Skip to main content

E-IWW Census: Visualising Inland Water Volumes

Introduction

The UNECE E-Road and E-Rail censuses collect infrastructure and traffic information on the main highways and rail lines of international importance, as defined in the AGR and AGC agreements. This allows an international view on traffic levels, enabling an understanding of key international connections, potential bottlenecks and infrastructure planning.

A similar exercise is not yet done on the E-Waterways network, as defined in the AGN agreement, which spans for more than 29,200 km. In recent years, the Working Party on Inland Water Transport (SC.3) agreed that an E Waterway census could be a useful analytical tool for policymakers in member States and will give a better understanding of the international inland waterway sector, also allowing to complement information for inland transport modes.  SC.3 has requested that the Working Party on Transport Statistics (WP.6) explore this idea. Member States preferred not to consider an additional statistics collection exercise for this purpose, therefore the secretariat explored ways to produce census-like outputs for the E-Waterway network using existing data sources.

Activity Data

For European Union countries, Eurostat conducts data collection of many aspects of inland water transport, including regional origin-destination goods flows. These data are available in the iww_go_atygofl table which gives data for NUTS2 region origin-destination flows, on both a tonne and tonne-kilometre basis, and  breaks the data down by type of good. Data are not available for Italy, Lithuania or Sweden (nor for the United Kingdom historically), but otherwise cover all European Union states with meaningful inland water transport volumes.

Geospatial Data

To complement their extensive sub-national data collection, Eurostat publish the geospatial shapes of the NUTS2 regions. These can easily be combined with the activity  data. And from this point it is somewhat straightforward to combine the activity data with the geospatial features, and plot tonnage as straight lines between the centroids of each origin-destination.

IWW Origin-destination straightlines of tonnage

Based on administrative boundaries: © EuroGeographics. The boundaries and names shown and the designations used on this map do not imply official endorsement or acceptance by the United Nations.

 

This first visualization is too noisy to provide meaningful insights. With sufficient filtering, the main routes can be ascertained, but this will still leave the challenge of understanding the magnitude of each route, as similar trajectories will overlap with each other and not give a full understanding of tonnage.

An additional geospatial data source is available from SC.3, namely the Shapefile of the E-Waterway network available from the Blue Book database. Using this Shapefile of the real shapes of canals and rivers, the Eurostat data can be manipulated and summarised so tonnage is projected onto likely routes that the traffic will pass on.

Process

The following is a simplification of the process to visualise the inland water traffic on the real network.

  • Turn the Shapefile into a network. A shapefile is not a network in the mathematical sense. Networks have nodes and edges, and each node knows what it is connected to.
  • Translate each origin and destination point onto the nearest node of the network (based on geographical distance).
  • Break each origin-destination pair into multiple edges.
  • Sum up the tonnage on each edge of the network.
  • Plot these results with thickness based on total tonnage passing through.

Data Quality

There are many caveats to consider with visualizing these data using this approach, as the output will not necessarily agree with reality. Some are listed below, but there may be others to .

  • The method takes the shortest path along the network from centroids of the NUTS2 regions. This will not always reflect the spatial reality of where industries and transport locations are located.
  • Similarly, as the method uses the shortest path, no account is taken of canal width, river depth, bridge height or other navigability issues, which may affect which path vessels of certain sizes and specifications take.
  • The Shapefile of the Inland Waterway network only covers inland waterways, therefore River-sea shipping quantities may either be misrepresented (taking complicated inland journeys rather than a simpler maritime route) or not shown at all (as there is no possible inland water route between e.g. the River Rhine and the River Thames.)
  • Quantities shipped from and to the same region. Currently the origin and destination points for these cases are shifted to slightly different geographical locations so that at least the quantities should appear on the visualisation in some form. These special cases are not trivial; 2 of the top 10 origin-destination flows in 2020 relate to these cases (RO22>RO22 and NL33 to NL33). 
  • The Eurostat data are not complete, with Italy, Lithuania, Sweden absent, as well as the United Kingdom historically not included.
  • Other statistical differences.

Given these possible reasons for differences with reality, national sources were sought to make comparisons. There are only limited national sources that publish traffic data on specific parts of the network (that the secretariat is aware of). Two such sources are the French Voies Navigables de France and the German Statistical Office Destatis. Data for regions within France are available (for example). While Destatis publish a table with flows on named parts of their network (see table 46321-0013 here).

The secretariat has conducted limited order-of-magnitude checks with these sources. There are significant differences in the French data in the Nord-Pas de Calais region, with the produced map only showing volumes between Cuinchy, Lille and the Belgian border at Deûlemont, while the map from VNF shows significant quantities also to the sea at Dunkerque. This is likely due to the data quality issues listed above (although the quantities shipped in Lille appear to be similar to the national statistics, with quantities from the national source of around 5204kt at Lille compared to 5808 kt on the map.)

France Nord-Pas de Calais IWW volumes, 2020
France Nord-Pas de Calais IWW volumes from method, 2020

Source: https://www.vnf.fr/vnf/brochure-et-lettress/observatoire-transport-fluv…. The boundaries and names shown and the designations used on this map do not imply official endorsement or acceptance by the United Nations.

For Germany, the complexity of the canal network in the North of the country also lead to larger differences between the sources; for example, on the Elbe Seitenkanal, Destatis reports 6357kt and the map shows 4082kt. Meanwhile, quantities on the start of the Rhine river close to Strasbourg show a much closer alignment, with just a 4% difference between the two sources between Strasbourg and Karlsruhe.

Due to the above reasons, the map should not be considered an official statistics output. It is rather a visualisation of an official statistics dataset, that hopefully provides value in understanding the main flows on European waterways.

Results

Inland water volumes, 2020

Please browse the map of 2020 inland water freight volumes at : https://gis.unece.org/portal/apps/webappviewer/index.html?id=56e9186ac6e8400f8e3177957a72b9ab. The boundaries and names shown and the designations used on this map do not imply official endorsement or acceptance by the United Nations.

Uses

The secretariat hopes that the visualization can be useful in a number of ways.

  • Understanding goods quantities on specific network segments allows the value of the network to national and regional economies to be ascertained.
  • Having the statistics presented geospatially allows potential identification of modal shifting opportunities, especially when compared with other geospatial data from road and rail networks.

Further work in this field is possible, depending on the needs of member States. For example, trends over time or analysis by type of good can be produced. The data can of course be updated on an annual basis as well.

References

For an understanding of how to conduct similar geospatial analyses using an existing Shapefile, see https://r-spatial.org/r/2019/09/26/spatial-networks.html.