AN INTEGRATED GIS METHOD TO EVALUATE SUSCEPTIBILITY AND HAZARD OF SEA LEVEL RISE IN COASTAL PLAINS:
THE VERSILIA PLAIN

Giuliano Gallerini (1), Mauro De Donatis (1), Saverio Devoti (2), Massimo Gabellini (2), Marco Fulvio Nisi (2), Sergio Silenzi (2)

(1) University of Urbino, Environmental Sciences Faculty (IT)
(2) ICRAM, Central Institute for Marine Research, Roma (IT)


ABSTRACT
A large amount of scientific evidences suggest that an accelerated sea level rise will have a significant impact over the next 100 years because of current changes in the climate. The prevision of impact on coastal plain areas becomes more important as a demographical increase in the coastal regions is evident. The aim of this work is to provide a methodological approach for the evaluation of susceptibility, hazard and risk related to the Relative Sea Level Rise (RSLR). This method, developed in a G.I.S environment, will predict the future physiography of the territory over the next 25, 50 and 100 years (with optimistic, intermediate and pessimistic scenarios), evaluating the susceptibility of an area to possible damaging events resulting from RSLR. The model has been tested on the Versilia coastal plain (NW Tuscany), a particularly vulnerable area to RSLR because of its morpho-altimetric asset and subsidence rates. The methodology could have a wide application range, being well suited for the evaluation of integrated hazard of coastal zones with different characteristics. The proposed method could be used by local government authorities for further detailed studies and as a powerful tool for territorial planning and management.

1. INTRODUCTION
Various possible impacts of climate change on the earth depend on the increasing concentration of greenhouse gases (IPCC [Intergovernmental Panel of Climate Change] 1990, 1996, 1998, 2001). One of the possible impacts associated with climate change is an acceleration of the Sea Level Rise (SLR). The IPCC has developed different future scenarios on the possible rate of the sea level rise until the year 2100.
Eustatical and geo-morphological researches on Italian coastal zones show that many plains are susceptible to flooding as results of Relative Sea Level Rising (RSLR) (Aminti et al. 2001; Nisi et al. in press).
The aim of this study is to develop a preliminary methodological guideline for the evaluation of the Susceptibility, Integrated Hazard and Risk of coastal areas to relative sea level changes, which would provide indications for a compatible territorial management through editing thematic maps.
We adopt a multidisciplinary approach, utilized for the pilot studies carried out in the plains of Versilia. This includes geological, topographical, geomorphological, hydrological, and land use surveys, forecasts ground level changes and accelerated beach erosion with respect to sea level rise.
All data are processed using Geographical Information System (GIS) because of its advantage in handling data and easily accessible spatial data, its usefull tools for modelling and integrating data from various sources, and for transforming data and maps in new informations that is useful for decision making. Also integrating 3D analysis and 3D views it's possible to obtain a more realistic vision of the future scenarios useful for specific evaluation and support decision makers. 3D views represent a synthesis of scientific documents, offer new information to integrate with the input data and increase the implicit value of the data.
In particular, we process and edit the following map-layers (GIS):

The main outcomes of this integrated approach are: hi-resolution zoning of land for potential future assessment of the coastal plain; evaluation of Risk connected to RSLR; planning of best land use, sedimentary deficit and coastal management; a relational database concerning all survey data and thematic maps, implementable with new set of data .

2. CONCEPTS AND METHODS
The methodological approach to RSLR evaluation is not exclusively a deterministic approach of geomorphological processes.
Infact, deterministic models (Hybrid Bruun, Aggradation and Translation models) are implemented into GIS as a set of equations, which are then applied to the terrain data. Changing parameter values in the models allows the user to run a number of impact scenarios for each model and locality. In practice this deterministic approach may not be readily applied, because:

The modelling of geomorphological processes, which are responsible for the erosion and deposition of coastal sediments, is difficult as these processes are poorly understood, especially at time scales relevant to engineering and planning (Cowell and Thom, 1994). Therefore, long-term morphological predictions need to be met with knowledge that is available today (Stive and de Vriend, 1993).
The multidisciplinary approach adopted in this work analyzes the causes of risk from RSLR, using a deterministic method for calculating spatial and temporal evolution of RSLR and a point count system model (Civita, 1994; Civita and De Maio, 2000) for evaluating susceptibility, hazard and risk. The overlay GIS operation and the computation of the original scores produce the final maps properly classified.

The methodological steps consist of:

3. GIS STRUCTURE
To represent and simplify the complex interaction of factors that influence the susceptibility, the hazard and the risk of a territory from RSLR, this conceptual model has been built up in order to graphically represent this methodological work-flow (see fig.1).

Figure 1 - Methodological scheme for RSLR Risk evaluation and study area.

The first step involves the territorial analysis.
Data Surveys (D): acquisition of data on geology, lithology, geomorphology, altimetry, palaeoenvironments, hydrology, hydrogeology and land use.
Homogeneous Territorial Units (HTU): spatial distribution of the parameters predisposing to local RSLR susceptibility. The parameters have been identified among the territorial data of the surveys, in order to distinguish elementary parts of the territory where every parameter can be considered as a constant.
Susceptibility (S): assessment of the predisposition of the territory to instability. Appropriate scores and weight multipliers are attributed to the single predisposing parameters, and all informative strata are then intersected by means of a susceptibility function. This process has been widely used for the evaluation of environmental vulnerability.

The second step predicts the physiographic changes.
Future Scenarios (F): modeling the future physiographic and altimetric placement of the studied coastal area, aimed at determining the future coastlines and the 0 m a.s.l. isohypsa (development of areas below sea level in the emerged sectors) according to three different scenarios (called optimistic, intermediate and pessimistic), respectively connected to minimum, medium and maximum sea level rise for each predicted period (years 2025, 2050, 2100). The shoreline regression has been calculated on the basis of Bruun's Rule of Erosion (Bruun, 1962). "Sea Level", an automatic calculation software, was used for the construction of the future coastal regression scenarios (Pranzini & Rossi 1995; Aminti et al. 2001). Specific Hazard (Hs): each of the various scenarios obtained through the Bruun Rule Model and the DTM Elaborations receives a "score" indicating the degree of danger. The scores are assigned to the three scenarios of one same prediction year, related to the mutual occurrence probability (Optimistic, Intermediate and Pessimistic Scenario).

The third step calculates the Integrated Hazard and RSLR Risk
Integrated Hazard (H): the intersection between Specific Hazard and Susceptibility; it is used for the evaluation of the occurrence probability of several potentially harmful phenomena connected to the RSLR.
Risk (R): delimitation of areas with homogenous risk values through the intersection of the Integrated Hazard scores and of those linked to the risk elements. This method is applied to three different prediction periods (year 2025, 2050, 2100) because the increase in RSLR impact on the shoreline will not be linear, but it will tend to adopt an asymptotic trend; though it seems likely that such a development will assume impressive proportions only on a secular scale, it is also necessary to know what the short term situation in the territory will be.

The first and the third steps of this methodology use a point count system model. This system is built up on a rating given to every selected parameter as a function of its importance within the final global assesment of the natural phenomena. Ratings of each parameter may be summed (Rating Systems) or crossed into a matrix (Matrix System) or multiplied by weight strings that are able to describe each impact situation to magnify the action and the importance of the parameters in various levels (Point Count System Model) (Civita, 1994; Civita and De Maio, 2000). This system model offers a great advantage because it compares different physical sizes evaluating only the importance of the parameter within the natural phenomena.
The application presented in this work is the first application of this new methodological approach in RSLR study. Points and weights used in this application may not be adapted for a different coastal zone. However the methodological schema is always applying.

3.1 The Geo-DataBase
The data model is organized in a DataBase Management Relational System (DBMRS), using new functionalities of storing coordinates of the graphic elements into a Binary Long Object (BLOB) field. It realizes a GEO-DataBase (GEODB) in Microsoft Access MDB format, using Intergraph Geomedia GIS software.
The GEODB offers:

The GeoDataBase realized contains these data that integrate the Survey Data (SD):

The GEODataBase is implementable with new survey data because of its open structure. Data point could be elaborated with geostatistical tecniques for obtaing new continuos themes usefull to integrate with others theme just stored in the database and for develop GIS analysis. New data can offer, also the possibility to develop new specific elaboration: for istance hazard and risk evaluation from extreme coastal storm surge events.
Infact susceptible coast are defined as those which are at present or have been in the timescale chosen normally frequently affected by storms and/or where storms are belived likely to cause significant coastal change. Extreme coastal storm surge events are a potential consequence of climate change. In addition to the effect of rising mean sea level, coastline future changes in local meteorology will lead to further significant reduction in the return periods of extreme storm surge events. Account is being taken of the resilience of such coasts, the ability of the coast to resist change due to storminess or to achieve equilibrium in its physical development following change. (Lowe et al. 2001).
So the GEODataBase could be implemented with these data:

4. A CASE STUDY IN VERSILIA PLAIN
The studied area spreads along Tuscany's northern coast, stretching between the mouth of the Cinquale stream and that of the Serchio River. It comprises the Versilia plain and the northern margin of the Pisa Plain.
The Pisan-Versilian plain is part of a single wide tectonic subsidence basin with apenninic direction, known as the Viareggio Basin (Della Rocca et al. 1987). The sector's geologic skeleton is represented by the Apuan chain, composed of a nappe orogene of Alpine Age; The strong, vertical motion have resulted in the creation of a dome-shaped structure, characterized by thrusts.
The subsequent tectono-genetic phase, which mainly occurred in Neogenic Age and in the whole area, is characterized by direct faults with high vertical displacement, and it has caused the subsidence of the Versilian tectonic basin. The whole coastal plain is also characterized by soil lowering due to anthropic activity (Antonioli et al. 2000).
The outcropping lands in the plain comprise elements which range in age from the Upper Pleistocene to the whole Holocene and they are mainly composed of conglomerates, gravel and silty-sands of the piedmont fans, alluvial deposits, siliceous sands and eolic silty sands of the dune bars, medium-fine highly siliceous beach sands. The surface water table in the plain emerges at various times, both at the base of the alluvial fans and by the topographic depressions.

4.1 RSLR values
On the basis of the sea level rise values, proposed by the IPCC report (2001), it has been possible to extrapolate the SLR in cm predicted for the years 2025, 2050, 2100. For each of these periods three different values have been considered, respectively corresponding to a minimum, an intermediate and a maximum sea level rise scenario (Table 1).
These rates must be added to the effect of the dislocations' vertical component acting locally, substantially represented by a differential subsidence, which is due to neo-tectonics as well as to the compacting of recent sediments (Table 2).

Table 1: Minimum, intermediate and maximum SLR values predicted for the years 2025, 2050 e 2100

Min SLR (cm) Int SLR (cm) Max SLR (cm)
Year 2025 3 8 14
Year 2050 5 18 31
Year 2100 9 48 88

Table 2: Subsidence Values considered for the years 2025, 2050 and 2100. Coastal area - Massaciuccoli basin

Year 2025 5 - 17.5 cm
Year 2050 10 - 35 cm
Year 2100 15 - 70 cm

4.2 Surveys
On the basis of the characteristics analyzed in the studied area, are identified the predisposing parameters (stored in the GeoDataBase) directly or indirectly influencing the territory's evolution, which is linked to sea level rise and related phenomena.
From Geological survey derive the definition of predisposition to erosion factor (considering also the protection effect of the type of land cover) and permeability factor.
From Geomorphological survey derive the definition of coastal dynamic factor and morphology with propension to be modified factor (depending on altimetry classes).
From Hydrological survey derive the definition of flooding area factor.
From Hydrogeological survey derive the definition of critical water table depth factor (depending on the worst scenario of SLR for each one forecast period)

4.3 Susceptibility
Susceptibility represents a territory's tendency to undergo and/or to contrast a potentially destructive phenomenon. The Susceptibility maps produced for the Versilian plain originate from the scores attributed to the single Homogenous Territorial Units for predisposing parameters found in the area. The results of the final overlay have been subdivided into three classes: high, medium and low susceptibility. The analysis of the maps shows that for all the three prediction periods the area of greatest susceptibility is, obviously, the coastal zone (which comprises the beach and the first dune bar) especially in the non-urbanized stretches. The presence of discontinuous urban settlements along the coast represents instead a protection for the shoreline, but it causes a further sedimentary deficit that worsens the erosion processes in the neighboring areas. The backdune area presents medium susceptibility, being characterized by low heights and lithologies, which favour water stagnations. In such a sector it is therefore possible to observe an expansion of the areas with high susceptibility as a function of the increase of the prediction period. This is because, while the other parameters can be considered to be substantially invariable in time, at least on a secular scale, the critical water table depth value becomes more and more influential as the RSLR values increase.

4.4 Future Scenarios
The maps of the Future Scenarios in Versilia describe the predicted physiographic look of the plain (plano-altimetric modifications, areas submerged by the sea, floodable inland areas, etc.) in relation to eustatic variations and to the differential subsidence rates (Figure 2). For each prediction period (years 2025, 2050 and 2100) it has therefore been possible to determine three different scenarios as a function of the three different RSLR rates (maximum, intermediate and minimum).

Figure 2: Detail of the pessimistic scenario for the three periods.

4.5 Integrated Hazard and Risk
The Integrated Hazard maps aim at grouping the Versilian coast into zones depending on whether they may be affected or to contrast the occurrence of potentially harmful events connected to RSLR. These maps are obtained through the intersection between the territory's Susceptibility and the Specific Hazard values, which are derived from the scores awarded to the Future Scenarios. These scores are univocal for all of the three prediction periods, though they differ depending on whether the area they refer to is reached by sea ingression or if it will not be submerged even though the area is below sea level; the Specific Hazard values will be equal to zero in the areas placed above sea level.
The Integrated Hazard Zonation, obtained in this way, leads to the subdivision into 6 classes of hazard, ranging from extreme to very low. From the analysis of the produced maps it is obvious that certain susceptibility parameters greatly contrast shoreline regression. An example of these parameters are the fossil dune systems or the urban centers; the most sensitive areas are instead the depressed areas and those situated by the river mouths, where the extreme Hazard class is most evident.
The Risk maps produced provide an evaluation of the expected loss according to a relative economic evaluation. These maps are obtained through topological analysis of the informative levels of Integrated Hazard and the map Land Use, subdivided depending on the evaluation of the economical value of the exposed lands. For this purpose, the various types of Soil Use have been grouped into five classes: the humid areas, forested and natural vegetation areas, agricultural land, productive areas and infrastructures, and urban areas/bathing establishments.
Though this is an approximate first evaluation, the RSLR risk evaluation in Versilia clearly indicates that if the recent predictions of sea level rise should turn out to be reliable, the economic loss linked to beach erosion, infrastructure damage, injury to the urban centers and to the agricultural areas would be extremely high.

5. G.I.S. WORKS
The methodological approach and the application to the Versilia plain, as described above, has been carried out through GIS operations and analysis.
The database used is Microsoft Access. This choice offers the advantage to store all the tables of the project in a MDB file, while GIS acquisition data and analysis are realized using different Gis Software: Microstation Geographics by Bentley Systems for goereferencing original maps survey, digitizing, linking to DB, checking topology of graphical elements; Geoterrain by Geopak, a specific module of Microstation, is used for DTM analysis; Arcview by Esri and Geomedia by Intergraph for topology analysis and for the distribution of the final maps; TN3D by Terranova for draping orthophoto on DTM and exporting photos and videos of the attended scenarios.

One of the main GIS operation conducted is the DTM elaboration for elaborate future scenarios. The data entry is the Regional Technical Map (CTR) of Regione Toscana, scale 1:5.000 in DXF format and bathymetric source points. The TIN model realized counts over 245.000 3D points, over 735.000 3D lines as data input , for over 500.000 triangles.
The first step consists in exporting 24 topographic sections from the -50 m isobath to 4 km landward and conversion to excel format, for determining the future coastlines and the 0 m a.s.l. isohypsa using the "Sea Level" application (Pranzini & Rossi 1995; Aminti et al. 2001).
The second step consists of obtaining a provisional Digital Terrain Model (DTM) of pessimistic, intermediate and optimistic scenarios for the years 2025, 2050 and 2100 (for a total of nine different DTMs), by considering sea level rise and differential rates of subsidence (Table 1 and 2).
Such elaboration developed according to the following operations:

The third step, conducted later, consists on georeferencing aerial orthophotos of the study area and draping them to DTM obtained for the each temporal prevision. The results of this elaborations offer a more realisitc vision of future scenarios, better evaluation the damaged areas and the worst possible outcomes. Photos and videos (simulating fly on the study area) are very usefull for specific evaluation and for support decision making. 3D views give a contribution to an incisive representation of the project results because they offer a simple and direct method of displaying structured and technical natural phenomena. Also 3D documents increase the information available from the data input and so represent a synthesis of scientific documents that increase implicit value of the data (Figure 3)

Figure 3: Versilia plain in 3D view from south. Coast line in the year 2100 wtih maximum scenario of SLR.

GIS overlay analysis represent the most applied operation in all the project phases.
Homogeneous Territorial Units (HTU) are a translation of survey data regarding those parameters which are predisposed to RSLR susceptibility (Predisposition to erosion protect by land use, Permeability, Morfology with propension to be modified disctinct by altimetry class, Coastal dynamic, Flooding area, Critical water table depth). Each theme has associated points and weights and the Homogeneous Territorial Unit (HTU) synthetic maps are elaborated, through a GIS overlay analysis using intersect operator.
For each forecasting period one susceptibility map is obtained. The result was achieved by processing HTU maps through a GIS overlay analysis. The final score is obtained, according to the following:


The results are subdivided into an appropriate number of classes and re-classified using an equal interval method (same range of value for each one class), to obtain Susceptibility maps.
The sets of colors used for the graphical representation of thematic maps respond to a color ramp or simbolic scale (i.e. traffic-light color scale).

Using the same methodology and the same GIS techniques described, we applied another GIS overlay analysis between the Susceptibility map and the Specific Hazard Map to obtain an Integrated Hazard map.
The Risk map is extracted through the GIS overlay operation between the Integrated Hazard map and a re-classified Land Use map (assigning to each relevant element a normalized score, based on its economical value).

6. CONCLUSIONS
The proposed methodology, implemented through GIS, allows to carry out forecasts on any temporal period and it predicts the realization of three different scenarios in relation to an expected minimum, intermediate and maximum relative rise of sea level. The entire cartographic set, which explains the whole methodological approach, is composed of 4 base maps (surveys), 6 Homogeneous Territorial Units (HTU) and of 3 maps derived from prediction interval (years 2025, 2050, 2100) for susceptibility, hazard and risk.
The main points of the integrated GIS methodology and techniques used are:

The model's effectiveness has been tested in the Versilian plain, an area that is particularly vulnerable to the effects of the RSLR. In particular, the tourist area of Viareggio would seem to be subject to a high probability of submersion, unless remarkable preventative measures are taken. Moreover, wide portions of the versilian backdune areas turned out to be highly exposed to flooding and to sea ingressions as well as to the worsening of meteorological events and wave phenomena.
It is however important to underline that the results obtained from any model concerning climatic changes must be dealt with cautiously because of the uncertainties regarding the effects of global warming and the results of the models in general. Also, in the case of the Versilian plain, the reliability of the results decreases as the predicted time interval increases. This is essentially due to the fact that the values of subsidence for compacting of organic soils are analyzed over large temporal intervals (more than 25 years); such values represent in fact an important share in the total RSLR rates, but the data they originate from is not sufficient to determine the time of the end of phenomenon when related to the sediment's thickness.
By applying the precaution principle the maps created can be used for a fast identification of the high risk coastal areas, as well as for planning nourishments and other coastal defense works, or for deciding possible restrictions to the construction of new infrastructures in the areas characterized by the greatest hazard classes.
Interesting develop are presented about the possible application of the method to evaluate susceptibility, hazard and risk from extreme storm surge events by its integration and GEO-DB implementation.

References