Creating a hydrographic network from its cartographic representation: a case study using Ordnance Survey MasterMap data

• Published in: International journal of geographical information science : IJGIS Vol. 20; no. 6; pp. 611 - 631
• Main Authors: Regnauld, Nicolas, Mackaness, William A.
• Format: Journal Article
• Language: English
• Published: Taylor & Francis 01.07.2006
• Subjects: Data modelling; Hydrographic network topological modelling
• ISSN: 1365-8816; 1365-8824; 1362-3087
• DOI: 10.1080/13658810600607402

A meaningful hydrological network is critical to spatial analysis and modelling. 'Meaningful' in that it is topologically correct, provides a basis for modelling flow and differentiates between different types of water features. In Great Britain, large-scale digital mapping of hydrological features was captured from paper maps and had a cartographic emphasis that had poor attribution, and no underlying model that supported geographical modelling. This emphasis gave rise to rivers and lakes that were variously 'broken' into sections by features such as dams, bridges, and culverts. This paper reports on research to create automatically a topologically connected hydrological network that underpins the detailed cartographic representation of such features. The network was created by joining these hydrographic features together according to rules of both continuity and proximity between river sections, and their flow direction (using an underlying digital elevation model). Confidence values were associated with each section link reflecting the certainty of that connection. The confidence values provided a basis for directing human intervention to uncertain connections as part of the final editing process. The project took as its input OS MasterMap 'water feature' data. A skeletonisation process was used to create the medial axis of the network. The paper reports in detail the methodology, the implementation and evaluation. The algorithm worked well in rural areas where interruptions are small and there is greater variation in height. In urban areas the challenges were greater where typically relatively long sections of river may be re-engineered and culverted, and where the fidelity of the digital elevation model was insufficient to discern the subtle changes in elevation. *Present address: Research and Innovation, Ordnance Survey, Romsey Road, Southampton SO16 4GU. Email: Nicolas.Regnauld@ordnancesurvey.co.uk