Main Page

Open Street Map for Great Britain: OSM-GB - Initiated at The University of Nottinghams Nottingham Geospatial Institute (NGI) , the OSM-GB' project is a collaborative research programme with 1Spatial and KnowWhere. The overarching research purpose is to examine and develop the methods used to model the map features represented in the OpenStreetMap (OSM) crowd-sourced database in the context of Great Britain for the application of country-wide data quality improvements. Proposed outputs include the development of standard Web Services (Tile, WMS and WFS) over the improved data in a number of reference systems including British National Grid projection, a rules based catalogue for checking and improving the geometry and attribution consistency, a methodology for change-only updates to the OSM-GB database. In addition, the project aims to support the use and uptake of the OSM-GB product and to supply data and comments regarding data quality back to the OSM project.

This wiki is the home of sharing the guides, comments and documents about this project. In particular, there are special pages for for the implemented Quality Improvement and for using the Web Services.

= Project Overview = The project is divided into three board areas:
 * Infrastructure development and implementation
 * Rules catalogue and action research
 * Community engagement

Infrastructure development and implementation
The aim was to establish a web mapping infrastructure across multiple servers that is capable of importing large OSM datasets of the UK for storage, processing and delivery of the OSM-GB product. Processing includes application of the rules catalogue, re-projection from WGS-84 (World Geodetic System) into OSGB36 (British National Grid), with outputs provided to the public in WMS and WFS formats under the Creative Commons Attribution Share Alike (CC BY-SA) licence.

Rules catalogue and action research
This aspect centres around the application of 1Spatial's rules based geodata quality tools to the GB extract of OSM. Radius Studio will be used to establish a processing system that produces and feeds quality controlled OSM-GB data into the web mapping infrastructure.

Community Engagement
The OSM–GB team are keen to introduce this project to the established OSM community as well as the wider community. They have engaged with early adopters and a steering group so that end user and community needs can be established early on and to engage in a two-way exchange of information and ideas. Project updates are maintained via press releases, a live web page, blog, twitter and of course the project wiki.

=Infrastructure = The OSM-GB system operates over three servers. OSM-GB1 and OSM-GB3 are primarily used for the processing and storage of the project data, while OSM-GB2 is the public-facing server that will deliver the OSM-GB product and project information.

More details are available in Information Management page.

= Inputs =

OpenStreetMap
OpenStreetMap was founded in 2004 and is a crowd-sourced free-to-use product providing the means to access, use and edit geographical data and information at a global scale. It employs a peer production model similar to Wikipedia and is estimated to have 500,000 registered contributors globally (as at November 2011), mapping and submitting features such as roads, parks and buildings to be shared (Coast, 2011; Neis et al., 2011). The nature of the OSM data, including its freedom of use and distribution (see below), its attribute richness and the ability of the community to rapidly update changes in real time has placed OSM at the heart of many frontiers of research and technological development, including route application, 3D city modelling and Location-Based Services (LBS). However, issues with completeness, attribute ontology and consistency and geometric accuracy have rendered the product inaccessible / unusable to those agencies and organisations requiring a more authoritative dataset. OSM datasets are extracted by our region of interest (Great Britain), and downloaded and stored on our server ready to be processed.

Open Street Map License
OSM data is available for anybody to use as they see fit under an open license. There are, however, two conditions:
 * 1) 	The source of the data must be cited
 * 2) 	Changes to the data are released under the same creative commons license, CC BY-SA.

= Data Processing =

OSM Data Management
The use of “tags” to describe geographical features gives a semi-structured nature to the OSM data model, providing a simple yet rigid data structure that facilitates a huge amount of richness to be added easily to the product. However, this also provides a challenge to be overcome - the openness of the tagging allows any key-value pair to be associated to any feature. Something of a Meta data headache.

The scope of this project allows us to study six aspects of OSM data modelling:
 * 1) The conceptual data model: tags are key-value pairs associated with nodes, ways and relations. These are the three data primitives that OSM data consists of. Recommended key-value is describing geographical features as tags is provided here: http://wiki.openstreetmap.org/wiki/Map_features
 * 2) The way that data are structured within the OSM internal database:OSM has an established data structure for the internal storage of their data as illustrated in: http://wiki.openstreetmap.org/wiki/Database/Model
 * 3) The OSM-XML schema: this is the way that OSM makes the dump data files and update files, the schema can be found at http://wiki.openstreetmap.org/wiki/API_v0.6/XSD
 * 4) The storage of the OSM data in the OSM-GB servers for analysis: this is a trade-off between optimum required structure for rules application and the structure created by loader tools. Ultimately, Snowflakes Go Loader structure has been selected due to high performance and its simple and effective use of tags. Oracle Spatial 11g is the database being used because not only is the University of Nottingham an Oracle Spatial Centre of Excellence, but the Radius Studio tool has been proven to work effectively with this database. During the importing process, the OSM-GB team have kept to the original OSM conceptual model as possible.
 * 5) The storage of the OSM data in the OSM-GB servers for delivery to the public: PostGIS has been selected to store and deliver the most up-to-date version of OSM-GB. This is because, although it is a lossy and in the view of the OSM -GB project team, non-normalised, the openness of the database and its proven track record for adaptability has meant it can be efficiently used by the Mapnik tiling service. The structure is as defined by common conversion tools in the OSM community (e.g. Osm2pgsql and Osmosis).
 * 6) Data outputs: Web Map Service (WMS) and GML schema through a WFS service will be the main way that data is encoded and exported to the community. This is because of the widespread acceptance of these to open standards within the open geospatial community.

In addition, the team has been researching the practicalities and mechanics of importing and exporting to and from some of the above items. A summary is provided below.


 * Importing from OSM–XML to Oracle Spatial: This is the first stage of the process, and a critical step in the research project. All data validation is based around querying and optimised data structure of the converted OSM data. This shall be updated regularly. An array of importing tools were reviewed and included Osmosis/OGR2OGR, QuantumGIS plugins, FME and GO Loader/GO Publisher. Snowflakes GO Loader was settled upon the reasons outlined above, and was procured through an academic license. At is used to capture and store the OSM data for England, Scotland and Wales from the GeoFabirk’s dump OSM XML files into the projects Oracle Spatial database. Research is being conducted with the snowflake team in order to optimise the conversion procedure.


 * Importing from OSM-XML to PostGIS: Osm2pgsql was used to populate the PostGIS database from OSM-XML data in order to study and test the database structure. Although during the OSM-GB project the OSM-XML data will not populate the PostGIS database directly, but will first be loaded into the projects Oracle Spatial database for processing, it was necessary to become familiar with the PostGIS database structure. This is because Mapnik has already been successfully used to render OSM data out of PostGIS, and the OSM-GB project team wanted to emulate this already tried and tested process.


 * Converting between Oracle Spatial and PostGIS: a Java tool is to be developed that will ingest the OSM-GB product from Oracle spatial into PostGIS in a similar data structure as when Osm2pgsql is used to populate PostGIS from OSM-XML data.

OSM Quality Improvement
= Press Releases =

1Spatial and University of Nottingham collaborate on OpenStreetMap data quality project

Association for Geographic Information

RCS - 1Spatial Holdings - 1Spatial and University of Nottingham collaborate

1Spatial and University of Nottingham, OSM Project

= Papers and Conference Material =

How authoritative can the crowd be?

Can the crowd be authoritative?

OSM-GB - The Elevator Pitch

OSM-GB Stickers from the AGI Geocommunity Event 2011

= Project Wiki = This project wiki is designed to be a Web 2.0 tool for creating, editing and accessing the wiki-style materials related to the target of the project.

= External Links = Project website

Project blog