hazardous sandbars, and the performance of coastal defence structures. In order to assist in decisionmaking, the CoastView project aims at reducing the complex information about the physical state of the coastline into a simplified set of Coastal State Indicators (CSIs),
236 MEDCOAST 03 upon which management decisions and policy can be based. This paper describes the project, its technological tools (Argus System) and the studies undertaken on the Lido di Dante site (Ravenna, Italy). Research and studies are carried out in order to assess the validity of video data products concerned with the breaking zone hydrodynamics and the evolving coastal morphology. The analysis of the video data will involve the development of new algorithms, the testing and refining of these algorithms using in-situ data. Problems to be addressed include the optimum definition of the waterline for different levels of ambient illumination and wave conditions, estimation of inshore water depth through wave phase velocity, quantitative determination of the location of offshore bars from breaking wave patterns, and the potential estimation of surface currents. Introduction The European CoastView Project, financed under the V Framework Programme of the European Union, involves 12 partner institutions from six countries, including four national-scale coastal managers from Spain, Italy, Netherlands and the UK. The institutions working on the Italian site are the Emilia-Romagna Regional Government and the Universities of Bologna and Ferrara. The CoastView consortium studies four morphologically dissimilar field sites that typify the sort of management issues that are common to European coastlines. These sites include a continuous-undefended coastline (Egmond, Netherlands), a continuous-defended coastline (Lido di Dante, Ravenna, Italy), a coastal inlet with a single bar or spit (El Puntal, N. Spain) and a coastal inlet with multiple complex bars (Teignmouth, UK). Fig. 1: CoastView Sites map
Albertazzi et al. 237 The site of Lido di Dante (near Ravenna, in the Emilia-Romagna region), facing the Adriatic Sea, was chosen because of the opportunity to study the dynamics of equivalently situated defended and undefended beaches. Here a video system monitors a stretch of coastline defended by a shore parallel breakwater and by three groins, adjacent to an undefended length of coastline (Figs 2 and 3). Fig. 2: Defended part of the area of study in Lido di Dante, Ravenna (Italy) Fig. 3: Undefended part of the area of study in Lido di Dante, Ravenna (Italy) The natural part of the beach at Lido di Dante is almost 2 km long, stretching between the southern groin and the Bevano River mouth. The principal characteristic of the area is the presence of vegetated dunes and a pine forest behind them. The Bevano is a very dynamic stream and it rapidly moves northwards, laterally eroding the dunes.
238 MEDCOAST 03 In the unprotected beach dunes are eroding in the northern part and quite stable in the southern one. The principal causes of the dune erosion are trampling by beach users and wave run-up that touches the dune foot during storms. Regarding the submerged beach, the area presents a system of two bars: the external ones are crescentic bars: their morphology and position change rapidly. The internal ones are difficult to identify: in some parts, they seem to be linear, in others they seem to be crescentic. Moving towards the Bevano the internal bars are connected with the submerged delta of the river itself. A low tide terrace/bar is present in the northern and central portion of the coastal stretch. These video systems are capable of monitoring waves, surface currents and the evolving coastal morphology (Holland et al., 1997). Potentially they could provide useful information on coastal erosion and recession rates, the location of submerged sandbars and the performance of coastal defence structures. Project Description The two primary scientific results that are expected from the CoastView Project are: 1) To develop resource-related Coastal State Indicators (CSIs) for describing the dynamic state of the coast, in support of coastal zone management; 2) To develop and verify video-based monitoring methods and associated analysis techniques for estimating and interpreting these CSIs. New video systems, theory and software for data collection and evaluation of CSIs will be delivered. Video cameras are mounted on a high vantage overlooking the coastline and are capable of remotely sensing waves, currents and beach elevation at high frequency, over a scale of several kilometres. These systems relay information recorded onsite through telephone lines to a remote computer and the Internet. The development of these systems has been largely motivated by scientific research into the evolution of natural beaches due to the action of waves and tides. However, it is becoming increasingly evident their potential to contribute to the effective management of the coastal environment as well as science (http://coastview.ims.plym.ac.uk//project.html). The primary focus of the CoastView project is to simplify the task of the coastal manager. The information currently available to the coastal manager about the physical state of the coastline from observations, models and scientific interpretation is often complex and difficult to use directly. In order to assist in decision making, this complex information needs to be delivered promptly and in a simplified form. The project aims to develop a set of video-derived Coastal State Indicators (CSIs) upon which management decisions and policy can be based. CSIs can be defined as a reduced set of parameters that can simply, adequately and quantitatively describe the dynamic-state and evolutionary trends of a coastal system. The different sites are expected to yield CSIs that are both generic (appropriate to all coastal locations) and site specific. Video-derived CSIs will be validated with ground-truth measurements at each field-site and their confidence limits evaluated. The implementation of video-derived CSIs is expected to have an impact on several aspects of Coastal Zone Management on a time-scale that is consistent with the rate of coastal change, including coastal protection, recreation, ecosystem protection and
Albertazzi et al. 239 navigation. In the context of coastal protection, for example, a manager might be concerned with preserving beach volume or the width of the beach. If these parameters were frequently monitored using video systems, it would facilitate management decisions to know when beach nourishment or protection measures are required, what areas are particularly at risk and what the most appropriate defence measures might be. Accepting the guidelines elaborated by the UE, the Emilia-Romagna Regional Government promoted in 2002 the plan Integrated Coastal Zone Management (ICZM). This Project aims at recognizing and analysing all the different components of the coastal environment, making them compatible in an integrated view, and driving the various activities of the coastal system towards the environmental social and economic sustainability. In the following table one can see, as an example, a series of CSIs identified for the non-protected beach. Table 1: Series of CSIs identified for the non-protected beach BEACH STATE DUNE STATE HYDRODYNAMICS HUMAN USE Shoreline Location Dune Height Wave Height Beach Recreational Use Bar Location Position of Dune Foot Wave Period Water Quality Emerged Beach Width Run-up Wave Direction Beach Nourishment Location of Scarps Overwash Breaker Location/Depth Trampling of Foredunes Beach Volume Salt Spray Surf Zone Width Sub-tidal Water Depth Dune Flora Longshore Current Intensity and Sediment Transport Bar Height Dune Vegetation and Sand Migration Rip Current Intensity and Offshore Net Transport Slope of Beach Fire Hazard Wind Intensity and Direction Position of the Bevano River Mouth Sediment Fluxes The role and stability of dunes is clearly a major issue for the non-protected area. Apart from the erosion of the dunes caused by wave run-up and by trampling, it would also be interesting to study seasonal variations in vegetation density and biodiversity and the aeolian sediment transport. The migration of the Bevano River mouth is also a fundamental CSI. The lateral erosion of the dunes and the marine ingression inside the pine forest are two of the main problems of the site. Environmental managers are interested in knowing how the river behaves and which solutions are to be undertaken. The two previous CSIs are linked by the role of nearshore bars. The river provides sediment that is reworked into the bar system. On the other hand, the bars decrease nearshore wave energy and control the distribution of run-up events.
240 MEDCOAST 03 Aims of the project The principle aims of the project can be summarised as follows: 1. To develop a set of issue-based CSIs in support of coastal zone managers 2. To develop improved video systems for delivering CSIs promptly to the coastal manager at the appropriate temporal and spatial scales. 3. To develop algorithms for the estimation of CSIs 4. To ground-truth and evaluate confidence limits for video-derived CSIs 5. To produce schemes for the interpretation of CSIs and prediction of coastal state The Argus System Holman of Oregon State University (USA) originated the video remote sensing technology and set up the Argus programme (Holman et al., 1993). This international research project involves a network of collaborators, who have deployed cameras in a wide range of radically different environments world-wide, including the east and west coasts of mainland America, Hawaii, Australia, New Zealand, UK and the Netherlands. The Argus network takes its name from the mythological Greek figure with multiple eyes: the multiple eyes in this case are the numerous cameras stationed around the globe. These systems record oblique images of the coastline, often from numerous cameras, that are subsequently rectified and digitally merged to produce undistorted planviews. This process uses theory originating from the fields of photogrammetry and robotics. The rectified images facilitate the accurate quantification of visible features within the camera field of view. Video remote sensing systems are capable of evaluating the shoreline position, the location of submerged sandbanks, inter-tidal and sub-tidal bathymetry, wave statistics (e.g. period, speed and direction) and surface currents (e.g. flow in channels, longshore and ripcurrents) (Lippmann et al., 1989, Stockdon et al., 2000). With the Argus image processing system one can identify coastal features and processes and map them into any desired coordinate system. One can then perform time series analyses, to track changes in the locations of these features and their degree of variability, and quantify important erosion processes such as wave run-up and total wave energy (Aarninkhof et al., in press) The Argus Station in Lido di Dante, Ravenna Lido di Dante, situated in the North-eastern coast of Emilia-Romagna facing the Adriatic Sea near Ravenna, was chosen as site of interest because of the excellent opportunity of synchronously studying the dynamics of equivalently situated defended and undefended beaches. Some sections of this beach are, in fact, unprotected coastline, while others stretches of coastline are defended by groins and shore-parallel breakwaters.
Albertazzi et al. 241 Consequently the video System here consists of four cameras hold up by a wooden tower (Fig. 4), 18 meters high, on purpose installed in situ, with the most southern camera looking at the non-protected beach, with its littoral dunes and submerged sand bars, while the other three cameras monitor the evolution of the defended beach. Fig. 4: The four cameras and the Argus tower. Argus Image Data Data types include traditional image data as well as supporting photogrammetric meta-data. Argus collects three basic image products for beach studies. These include the traditional snapshot, showing wave activity, ten-minute time-exposure (timex) images of the wave dissipation patterns (revealing submerged sand bars and rip channels) and variance images (separating dynamic from steady areas of the image) (Fig. 5-a, b). The greatest scientific value currently comes from the timex images.
242 MEDCOAST 03 The simplest image type is the snapshot image, which is a single individual image where no averaging is applied. This is the same image that you would see if you took a picture of the beach using a conventional camera. Snapshot images provide simple documentation of the general characteristics of the beach, but they are not so useful for obtaining quantitative information. (Fig. 5-a). Fig. 5: a) Snapshot image from camera 2; b) Time-exposure image from camera 2. A much more useful image type is the time-exposure image. Time-exposure images are created by digitally averaging the individual pixel intensities (mean I u,v = I u,v /n ) of 600 individual snapshot images that are collected at the rate of 1 picture every second, for a period of 10 minutes. By removing visual noise they reveal the location of key coastal zone features such as shoreline location, dry beach width, and sand bar location. Time exposures of the nearshore wave field have the effect of averaging out the natural variations of breaking waves, to reveal smooth areas of white, corresponding to the locations of enhanced breaker dissipation, which has been shown to provide an excellent indicator of the shoreline and nearshore sand bars. In this manner, a quantitative map of the underlying beach morphology can be obtained. (Fig. 5-b). An appropriate tool, besides, gives the opportunity of making also day-timex images where individual pixel intensities are averaged over all visible sampling times to reveal mean daily features, such as the locations of bars and rip channels. They are useful for creating movie loops of timex imagery showing the dynamic evolution of the nearshore. While the time-exposure images are being collected, an image type called a variance image is also created. Whereas the time-exposure is an average of many individual snapshot images, the corresponding variance image displays the variance (σ I 2 ) of light intensity during the same 10 minute time period. Variance images help identify regions which are changing in time, from those which may be bright, but unchanging. For example, a sandy beach will appear bright on snapshot and time-exposure images, but dark in variance images. Because of this, variance images are useful for analysis techniques such as the identification of the shoreline, as the (bright) changing water surface is readily identifiable against the (dark) beach. Snapshot, time-exposure and variance images are automatically created every hour at the remote site, and the images transferred to the laboratory for archiving and analysis.
Albertazzi et al. 243 Argus Pixel Data The real power of Argus comes from the use of pixel intensity time series. Pixel is the acronym for picture element, the basic unit of measurement in a digital image. A digital video image consists of an array of 640 by 480 picture elements (pixels), where the light intensity of each pixel is resolved to 256 (0-255) individual values (i.e. one byte per pixel). Pixel values are resolved into red, green, and blue (RGB) values on a scale of 0-255, where 0 is no intensity and 255 is full intensity. The real-world size of the volume contained in a pixel (width, depth, height) is determined by lens focal length and the geometry of the camera, notably height, tilt, azimuth, and yaw (camera roll). The underlying assumption is that time series of optical radiance provide exploitable time series that allow measurement of a range of nearshore variables. In essence, we are simply taking advantage of the fact that the eye can see waves, foam, currents and other phenomena. Traditional nearshore measurements require an array of in-situ instruments that can be costly to purchase, maintain, and must survive in a hostile environment. By contrast, pixels are cheap and rugged. Time series of optical radiance (intensity from a particular pixel) are usually strongly coherent with those of a submerged pressure sensor at the same location. In this case, the time series are visually very similar. A data set containing the time history of pixel intensities along a transect is called time stack. Time stacks are especially useful for tracking features such as shoreline migration, wave run-up and swash excursion, and wave phase speed and direction. High frequency (2 Hz for currents, 1 Hz for waves) video sampling are used to measure wave phase speed (c) and direction. These time series are returned to the laboratory and analysed through appropriate signal processing algorithms, and will be assimilated into numerical models or direct data collected through intensive field experiment for a complete nearshore prediction system. Image Analysis Analysis tools include the ability to rectify images to real-world coordinates, merging of images from multiple cameras to produce a single image, and the mapping of the shoreline, bars, rips and many other features of the coastline. To make scientific measurement from image data, in fact, the images must be rectified onto a map plan. Rectification requires a good understanding of the photogrammetry geometry of the camera-image system as well as accurate knowledge of the cameras pointing parameters (azimuth, tilt, HFOV (Horizontal Field of View) and roll). Both single-camera and multiple-camera rectified images show bright, longshore bands, clearly indicating the locations of breaking waves. As the breaking of waves is caused by shallower water, the bright intensity patterns can be used to map the underlying bottom topography. If wave and tide conditions are known, the water depth over bars can also be determined. Two kinds of in-situ data will be collected. Firstly, a continuous monitoring will include tide, wave and meteorological measurements, and regular conventional beach
244 MEDCOAST 03 surveying, both sub-aerial (above the low tide line) and offshore. Secondly, a period of intensive in-situ field campaigns will be carried out, designed to measure parameters relevant to the CSIs such as the wave and current field and sediment transport. The analysis of the video data will involve the development of new algorithms, and the testing and refining of these algorithms using the in-situ data. The CSIs listed for the unprotected part of the beach that can be easily studied with the Argus System are those regarding the beach state, the dune foot and run-up position. Image analysis with the IBM (Intertidal Beach Mapper) toolbox and AMT (Argus Merge Tool) toolbox at the moment gives as results the hourly shoreline position during the daylight hours and the position and migration of the bars (Plant et al., 1997). More detailed analysis will be carried out once the images will be validated and longer time series become available. Fig. 7: Rectification and merge of the 4 cameras Fig. 8: Planview of the unprotected part of the beach (21/05/2003). Notice the external complex bars, linear and at some stages oblique internal bars. The grid is in meters. North is positive towards the left hand side.
Albertazzi et al. 245 Fig. 9: Shoreline positions of the unprotected part of the beach at different times (09:00, tidal level: +0.5 m; 17:00, tidal level: +0.21 m) for 17/04/2003. The grid is in meters. North is positive towards the left hand side. Ground truth measurements have been carried out in the unprotected part of the beach starting from December 2001 to study those CSIs that are difficult to monitor with the Argus System and to test the processing algorithms. Beach profiles, including dunes, were surveyed almost monthly with a Total Station. Shoreline position was monitored with a DGPS. Long term monitoring (last 20 years) of the shoreline variation was done using GIS techniques (Moore, 2000). Bathymetric surveys were done on the whole area (from -3 m to -1 m) and on the submerged delta of the Bevano. Sediment sampling was undertaken along significant beach profiles. In Spring 2003 tracers were injected in the river mouth, to assess river sediment transport, and on a swash bar situated in the southern part of the site, to study tidal effects on the beach morphology. Topographic surveys of the Bevano were done starting from October 2002. Sediment sampling of few sections of the river was done in Spring and Summer 2003. Conclusions The CoastView project will develop coastal video systems and associated tools that are capable of collecting high quality data for the evaluation of CSIs. New theories, algorithms and software will be developed for the estimation of CSIs from video data. A combination of the developments in video technology and the definition and implementation of CSIs will clarify coastal management tasks, facilitate better resource planning, improve project designs and allow better post project evaluation. The project is expected to form the basis for future monitoring and management of coastal systems. Acknowledgements
246 MEDCOAST 03 The work presented in this paper was carried out within the framework and with the financial support of the European Union Project COASTVIEW (contract EVK3-CT-2001-00054). Besides, it is part of the Emilia-Romagna Regional Government Plan for Integrated Coastal Zone Management. The project web site is http://www.thecoastviewproject.org References Aarninkhof, S.G.J., Plant, N.G., Turner, I.L. and Kingston, K., Shoreline identification from video imagery. Intercomparison and ground truthing of four detection models, Coastal Engineering, in press. Holland, K.T, Holman, R.A., Lippmann, T.C., Stanley, J. and Plant, N. (1997), Practical use of video imagery in nearshore oceanographic field studies, IEEE Journal of Oceanic Engineering (special issue on image processing for oceanic applications), 22(1), 81-92. Holman, R.A., Sallenger, Jr. A.H., Lippmann, T.C. and Haines, J. (1993), The application of video image processing to the study of nearshore processes, Oceanography, 6(3), 78-85. Lippmann, T.C. and Holman, R.A. (1989), Quantification of sand bar morphology: a video technique based on wave dissipation, J. Geophys. Res., 94 (C1), 995-1011. Moore, L. J. (2000), Shoreline mapping techniques, J. Coast. Res., 16(1), 111-124. Plant, N.G. and Holman, R.A. (1997), Intertidal beach profile estimation using video images, Marine Geology, 140, 1-24. Preti, M. (2002), Ripascimento di spiagge con sabbie sottomarine in Emilia-Romagna, Studi Costieri, 5, 107-134. Stockdon, H.F. and Holman, R.A. (2000), Estimation of wave phase speed and nearshore bathymetry from video imagery, J. Geophys. Res., 105 (C9), 22,015-22,033. http://coastview.ims.plym.ac.uk//project.html http://www.wldelft.nl/argus/sites/lidante/2003