Generalisation and Multiple Representation

Production, Generalization, and Conflation

6704.2 - Road data conflation – the key step to geospatial data enhancement

Thursday, July 6
4:30 PM - 4:50 PM
Location: Virginia B

The quality of road information is always a vital part of GIS databases maintained by government or mapping organizations. Not only infrastructures are built around roads, but also roads are so essential on majority maps. Urban development often comes with road changes. It is critical to keep road data up-to-date to support city plan-ning, routing, and other spatial analysis in which roads are one of the important factors. Knowing that various free or commercially available road datasets are more accurate and up-to-date than existing data, many organizations are considering adopting road data from the new sources. However they are facing challenges in dealing with the spatial and attribute inconsistency between their legacy road data and the road data from the new sources. The common goals include to transfer the rich attributes from the legacy roads to the roads of better spatial accuracy and to replace the old roads by the changed roads. This process of reconciling multi-source data into one unified and optimal dataset is called conflation. This paper introduces the conflation tools and workflows in ArcGIS Desktop software and presents how some of them are used in different real world projects to complete conflation tasks. Road data conflation can be accomplished efficiently using highly automated tools and manageable interactive processes. It is the key step towards updating geospatial data for better quality.
The paper will be devoted to the following sections:
1. Introduction on conflation tools and workflows. It briefly explains the conflation tools developed in ArcGIS Desktop, based on in-house feature matching techniques. These tools can produce highly accu-rate results, while potential errors are automatically identified and flagged for review. Therefore, a typical conflation workflow consists of the automated steps and interactive inspections and corrections.
2. Real world scenario one: enhancing county roads by city roads. It presents how county road centerlines are enhanced by the more accurate city centerlines (Temecula, California), using a processing strategy: the Detect Feature Changes (DFC) tool finds matching features and line direction differences (estimated 98%+ accuracy); subsequent steps are taken to make line segmentation and direction consistent so the city road geometry can be transferred onto county roads with county road attributes and directions in tack. The gain in positional accuracy with reasonable time involved was worth pursuing.
3. Real world scenario two: attribute transfer between road datasets. It explains the workflow strategy for attribute transfer from source roads to more accurate target roads. The Transfer Attributes (TA) tool transfers a unique ID field from source to target, followed by automatic evaluations of the result which flag potential errors. Through an interactive review process, the transferred unique ID values are updated. The final transfer of additional attributes can be easily done through the common unique ID in both source and target data. This conflation work is typical needed by many GIS and mapping organizations and has proven efficient using this workflow.
4. Conflation of non-road features. Challenges and ideas for getting other features that don’t match newer sources conflated with roads will be discussed. Spatial adjustments may be applicable to move features while maintaining their relative positions with roads. Hybrid approaches with imagery or other reliable sources as references can be incorporated for updating existing features or extracting new features.
5. Conclusions and future work. Road data quality can be improved efficiently by conflation with better data sources. It is critical to take this step to lay the ground so other related features can be updated accordingly. The current conflation tools can be further improved by better pattern recognition and matching analysis and be extended to other feature types with the consideration of contexts. Our efforts on formalizing and optimizing conflation workflows will continue in a number of areas (details to be given in the full paper).

Dan Lee

Senior Product Engineer; Conflation Project Lead
Esri Inc.

MS degrees in Geodetic Sciences and Surveying (Ohio State Univ.) and Geography/Digital Cartography (Syracuse Univ.); BS in Physical Geography (Peking Univ.). Worked at Esri (20 years) and led project teams with research and development of generalization and conflation software. Involved in core Analysis and Geoprocessing team, responsible for tool designs, quality, workflows, and documentation. Active participant and contributor to ICA conferences, ICA Generalization and Multiple Representation Commission workshops, and related activities. Coordinated the Cartography SIG of Esri during 2002 - 2009. Worked as cartographic consultant at Intergraph (1990 - 1995) on generalization research and software product.


Send Email for Dan Lee

Nicolas Regnauld

Product Manager

Nicolas Regnauld has many years of experience in researching solutions for automating generalisation. After a PhD on generalisation at IGN (Paris), and a post Doctorate at the University of Edinburgh, he led the research team on automated generalisation at Ordnance Survey for 11 years. During this time, his team successfully developed a fully automated generalisation process that was used for the creation of OS VectorMap District. Nicolas is now Product Manager at 1Spatial, responsible for 1Generalise and 1Publish. These two products form the core of 1Spatial map derivation solutions, for deriving high quality maps from large geospatial databases.


Send Email for Nicolas Regnauld


6704.2 - Road data conflation – the key step to geospatial data enhancement

Attendees who have favorited this

Please enter your access key

The asset you are trying to access is locked. Please enter your access key to unlock.

Send Email for Road data conflation – the key step to geospatial data enhancement