The changing influence of city-systems on global shipping networks: an empirical analysis – Journal of Shipping and Trade

Maritime network construction

Among all existing maritime data, the Lloyd’s List, a global drawing card in shipping news, is the only potential source capable of documenting the global distribution of nautical flows in a disaggregated manner and over meter. The Lloyd’s Shipping Index had been published daily or weekly since 1880 on a regular basis since the late nineteenth hundred. It contains data about vessels and their latest inter-port movement at the date of the publication. For the purposes of this research, it was decided to extract from paper sources one issue every 5 years between 1950 and 1990, around April-May, and to compute the number of vessel calls per port and per inter-port radio link. Each 200-page publication frankincense provides a comparable snapshot of ball-shaped maritime action covering approximately 1 week of movements. The unmanageable legibility of the print original documents could not even allow for the origin of all information, namely the tonnage capacity of vessels, with conventional capabilities of Optical Character Recognition ( OCR ) software. The extract information went through a harmonization process whereby all port names were verified and disambiguated to avoid errors, as many of them changed over clock, aboard decolonization trends for example. The leave tables were merged into one single nautical database, which served to construct a ball-shaped origin-destination ( or adjacency ) matrix of inter-port maritime flows. In the network ( or graph ), ports are considered as nodes ( vertices ), and flows between them as links ( edges ), to allow the calculation of standard net measures originating from graph hypothesis ( Ducruet and Lugo 2013 ). The adopt measures ( see Appendix 1 for a graphic exemplification ) were retained to be calculated at the flat of urban areas, the latter being defined in the following section. Degree centrality is the issue of adjacent neighbor nodes connected to node i as in formula ( 1 ), the most common measure for any node that expresses its actual connectivity at the local level of its neighborhood. Betweenness centrality is the number of short paths among all nodes on which node i is located ( 2 ) ; it is more a global-level measure of net approachability taking into bill the stallion graph. The clustering coefficient ( 3 ) is the proportion of observe triangles ( or triplets, cliques ) in the maximum potential total of triangles ( or triplets, cliques ) in the neighborhood of node i, which has the like distribution than the share of observe links in the utmost possible number of links in this neighborhood. This coefficient is low at ports being bridges or hubs in the network, i.e. which neighbors are ailing connected with each other, as in the star or hub-and-spoke shape. conversely, high values indicate that ports are share of more dumbly connected or meshed patterns. last, the rich-club coefficient ( 4 ) corresponds to the concentration of the subgraph of larger cities ( i.e. with a population higher than world average ) divided by the concentration of the whole net ; the concentration being defined as the proportion of actual links in the maximal possible number of links. Values over 1 signify that larger nodes ( here in terms of population ) are more densely connected with each other than the rest of the network, a phenomenon besides known as the “ rich-club consequence ”. At the flush of urban areas, such measures can reveal to what extent is maritime centrality of unlike kinds influenced by the urban system of weights of port nodes.

$$ \begin{array}{cc}\hfill \mathrm{Degree}\kern0.5em \mathrm{centrality}\hfill & \hfill {k}_i={C}_D(i)={\displaystyle \sum_j^N{x}_{ij}}\hfill \end{array} $$

( 1 )

$$ \begin{array}{cc}\hfill \mathrm{Betweenness}\kern0.5em \mathrm{centrality}\hfill & \hfill {C}_B(i)=\frac{g_{jk}(i)}{g_{jk}}\hfill \end{array} $$

( 2 )

$$ \begin{array}{cc}\hfill \mathrm{Clustering}\kern0.5em \mathrm{coefficient}\hfill & \hfill {C}_i=\frac{2\left|\left\{{e}_{jk}\right\}\right|}{k_i\left({k}_i-1\right)}\hfill \end{array} $$

( 3 )

$$ \begin{array}{cc}\hfill \mathrm{Rich}\hbox{-} \mathrm{club}\kern0.5em \mathrm{coefficient}\hfill & \hfill \phi (k)=\frac{2{E}_{>k}}{N_{>k}\left({N}_{>k}-1\right)}\hfill \end{array} $$

( 4 )

Ports and urban spatial structures

This research benefited from the handiness of demographic data in the Geopolis database, which provides the demographic size in urban areas over 100,000 inhabitants in 1990 based on geomorphologic criteria over the period 1950–1990 ( Moriconi-Ebrard 1994 ). As presented in Fig. 1, two levels of urban activity have been distinguished, city and urban area. The city level is the municipality where the port is located, i.e. the smallest administrative area that is frequently the eponym of the port itself. The urban area level is the agglomeration or urban morphologic area, with two possibilities : the urban area to which the city belongs, or a more distant, inland urban area that connects by road the city, the latter being the nautical mercantile establishment of the early .Fig. 1figure 1 methodology for port-city match. generator : own realization Full size persona Each port or terminal was associated to the nearest urban center taking into history urbanization patterns, physical proximity, road approachability, and urban system layout ( see Appendix 2 for a description of quantiles ). This manual method acting was preferred to any automatic pistol match in a Geographical information System ( GIS ) to avoid putting together cross-border locations belonging to radically different historical or socio-economic context. In addition to manual match using the web site Google Maps for locating each port within or near a given city or urban area, we used versatile port-specific websites to retrieve them, such as World Port Source, Maritime-Database, and Portfocus, a well as numerous websites of individual port authorities. In many cases, it had been necessity to verify the probable geographic extent of interface hinterlands by consulting a wide variety of historic documents, which can not be listed in this newspaper due to their numeral and diversity. unfortunately, the absence of taxonomic information about backwoods flows could not help to delineate them with preciseness, which is a perennial trouble in port geography ( Guerrero 2014 ), particularly for studies having a historic concentrate. In any font, this method is a necessary reduction of reality to allow discussing the distribution of flows in relative to the size and dimension of the places of cargo ( coastal urban area ) and in some cases, the likely places of consumption/production ( inland urban area ). Yet, vessel movements match to inter-port segments within a wider sequence of port calls, in which there is no information about the true lineage and finish ( and measure ) of the transport cargo. however, Fig. 2 introduces four perennial cases of port-city match, for exemplify with a big upstream urban center exerting its dominance upon a large number of estuarine ports up to the slide ( London ) ; a more polycentric, coastal urban system including cities of equivalent size and evening composing conurbations ( Southampton-Portsmouth, Bournemouth-Poole ) ; an inland urban center located near the coast but having a road access to maritime transportation through smaller coastal urban settlements ( Chiclayo in Peru ) ; and two major urban areas being connected over land with one individual nautical access ( Sao Paulo, Santos in Brazil ). such cases can well be extended to other examples worldwide due to the general character of urban liquidation patterns. Yet, the only drawback of this methodology is to ignore the historical development of urbanization, as colony patterns were quite different in the 1950s compared with nowadays. In summation, certain cities ( or communes, districts ) at the administrative level did change their boundaries over time, creating a bias or modifiable Area Unit Problem ( MAUP ). To enable our analysis on the 1950–1990 it was chosen to work alone at the urban sphere ( agglomeration ) charge and to consider that urban structures have been spatially relatively stable over prison term, despite urban conurbation and suburbanization ( Bretagnolle 2009 ). As a consequence, it was possible to aggregate many terminals, ports, and cities wholly that in fact serve the lapp urban area, and profit in spatial coherence .Fig. 2figure 2 Locational aspects of coastal and inland port-city meet. source : own realization based on Google Maps Full size persona As a consequence and based on Fig. 1, the ball-shaped port-urban database consisted in 529 urban areas having at least one vessel call between 1950 and 1990, such cities being directly matched with a interface ( a and b in Fig. 1 ). This come increased to 628 when matching extra ports to the closest urban area ( c and inland in Fig. 1 ). These 628 cities concentrated a growing plowshare of the total count of ports in the maritime database, from 51 % in 1950 to 63 % in 1990, but a slenderly declining contribution of total populace population ( from 53 to 47 % ) and global vessel calls ( from 82 to 78 % ). The extra hundred cities added a mere 14 % of world traffic to the sample distribution on average compared with the 529 cities. Despite the dangle in traffic share, the latter remains very high and suggest that most of the earth ’ sulfur nautical natural process in fact concentrates at a limited number of urban places. Such a preliminary result already answers, at least partially, the initial hypothesis as a very high proportion of nautical flows concentrate at larger cities. The little decline over the period is attributable to the ejection of smaller cities from the Geopolis database, which tended to attract more traffic over clock. In accession, we calculated that the urban areas under study are three times larger on average than other cities in terms of demographic size. extra preliminary results are provided in table 1 and Fig. 3, which help to appreciate how much the geographic distribution of traffic and population has changed between 1950 and 1990. In 1950, most of the global ’ s nautical traffic concentrated in the North Atlantic region, whereas in 1990, East Asia had become the leave region. The comparison with population implies that the two main indicators haven ’ metric ton evolved along similar ways, as seen with the fantastic urban population growth for example in Latin America and South Asia that did not result in an equivalent dealings growth. Contrastingly in Europe, British port cities maintained their population but lost a considerable traffic share to the Le Havre – Hamburg range, the London case being a distinctive example of a declining port in that period, with the symbolic reconversion of its Docklands .Table 1 Distribution of vessels calls at the world’s demographically largest cities, 1950–1990 Full size board

Fig. 3figure 3 World distribution of vessel calls and population by urban area, 1950 and 1990. source : own realization based on Lloyd ’ s Shipping Index and Geopolis data Full size image last, the analyses proposed in this wallpaper rested on two extra calculations. One of them consisted in distinguishing six classes of urban areas based on their demographic size ( see Additional file 1 for a complete list of ports and urban areas ). Using quantiles rather of arbitrary thresholds ( e.g. over 1 million inhabitants ) avoided the possible bias caused by the general increase of city sizes over time, and consequently the incomparability of city-systems from one period to the early. Quantiles depend on different population thresholds between 1950 and 1990 but can be compared as each class contains the same proportion of cities, i.e. around 16.7 %. The second base approach is the measurement of orthodromic distances ( or great-circle distances, i.e. crow ’ s fly distances taking into report the sphericity of the Earth ) for each copulate of plug in urban areas in the nautical network. Such a measurement is very helpful to verify to what extent larger cities connect geographically far-reaching nautical forelands, as it was demonstrated earlier in the case of airports in airline networks ( Guimera et alabama. 2005 ) and of container ports in lining ship networks ( Ducruet and Zaidi 2012 ) but entirely in late times. Further research may consider using nautical distances in holy order to better respect the contour of continents and coastlines .

Rate this post

Bài viết liên quan

Theo dõi
Thông báo của
guest
0 Comments
Phản hồi nội tuyến
Xem tất cả bình luận