Globalization, regionalization, urbanization: an analysis of the worldwide maritime network since the early 18th century
Mapping and modeling maritime flows
The first research direction is the elaboration of a geomatics platform with strong emphasis on Geographic Information Systems (GIS) methods of data representation and analysis. Flows will be represented in diverse forms such as 2D and 3D representations on a world map and sphere using various cartographic projections. This is in itself an innovation of the project that has many pedagogical outcomes. Each port of call is a location (node) and each voyage is a link between two or more ports of call at a given period. Each vessel circulation creates both a "chain" (successive stops) and a "complete graph" (all stops connected directly and indirectly). The global maritime database is thus built from successive port-to-port matrices to be analyzed as a non-planar, weighted, and directed graph on various levels of node aggregation (e.g. port, port city, country, macro-region, continent) and where nodes and links have several attributes based on vessel characteristics (i.e. flag nationality, vessel size and type) and movements (i.e. duration, length, frequency).
At the global level, the project will measure the size of the network (i.e. number of nodes and links, total tonnage) as well as the changing structure of flows through various statistical and network analytical tools. Classic concentration measures (e.g. Gini coefficient, HHI) will be completed by network-specific ones such as those provided by graph theory, social network analysis, and complex networks. The changing size and properties of the network will be one first evidence of the globalization process and its ongoing trend of transformation. The capacity of flows to reflect and/or anticipate major global events will be examined in terms of the network's robustness and vulnerability, with reference to major works on scale-free / small-world networks and their cumulative dynamics. Preferential attachment can be tested to explore whether newly added ports generally connect to already established and larger ports. Innovation diffusion in the maritime and ports sector has been highly selective at different stages; many ports were dropped from the network, but new ports were also created or reactivated. Other global indicators can be produced such as the average duration and distance of links so as to verify the concept of time-space shrink in world exchanges. Simulation experiments will be tested with the help of other institutions in order to forecast future changes and confront past dynamics with some models of network evolution, but this is not a core objective and competency of the principal investigator. Another important dimension of the analysis will be the search for coherent substructures in the network, with reference to the buoyant research field on clusters and communities. Various methods such as single linkage analysis, hierarchical clustering, modularity, trajectory analysis, blockmodeling, and structural equivalence (see for instance Snijder and Kick, 1979) will reveal in different ways the emergence and resilience of subsystems in the pattern of flows. The idea is to map the cores and the peripheries as well as to examine the influence of various proximities in their internal and external links (e.g. spatial, commercial, political). The role of physical distance in the evolution of connectivity can also be tested with reference to small-worlds where only a few nodes act as connectors between different and/or distant communities within which most links are local. This will contribute to the fast growing research field on spatial networks where territorial embedding plays an important role (see Barthélémy, 2010 for a review).
At the local level, comparing the evolution of maritime flows and urban population data (i.e. the most widely, if not the only, accessible indicator of urban importance on a world level and over time) will allow for comparing and categorizing individual trajectories of port cities. The interdependence between port growth and urban growth will be analyzed by measuring the changing statistical correlation between city size and traffic size as well as through applying other methods such as Granger causality tests. Traffic flows per location will be normalized for the comparison of local dynamics (e.g. London vs. New York 1734-2010). The specialization of cities' traffic will also be measured in terms of commodities and vessel types (e.g. raw materials vs. manufactured goods, steamers vs. sailing vessels) as well as geographic coverage (e.g. European vs. Asian forelands, long-range vs. short-range connections) and centrality (i.e. accessibility measures, degree, closeness, betweenness, etc.). An example of cartography based on one Lloyd's register in published in 1890 (next figure) is provided for illustrating possible treatments. Specific GIS methods can be used to map the results such as cartographic distortions (e.g. dynamic anamorphic visualization of traffic distribution), traffic isochrones to/from certain centres (tributary areas), edge aggregation based on Kernel density, etc. This will help to determine the dominance of certain cities not only in terms of traffic volume but also in terms of the geographic reach of their maritime connections. A complementary approach will be to measure the centrality of cities on the level of a combined sea and land network, where port cities act as connectors between the two spaces (foreland and hinterland). Based on the modelling of land-based transport networks, such an approach will better assess the role of hinterlands on the emergence of some gateways as opposed to maritime hubs as well as regional particularities in this interdependence (e.g. North European gateways having wide hinterlands, Asian port cities being more maritime-oriented). This will complement and challenge current works on the structure and dynamics of coupled and interdependent infrastructure networks (Vespignani, 2010).