CHAPTER TEN Decision-Making in the Coastal Zone Using Hydrodynamic Modelling with a GIS Interface Jacques Populus, Lionel Loubersac, Jean-François Le Roux, Frank Dumas, Valerie Cummins, and Gerry Sutton 10.1 INTRODUCTION AND CONTEXT There are many kinds of coastal water quality issues. They arise from various sources of inputs to the coastal zone that are either chronic or accidental. Pollution of seawater mostly results from a) point source and diffuse pollution originated from agricultural, industrial and urban activities, or b) pollution from maritime activities, e.g. waste oil as well as all types of toxic substances being dumped into the sea, including radioactive ones (Kershaw, 1997). Toxic phytoplankton blooms are a new plague. Although they are not the direct result of human behaviour, they are probably linked to human activities, and particularly contaminated ballast water. These harmful disruptions have severe effects on shellfish stocks, entailing long production shutdowns. Problems also currently exist with open sewers discharging into places designated for shellfish production or into recreational waters (Boelens et al., 1999). After flooding, there is run-off pollution from agriculture and urban areas and nitrate concentrations and bacterial counts increase remarkably. The increase in nitrate concentrations can lead to dramatic eutrophication, whereas pathogenic bacteria can make aquaculture products hazardous to human health (Lees, 2000). Other activities or phenomena such as dredging, deep water sludge disposal or landfill seepage are concerns for water quality and marine living resources (Sullivan, 2001). While hydrodynamic modelling results and the handling of geo-referenced information are becoming more readily available to coastal management stakeholders within GIS (Geographic Information Systems), there is still a lack of direct interfacing of model results with baseline mapping data (BASIC, 2001). This paper discusses solutions to bridge this gap and illustrates them with two case © 2005 by CRC Press LLC studies where effective model outputs are being used for improved environmental management and decision-making. 10.2 HYDRODYNAMIC MODELLING BASICS Mathematical models (Garreau, 1997) can solve geophysical fluid mechanics equations using a number of simplifying assumptions. In essence, knowledge of the bathymetry, wind tide and currents are used to predict water levels and concentrations of conservative elements over space and time. The ability of these models to accurately and reliably reproduce the important features of complex systems has improved considerably along with rapid growth in understanding of the underlying physical phenomena, and the ability to quantify them in terms of valid mathematical formulations. The widespread and affordable availability of high-powered computing facilities has also contributed to this development, given the heavy processing loads involved in running all but the simplest models. Confidence in the predictive capacity of models is governed by the quality of fit between modelled output data and real field measurements. Standard practice in building a typical coastal model entails a number of discrete steps, the first of which involves reconstructing the morphology of the area of interest in the form of a digital terrain model. The next step involves adding water to the system, and then setting it in motion through replication of vertical tidal fluctuations and horizontal wind friction. Together these forcing factors create gradients, which in turn drive horizontal current flows. A range of assumptions is usually made in order to simplify the system, and allow it to run effectively. It is recognised that models do not strive to exactly reproduce in detail all the features of a natural system; however the aim is to arrive at a situation where the model is capable of reliably reproducing the principal features through an iterative process of validation and calibration. 10.3 GIS FOR COASTAL ZONE MANAGEMENT A Geographic Information System (GIS) is a computer-based information system used to digitally represent and analyse geographic features. It is used to input, store, manipulate, analyse and output spatially referenced data (Burrough and McDonnell, 1998). A GIS can be distinguished from database management systems or from visualisation packages through its specialised capability for spatial analysis. The use of GIS for coastal zone management has expanded rapidly during the past decade and references are numerous (Durand, 1994; Populus, 2000; Wright and Bartlett, 2000). For optimum efficiency, geo-referenced data should be properly stored in geo-databases built on spatial data model design (Prélaz Droux, 1995). Some of the greatest challenges currently faced by those handling coastal zone data are a) the land-sea interface, with different mapping references in both horizontal and vertical modes, b) water dynamics and the related temporal issues, and c) 3D display requirements. © 2005 by CRC Press LLC 10.4 TOOLS AND DATA 10.4.1 Technical details of regional and local models MARS-2D is a bi-dimensional model using a finite difference method called ADI (Salomon, 1995). A broad regional model, extending between 40˚N and 65˚ N and from 20˚ W to 15˚ E with a 5 km grid, is used as a framework in which to embed further models of smaller extent for areas of interest along French Atlantic and English Channel coasts. Commonly, the embedding system has up to 5 levels allowing phenomena to be examined at resolutions from 1km down to 50m. This type of model is suitable for applications in areas whose waters are typically well mixed (e.g., coastal or mega-tidal areas). The model is designed to solve for tidal and wind driven currents and the transport of dissolved materials. MARS-3D (Lazure, 1998) is a fully finite difference model, in both vertical and horizontal orientations, which uses a time splitting method based on MARS- 2D for the barotropic mode. Good coupling between 2D and 3D modes has largely been achieved through the use of iterative methods. MARS-3D is used at resolutions ranging from 5 km over regional areas of epicontinental seas, down to 100 metres over detailed areas, narrow bays and estuaries. It is currently run operationally at the regional scale (with a 5 km mesh) on all French seaboards. Table 10.1 gives an overview of MARS model features. Table 10.1 Comparison of MARS-2D and MARS-3D main features Model MARS-2D MARS-3D Area English Channel, Bay of Biscay English Channel, Bay of Biscay Grid and time step From 50m up to 10km ~ 5km, 5 – 20 minutes Period of time From days to decade Year to decade Applications Tide currents, dissolved matter, salinity under homogenous conditions Tide, currents, temperature, salinity, transport of dissolved matter Type 2D Finite difference 3D Finite difference 10.4.2 The modelling system A modelling chain has been developed at Ifremer over a period of many years. The system has a series of pre-processors and tools for the graphic display of results produced by the MARS computation kernel. © 2005 by CRC Press LLC This mathematical model uses some simplified hypotheses to solve the equations that govern how marine currents and sea levels evolve. In order to function, the process requires an input water level along the edge of the area of interest. These boundaries are usually unknown locally, since they are dependent on tidal and weather conditions, which are themselves usually derived from modelling over a much larger area. Thus, the modelling process advances through the generation of a series of sequentially nested models. An initial general model covering a large area of the continental shelf and the Channel is followed by a succession of intermediate models of increasingly smaller scope, but higher resolution. Boundary conditions for the wide-area model are resolved using world tide models, into which modelled meteorological forcing factors have been assimilated. The modelling chain can be summed up as follows: x A generated link calculates the position, extent and resolution of each sub- model, from the large-area model to the detailed high resolution model based on computational and hydrodynamic design criteria. Computational efficiency is optimised by maintaining a maximum resolution ratio (mesh size ratio) of between four and five between any two consecutively nested models. Hydrodynamic criteria are observed as far as possible in the design through avoidance of islands or zones with violent currents, although at present the system only works when model boundaries are strictly aligned to parallels and meridians. This link of the chain has a user-friendly graphic interface and generates a descriptive file of the entire nesting process. x The second link in this computation chain is a software program which calculates an interpolated bathymetry for each nested model. The link also has a graphic interface developed in UNIRAS, which restores a depth for each calculation link in a file. The bathymetry used for the large-area model has been validated, and is essentially taken as fixed. However, it is updated on an occasional basis, as new information becomes available. The MARS-2D computation kernel is used in both case studies below, where prevailing tidal currents have a relatively homogenous vertical structure, providing a good approximation of the mean current fields pertaining in the study areas. x Lastly, a range of graphic tools are used to display the resulting modelled outputs which are written in NETCDF format (Rew and Davis, 1990). NETCDF is a widely used self-documenting format, which also provided a suitable platform upon which the ArcView portal was subsequently developed. 10.4.3 Reference mapping data Currently, coastal practitioners must refer to common baseline or reference data, i.e., primary data to which secondary (or more application-related) data will be subsequently linked (Allain, 2000). The coastline, bathymetric data, and major administrative boundaries are examples of such baseline data that could be readily © 2005 by CRC Press LLC provided to a wider public under optimal conditions of accuracy, updating, and scalability to suit various needs. However, it is noted that the bathymetry of inter-tidal areas, and other near- shore zones, is often less easily available or poorly defined. Such paucity of data is usually associated with the high cost of acquisition and restricted accessibility. Bathymetric data for the two study sites under consideration was available from the French hydrographic service at a scale of 1:50, 000. ArcView™ was the main GIS platform used for the studies, within which the Spatial Analyst™ extension facilitated a number of operations on raster images such as recoding, resampling, changing extent, computing statistics and the use of algebraic image combination functions. Existing ArcView functionality also facilitated interactions between raster and vector data layers drawing on the attributes of file features attaching to vector data sets. However a major hurdle remains in the efficient handling of large numbers of raster data layers that typically accrue as the output from multiple model runs. 10.4.4 An Integrated GIS/model interface The primary rationale behind the creation of a GIS/model interface was to allow a) consultation and display of results contained in the output files produced by the MARS hydrodynamic model; and b) extraction of these results to import them to ArcView. The concept was initially tested through the development of ModelView (Loubersac et al., 2000), a prototype interface the functionality of which was successfully demonstrated in the case of hydrobiological contamination in the Bay of Marennes-Oléron, France. As a further development, in order to broaden the scope of application a platform-independent stand-alone interface, MODELCON, was designed for use in conjunction with a range of GIS packages. In order to ensure optimum compatibility with standard software packages and other GIS (e.g. Excel, MapInfo™ etc.), MODELCONV extracts NETCDF files to a standard ASCII format. The MODELCONV interface was developed in JAVA, with Microsoft’s Visual Studio 6 environment and the JAVA library to access NETCDF files (NETCDF JAVA version 2), ensuring maximum portability in anticipation of future use on Unix systems or via the Web. 10.4.5 A Geographic conversion module An additional processing module was developed in order to address specific geoprocessing requirements beyond those available in the standard version of ArcView. This module operates as an ArcView extension and was implemented in Avenue. It enables the user to perform a range of geodesic processing operations on point, multipoint, polyline and polygon data, as well as 2-D and 3-D related measurements (pointM, multipointM, polylineM, polygonM, pointZ, MultipointZ, PolylineZ, PolygonZ). © 2005 by CRC Press LLC This module also allows data to be projected or unprojected, i.e. as geographic co-ordinates (latitude, longitude), or projected Cartesian co-ordinates including 3-D (X, Y, Z). Other operations that are supported include: x switching between geodesic systems: WGS 84, NTF and Europe 50; x switching ellipsoids: Clarke 1880, IAG GRS 80 and Hayford 1909; x switching projections: Lambert (various), Transverse Mercator (UTM). 10.5 CASE STUDY ONE: SHELLFISH PRODUCTION IN THE GOLFE DU MORBIHAN, SOUTHERN BRITTANY 10.5.1 Water quality issues in the Golfe du Morbihan The Golfe du Morbihan is located in southern Brittany, France. Enclosing an area of 125 km 2 , with many islands and extensive intertidal flats, the basin connects to the open ocean via a 1km wide channel. Its perimeter is highly indented, and is traversed by numerous streams and rivers. The main anthropogenic pressures related to two main cities, Auray (11,000 inhabitants) and Vannes (50,000 inhabitants), are augmented by very significant seasonal tourist populations. Important natural shellfisheries ( Venus spp.) are commercially exploited, and the gulf also supports a valuable oyster farming industry. Both rely on suitable water quality being maintained within narrow sanitary limits. In general, water and shellfish quality criteria are set by the EU Shellfish Hygiene Directive (Lees, 1995). However, at national level slight differences are found in the interpretation of shellfish hygiene E. coli guidelines, specifically in stating what proportion of samples must fall under the concentration thresholds and in the treatments required prior to human consumption (BASIC, 2001). In France, these values are defined by the modified decree n° 94-340 of 28 April 1994, as shown in Table 10.2 below. Shellfish farmers must take different measures for depuration with respect to these categories. Table 10. 2 French shellfish hygiene categories Level of contamination in faecal coliforms* Categories 300 1,000 6,000 60,000 A t 90% d 10% 0% B t 90% d 10% 0% C t 90% d 10% D > 10% * per 100g shellfish © 2005 by CRC Press LLC 10.5.2 Microbiological modelling Faecal coliforms (FC), of which Escherichia coli is a major component, are good indicators of bacteriological contamination levels in seawater and shellfish. EU health standards for recreational waters and shellfish harvesting zones are based on organism counts. These bacteria mainly come from rivers and streams which receive waste water from various sources including surface runoff and soil outwash, sewage treatment plant discharges (especially in heavy rainfall situations, when function is impaired by lower residence time) and unauthorised discharges. Other direct sources are storm water drain systems discharging directly into the sea and diffuse contamination in the vicinity of moored boats. Coliform bacteria do not tolerate exposure to solar UV radiation, and thus have a limited life span in seawater. Their survival time will vary depending on their metabolic state and environmental conditions. Bacterial survival times are positively influenced by the following: x lower winter temperatures which can slow down bacterial metabolism and extend survival based on slower rates of consumption of energy reserves; x increased turbidity values, linked to levels of suspended solids. Turbidity has a dual effect, as a potential food source and as protection against solar UV (ultraviolet) radiation. Turbidity is usually higher in winter, due to greater discharge and sediment resuspension. In general modelling applications, bacterial survival is represented by the term T90 (being time at which 90% of the bacteria will have disappeared), which assumes an exponential rate of decline in numbers. Values for T90 are normally established on an empirical basis, and those used in the current study were based on experience at local sites (Guillaud, 1997). 10.5.3 Simulation descriptions In order to characterise coliform distributions under a range of environmental conditions, a series of MARS 2-D model simulations were undertaken in which the following parameters were varied: x Tidal conditions. Simulations were made over a period of three weeks under realistic tidal conditions, i.e. of sufficient duration to capture both spring and neap cycles. x Seasonal influence. This was investigated by varying discharge volumes, i.e. coliform flux. x Weather conditions. Three sets of commonly prevailing conditions were selected, in accordance with wind statistics derived from data supplied by the French meteorological office. These were 1) zero wind (baseline condition), 2) westerly 8m.s -1 and 3) north easterly 8ms -1 . © 2005 by CRC Press LLC x T90s. A summer value of 10h, with 24h for winter was chosen in accordance with local measurements. x The impact of major malfunctioning of water treatment plants in heavy rainfall periods. This was modelled by doubling the amounts of bacteria discharged over a 24-hour period, which is typical of what can occur during a summer storm. The objective was to investigate the impact of episodic events on coliform distribution, particularly with respect to existing zonation patterns. The above scenarios were investigated through a combined series of seven simulations (two seasons under three different conditions, plus one exceptional situation type). In order to assess the validity of the simulations, the resulting coliform concentration distributions were then categorised according to EU shellfish farming criteria (Table 10.2). Model-derived zonation patterns were found to be in broad agreement with existing zonation plans based on both tissue sampling and water quality monitoring network samples. 10.5.4 Simulation results Having established the overall validity of modelled results, priority was given to locating the zones likely to be subject to the highest contamination. These were found to be in upper reaches of the main rivers and in the bay of Vannes (Figure 10.1). Observations performed during model runs indicated that once steady state conditions are attained, the spring/neap tidal coefficient has little influence on subsequent coliform concentrations. The impact of a constant, moderate wind (8 m/s) on contamination plumes was imperceptible, both in summer and winter. In the summary of results, only the baseline “no wind” situation is referred to. T90 was found to exert the greatest influence on coliform concentration, and when set to 24hours gave rise to the highest levels of contamination. This is reflected in the final model run for which a 24h T90 linked to winter discharge rates was chosen. 10.5.5 Comparison between the actual classification and a simulation Figure 10.2 shows the results of zonation categories based on samples collected by the water quality monitoring network (Ifremer, 2003). Zonations are based around 30 such samples per site that are routinely collected over the course of each year at low water during each spring tidal cycle (theoretically least favourable situation). Dots indicate the locations of effluent discharge sources. Figure 10.2 clearly shows the main water body of the gulf officially ranked as category A, whilst the estuaries and their outlets (north and northeast) are in the B or D categories. It is notable that no estuarine areas have been zoned in the C category and have instead been allocated to category D as a precautionary measure in respect of the EU Shellfish Directive. This precautionary designation takes into account the obvious © 2005 by CRC Press LLC risk of contamination near urban and port areas, as well as the impact of reduced salinity on mariculture products in upper estuarine reaches. Figure 10.1 Screen capture from GIS showing the Golf du Morbihan. Coloured areas denote EU shellfish classifications based on simulated coliform concentration distributions Comparing the results highlights the good consistency between simulation results and those obtained from monitoring network observations. Whilst the general configuration of zonation categories (Figure 10.1) based on simulations is broadly consistent with the official classification scheme (Figure 10.2), the former logically appears to give rise to a less conservative regime under which the estuarine areas in the north and northeastern gulf are designated as category C, rather than the precautionary official D designation. This may be explained by the relative shortness of the period simulated (3 weeks), whereas the official designations are based on an annual monitoring cycle. Other reasons may be that the coliform flows used for the simulation did not include the annual maxima (in the case of water collection and treatment plant malfunctions). Furthermore the actual T90 may exceed the 24h value used in simulations, especially in naturally turbid upper estuarine areas, or as a consequence of winter storm induced sediment resuspension. By comparing the Figures where the discharge points are identified by a dot symbol, we also note that the model, which minimised coliform concentration levels as seen above, reveals two B category zones in the outer Bay area. These can be seen in Figure 10.1, located to the north and to the west of the Ile au Moines (the island is indicated by an M on the map). Their coincidence with © 2005 by CRC Press LLC effluent discharge points (dots) suggests the logical cause of reduced water quality in these areas. This result highlights the valuable insights that can be obtained through the use of realistic simulations based on numerical models in identification of localised water quality issues, which may then be addressed through the allocation of additional monitoring resources. In this case, priority was given to the northern most area (associated with effluent discharge Arradon) owing to its proximity to significant shellfish farming areas, resulting in the establishment of an additional hygiene monitoring station. Figure 10.2 Screen capture from GIS showing the Golf du Morbihan. Coloured areas denote EU shellfish classifications based on measured coliform counts. 10.6 CASE STUDY TWO: THE WRECK OF THE IEVOLI SUN 10.6.1 Context of the case study The Italian chemical tanker, Ievoli Sun, sailing from Rotterdam to Genoa sank around 9 am local time on 31 th October 2000 in the central English Channel, approximately 9 nautical miles north of Les Casquets (Channel Islands) and 20 nautical miles west-north-west of the French Cap de la Hague off the north coast of Normandy (Figures 10.3 and 10.4). © 2005 by CRC Press LLC [...]... Dynamics on Coastal Activity and Sensitive Marine Environments, Atlantic area INTERREG-IIC programme Final report, (Cork, Ireland: CRC) Durand H., et al., 1994, An example of GIS potentiality for coastal zone management: Pre-selection of submerged oyster culture areas near Marennes Oléron (France) EARSEL Workshop on Remote Sensing and GIS for Coastal Zone Management Delft, The Netherlands, 24 - 26 Oct... geocoded information on the land-sea interface, through a GIS interface closely connecting these data and simulations issued from digital models Integration of consistent geo-coded information on the land sea interface with model simulated outputs within a GIS environment has resulted in the creation of an active interface for pollution response decision makers that incorporates up-to-the-hour scientific...Figures 10 3 and 10. 4 Ievoli Sun being towed a few hours before she sank (French Navy sources) The vessel was carrying three chemical products: 3998 tonnes of styrene (Vinylbenzene), 102 7 tonnes of Methyl ethyl ketone (MEK or 2-Butanone) and 996 tonnes of Isopropyl alcohol (Propanol-2) Although the latter two products are considered to be practically non-toxic for aquatic life, styrene... ppm (medium-grey) in the raised area Figure 10. 5 3D perspective representation of the plume of dissolved pollutant dispersion after 10 days Dispersion of a continuous discharge of 20 l.min-1 was also simulated for the period between 30th October and 24th November, incorporating actual wind measurements up to the 19th November and forecasts for the remainder of the simulation period Plate 10. 1 shows... model for the Channel and Southern North Sea Special Issue MAST 52, part B and C Journal of Marine Systems, 6( 5-6 ), pp 49 5-5 28 Sullivan, N., 2001, The relationship between the disposal site and the SAC in Falmouth bay, south-west England Unpublished Dissertation, Msc in Estuarine and Coastal Zone Management Valuing our environment, 1999, National Trust (UK) Wright D and Bartlett D., 2000, Marine and Coastal. .. period of 3.5 days, and c) simulation of continuous discharge of 20 l.min-1 over 25 days (closely matching observed leakage rates) 10. 6.3 Setting up a customised GIS A GIS was established for the wreck site and surrounding maritime region, in which functionality was tailored towards handling and visually representing marine and coastal data The main steps in this process were: Building a reference geographic... interoperability; Conversion of all data into decimal degree geographic coordinates, and establishing a common vertical reference frame for marine and land-based topographies; Adoption of a uniform map template for all data layers optimised for a standard working scale of 1:50,000 10. 6.4 Results The Ievoli Sun sank in an area in which strong tidal currents and an irregular seabed give rise to intense vertical... Pommepuy M., 1997, T90 As a tool for engineers: Interest and limits Wat Sci Tech., 35(1 1-1 2), pp 27 7-2 81 Ifremer, 2003, http://www.ifremer.fr/envlit/surveillance/remi.htm (on-line Accessed October 2003) Kershaw, P.J., 1997, Radioactive contamination of the Solway and Cumbria coastal zone: The Solway Firth, ECSA/JNCC, pp 4 3-5 0 Lazure P and Jegou A.M., 1998, 3D modelling of seasonal evolution of Loire and... d'eaux In "Geomatics and Coastal Environment", edited by Populus, J and Loubersac, L (Éditions Ifremer/Shom), pp 17 3-1 85 Loubersac et al., 2002, Communication de l’information géographique maritime et côtière pour la gestion d’une crise environnementale Revue internationale de Géomatique 12(3), pp 35 5-3 71 Populus J and Loubersac L., 2000, (eds), CoastGIS’99: Geomatics and coastal environment (Brest:... establish a baseline for environmental contamination, and devise a plan for coastal environmental monitoring in the subsequent hours This meant being able to characterise the fate of pollutants in the water mass and identify sensitive areas and fish species likely to be affected Thus the main requirements were (Loubersac et al., 2002): to rapidly compile various forms of multi-thematic information from multiple . iterative process of validation and calibration. 10. 3 GIS FOR COASTAL ZONE MANAGEMENT A Geographic Information System (GIS) is a computer-based information system used to digitally represent and. 1998). A GIS can be distinguished from database management systems or from visualisation packages through its specialised capability for spatial analysis. The use of GIS for coastal zone management. potentiality for coastal zone management: Pre-selection of submerged oyster culture areas near Marennes Oléron (France). EARSEL Workshop on Remote Sensing and GIS for Coastal Zone Management . Delft,