GIS AND COASTAL CHANGE ON THE FRENCH RIVIERA.
50 YEARS OF COASTAL MANAGEMENT FROM NICE TO MENTON

Samuel ROBERT,

CNRS - UMR ESPACE 6012, Universitè de Nice Sophia Antipolis (FR)

Coastal areas have been studied for quite a long time throughout the world using GIS. Many concerns have led scientists, planners, decision makers and others to build geographical databases to handle coastal information and to facilitate their work. Sometimes dedicated to specific subjects such as seagrass meadows, coastline erosion, landscape evolution, etc., coast-oriented GIS can be more territory-oriented. It is therefore more global. It then works as an earth observatory, at a local or regional scale. Very often, such a system relies on diachronic data : the database consists of various types of information related to different dates and periods of time. This multi-temporal database allows temporal analysis and the study of changes, whatever they may be. Here below, we will be detailing such a system; which is still in the very first stages of construction.

Dedicated to the French Riviera, the Côte-d'Azur, this GIS is focused on the area that stretches between Nice and Menton (Figure 1). This famous coast, where the Mediterranean sea and the Alps meet, has been seriously transformed and developped for more than one hundred years. It is one of the major seaside resorts of the world with prestigious sites including Nice, Monaco-Monte Carlo, Saint-Jean-Cap-Ferrat and Beaulieu-sur-Mer. Geography is undoubtedly at the origin of the beautiful landscape which has led to the region's success, but it is also a major constraint to development. Space is scarce and terrain is often very critical. Here, our main concern is to build a tool reflecting 50 years of coastal change in the region, namely how society has been transforming this coastal area.

1-The process of coastal change and the related data

What is coastal change ?
Coastal change is a global phenomenon affecting coastal areas in such a way that many of their original characters are modified. Land and sea, landuse and landscape, demography and economy, are all interconnected in a geographical system, thus contributing to the identity of a region which is highly dependent on the seashore (Figure 2). At the junction of land and sea, the coast gives its originality to the region. It is so influential that it is able to initiate modifications on the local economy, on population distribution and structure, and on the landscape and landuse. The system is not stable, but always changing (dynamic equilibrium). There is coastal change when many of the original characters are highly modified and, as a result, the system shifts from one equilibrium to another.

As far as the French Riviera is concerned, there has been a tremendous coastal change over the past 50 years. This phenomenon is not unique to this area, nor to France or to Western Europe. Everyone knows that coastal regions have been more and more settled and developped in the last decades (for instance the Costa Brava and the Costa Blanca in Spain). Between Nice and Menton though, changes have been extreme, various and geographically concentrated (there are less than 30 kilometers between the mouth of the Var river and the Italian border) :

This part of the French Riviera happens to be a kind of laboratory of coastal dynamics on the Mediterranean sea. It is an area that can be studied to understand coastal change process, using GIS along with an appropriate database. The area of study is composed of 25 municipalities (Monaco is considered as such), 10 have a coastline and 15 are separated from the sea by only one municipality.

Primary data
To handle the process of change, our GIS project relies on primary data.
First, we need data refering to the landuse and landscape reality of the area of study. Aerial photos are then required as they provide objective images of the coast at various dates. These are all supplied by the French Institut Géographique National (IGN). While July 1999 coverage is already in a digital format (Bd-Ortho), all the other photos (from July 1955 onward) are raw images in paper format. Photogrammetric treatments are therefore necessary to take advantage of these images. Along with aerial photos, a digital elevation model (DEM) needs to be incorporated into the database in order to integrate terrain data. This file will be used to rectify raw aerial photos and to compute 3D views.

Another set of primary data is composed of statistics and related graphic files describing geographical units (i.e. : municipalities). Population and agricultural censuses are the main sources, but other statistical surveys such as in the field of tourism are to be added for more complete information.

Secondary data
From the primary data, other layers of information will be derived using GIS capabilities.
Once computed and transformed into orthophotomaps, aerial photos will then be used as sources to digitize geographical features reflecting pieces of change. Among many of them, the following can be quoted:

Since these features will be digitized at various dates, comparisons will highlight evolutions throughout the area of study.
Statistical information will be computed in order to provide indicators. During the last 50 years, there is no doubt about the link between population growth, urbanization and agricultural decline. But what kind of link exists ? How strong is this link ? The secondary data shall be produced to answer these kinds of questions. Here, GIS will allow two levels of analysis :

Technical solution
The project is based on a commercial platform combining three softwares : ER-Mapper, MapInfo and Access. They offer the possibility to create, to manage, to store and to analyse various types of geoinformation. MapInfo, the GIS oriented software, is indeed highly enhanced with the two others :

Orthorectification and mosaicing
Masterpieces of the database, aerial photos need to be orthorectified in order to be compliant with the GIS. As stated earlier, most of the aerial data are raw data in paper format, i.e. black and white as well as color pictures. These images do not comply with digital cartographic data because they are not geocoded and they must be corrected from 3 major distorsions :

The orthorectification process is achieved using : a raw image which is scanned (1), a previously rectified and geocoded image or map, which is used as a reference map (2), a camera file to handle optical parameters (3) and a DEM to take terrain morphology into consideration (4). This is performed with Er-Mapper software. Later on, rectified images (or orthoimages) are combined together to create a seamless orthophotomap covering the entire area of study (Figure 3). This is perfectly compliant with other geographic files already integrated in the GIS.

The French Riviera project will rely on several aerial datasets to be orthorectified. They are at various scales (from 1:10 000 to 1:30 000). All shall be rectified using IGN Bd-Ortho (digital ortophotomap 1999) as a reference map and a 20 m resolution DEM made from IGN Bd-Alti (digital height data, planimetric accuracy : 50 m, altimetric accuracy: 10 to 20 m). The July 1955 orthophotomap which is already completed, is made of 31 snaps, scale 1:25 000 from IGN 1955 F - 3743-3742 dataset.

Digitizing features
Once orthophotomaps are ready, they can be used as sources to digitise features of interest in our study. All these features are organised in a data model where they can be characterised. Then, the data base management system (DBMS) allows further investigations : thematic or spatial selections, buffer analysis, 3D views, etc. As in all GIS, three major kinds of geographical features are possible : polygons, lines and points. They are all stored according to the data model, based on the relational database management theory where features are "entities" having "attributes" and linked to each others by "associations". This means, among many other things, that three kinds of files will be created by digitization because one file cannot contain more than one feature type. The data model indicates what feature shall be digitized, how and with which attribute.

Data integration
Other geodata that will be part of the GIS will not be digitized. They will only be integrated since they are in a digital format. Some will be easily incorporated without any modifications as they are already in an appropriate format and in the geodetic system selected for the project (NTF, Lambert 3). Others will require geodetic or format conversion in order to be compliant. At the moment, these layers of information are essentially provided by IGN and INSEE (Institut National de la Statistique et des Etudes Economiques) such as : administrative boundaries, transportation networks, scanned maps, etc.

3-Early results

Only a few results can be presented as our work has begun recently. They show several changes, mainly from the comparison between the July 1999 orthophoto and the July 1955 orthophoto, and from the analysis of population censuses.

Demographic evolutions
From 1946 to 1999, the population of the area of study increased by more than 180 %. This increase in population occurred mainly during the first half of the period. At that time, intercensus growth rates were above 12 % (+ 19.9 % between 1954 and 1962). Recently, growth is more calm : only + 0.6 % between 1990 and 1999. Municipality-wise, population growth is more contrasted. If the second half of the period is considered, there are some interesting pieces of information to collect from the coastal change point of view. Since 1975 indeed, most of the population growth is located in non-coastal municipalities, whereas the number of inhabitants has decreased in many places along the coast : in Villefranche, Beaulieu, Saint-Jean-Cap-Ferrat and in Roquebrune-Cap-Martin (Figure 4). It seems coastal places are less attractive. Is there any spatial influence of this demographic stagnation ? Why do these municipalities have a decreasing population ?

Urbanization of coastal plains and coastline modifications
The important population growth that happened in the second half of last century has had many spatial consequences. Urbanization is one of them. It has spread over all the coastal plains like in Carnolès, municipality of Roquebrune-Cap-Martin (Figure 5), and has sometimes led municipalities to reclaim land over the sea. Upper beaches have been settled to install roads. In relation to this urabanization, coastal protections have been erected to prevent erosion from th sea. The process has been so huge that from Nice to Menton, the only places which are not built-up are steep cliffs.

Reclaimed lands in Monaco
One of the smallest states in the world, Monaco has always been faced to the geographical limits of its national territory. During the second half of the XXth century, tremendous developments have been made, the major of them being the reclaimed lands (Figure 6). In the South-West part of the country, the Fontvieille area is entirely buit on new lands wich have been conquered over the sea. Residential buildings, Louis II stadium, industrial estates, Monaco national heliport and a yachting harbour have appeared where formerly, in 1955, one could find a long beach and a quay.

Remainings of agriculture in hilly areas away from the coastline
Since 1950, agriculture has been severely declining on the French Riviera. This emerges from agricultural statistics but also from aerial images. From orthophotomaps indeed, one can see the expansion of urbanization over former agricultural lands. Using the DEM, it appears more clearly that, as a result of the competition between urbanization and agriculture, agriculture remains mainly either far from the coastline or in hilly and less accessible places. Figure 7 represents the area of Bellet in the north-western part of Nice. On slopes, it shows greenhouses almost exclusively. In flat places, other buildings compete : mostly houses on hill tops and large commercial or industrial buildings in large flat areas, like in the Var river valley. Closer to the sea, agriculture is relictual, even on slopes.

Conclusion
At this stage of our work, efforts are made to complete the features data base. The main target is an analysis of urbanisation, which will rely first on the comparison between buildings / houses spatial distribution in 1955 versus 1999. One concern here is to highlight the spatial patterns of urbanisation along the coastline. How dense is the urban factory within a 500 m buffer in 1955 and in 1999 ? How often new constructions have replaced others on the same places (urban renewal) ? What kind of urbanisation have been developped : residential areas with houses and gardens or block buildings, planned or spontaneous projects, integrated (yachting harbour - appartments - commercial center) or spontaneous projects, etc. Another concern is to highlight the relation between urbanisation and population growth rates at the level of municipalities. Though urban areas expand, population stagnates in several places indeed. This shall be analysed using demographic and housing indicators such as age structures, local immigration rates, housing sizes, housing vacancy rates, etc. Using GIS, many other links between urbanisation and the other components of the French Riviera coastal system are to be studied.

As previously explained, the process of coastal change is a multithematic field of research and, as far as geographers are concerned, it can be analysed using GIS. The system allows the storage and the analysis of various data types. It is a wonderful tool for territorial investigations and scenarios building.

Landscape and socioeconomic data are indeed combined in a unique information system, offering the capability to explore and understand the functionning of a coastal territory and the mecanisms of change.

References