Maps, mashups and metadata: geospatial standards for access and retrieval

Chris Fleet, National Library of Scotland

Introduction

In the last few years, the opportunities for combining spatial information and maps together and presenting them through new, easily customised interfaces has dramatically grown. A whole range of spatial visualisations or mashups have arisen using Google Maps and Earth, Microsoft's Virtual Earth, NASA's World Wind, OpenStreetMap, OpenLayers, and Yahoo Maps, which can often be customised with relatively basic programming skills. These mashups can harness an exponentially growing volume of freely available geographic information, including maps, satellite imagery, and geo-referenced features.

The scale and variety of these initiatives continually grows. Programmable web acts as a directory, and guide to some of these applications, and lists over 1893 mapping mashups (January 2009). Everything 2.0 describes itself as "the most complete directory of Web 2.0 initiatives", and includes over 270 Geo 2.0 projects (January 2009).

According to the enthusiasts, the mashing of data from disparate sources has the potential to open up new avenues of research, exploration and understanding. Web 2.0 mapping utilities and spatial visualisations form an integral part of this process as content becomes more remixable, and services and tools more freely available (Macdonald, 2008) [1]. However, these technologies have also facilitated a huge increase in information noise, with cluttered and impossible interfaces, badly designed maps, and difficulties of retrieval. There are also some contradictions between the privately-owned, commercial world with commodified geospatial data, IPR and copyright, and the free, shared information commons encouraged by Web 2.0 technologies.

In the last year, the National Library of Scotland Map Library has made available a range of geo-referenced historical maps as mashups (http://geo.nls.uk/maps), overlaid on Google and Virtual Earth maps and satellite imagery. This paper describes the main open source spatial organisations and the standards they have developed that have made this possible, before considering the emerging metadata and retrieval possibilities surrounding these new technologies.

Open source spatial organisations

The Open Geospatial Consortium (OGC) is an international consortium of companies, public-sector agencies and universities who collaborate to develop publicly available interface specifications. These OpenGIS specifications further internet-based geo-data services and applications and include the Web Map Service.

The Open Source Geospatial Foundation, or OSGeo, is a not-for-profit organisation whose mission is to support and promote the collaborative development of open geospatial technologies and data. For example, OpenLayers is a free, open-source utility for displaying map data in most modern web browsers, requiring no server software. The Tile Map Service is an emerging de facto standard for splitting images into a hierarchy of smaller units for rapid delivery.

Standards

The most important Web 2.0 standards for delivering historical maps are as follows.

Web Map Service

The Open Geospatial Consortium's Web Map Service (WMS) enables the dynamic production of maps on the web using specified parameters, such as map layers, mapped extent, data format, and projection. A WMS Server typically generates maps in the standard graphics formats such PNG, JPEG, GIF, and TIFF for raster files.

Web Map Service operations can be invoked using a standard web browser by submitting requests in the form of URLs. The content of such URLs depends on which operation is requested. In particular, when requesting a map the URL indicates what information is to be shown on the map (LAYERS), what portion of the Earth is to be mapped (BBOX), the desired coordinate reference system (SRS), and the output image width and height:

    http://geo.nls.uk/wmsconnector/com.esri.wms.Esrimap?SERVICE=WMS
    &VERSION=1.1.1
    &REQUEST=GetMap
    &LAYERS=Scotland_Small,OS_Quarter,Bart_Half,OS_Popular
    &STYLES=
    &SRS=EPSG:4326
    &BBOX= -9.25805308169, 54.4515932002, -0.605194997975, 60.8904036625
    &WIDTH=900
    &HEIGHT=900
    &FORMAT=image/png
    &reaspect=false
    
Example of a Web Map Service URL request.

When two or more maps are produced with the same geographic parameters and output size, the results can be accurately overlaid to produce a composite map. The use of image formats that support transparent backgrounds (e.g. GIF or PNG) allows underlying maps to be visible, and individual maps can be requested from different servers. So the Web Map Service facilitates the creation of a network of distributed map servers from which clients can build customized maps.

The Google Maps API can easily display an overlay through standard WMS requests (see the page source syntax at http://geo.nls.uk/os6inch/google.html for example). This delivers set of layers of historical mapping, from 1:1 million through to 1:10,560, and can be easily mashed up with other applications, such as the Gazetteer for Scotland. Unfortunately, as the NLS' WMS has become more popular, so it has lead to slower response times, typically taking 10-15 seconds to refresh a screen, and therefore the quicker Tile Map Service alternative has been preferred for future development.

Tile Map Service

The Tile Map Service (TMS), developed by the OSGeo, provides access to rendered cartographic tiles at fixed scales. Unlike the WMS which relies upon software processing to deliver imagery, image tiles are pre-created at a set of zoom levels for a specific projection and datum for more rapid delivery.

The TMS takes advantage of the fact that Google Maps, Microsoft Virtual Earth, and other geo-data APIs use the same projection and tiling profile. The coordinates presented to the user are Latitude/Longitude on the WGS84 datum, whilst internally the applications use the Spherical Mercator projection, with coordinates in metres, and treating the earth as a sphere rather than the WGS84 ellipsoid. The extents of all tiles as well as the zoom levels are predefined for the whole Earth on the Spherical Mercator projection, with the only differences being the mechanism by which the individual tiles are described: the Google XYZ, the TMS zxy, and the Virtual Earth Quadtree format (see Pridal & Zabicka, 2008 [2] and also Tiles à la Google Maps: coordinates, tile bounds and projection)

In terms of format, each TMS tile is requestable by a URL, made up of a root server URL, then a zoom factor, then the x and y coordinates. Effectively the zoom and x coordinates are subdirectories beneath the root, with the y coordinate reflected in the tile name. For example:

    http://geo.nls.uk/maps/11/1092/1397.png
    breaks down into
    
    {server root}     / {sub-directory } / {sub-directory }/ {tile image}
    {geo.nls.uk/maps} / {11}             / {1092}          / {1397.png}
    
Tile Map Service Format: server-and-url/zoom/x/y.png

The tile model means that map access is rapid: previews and images can be loaded very quickly, rather than the usual time that traditional map distributions (such as WMS) take to render and re-orient.

Fairly simple Javascript code pulls the correct tiles as an overlay on top of whichever global mapping tiles are in the browser.

MapTiler

NLS has used MapTiler, free open-source software, as well as its former version, GDAL2Tiles. But other tiling software also exists, such as MapCruncher which allows geo-referencing and then tiling into 256x256 pixel image tiles. However, MapCruncher is not open source, and it is also unable to process maps that have already been geo-referenced.

Keyhole Markup Language

The Keyhole Markup Language (KML) is an official OGC standard with many geo-enabled web applications now able to display KML.

Javascript

Another important standard programming language is Javascript, used for customising most of the Application Programming Interfaces (APIs). AJAX (Asynchronous JavaScript and XML) allows web applications to retrieve data from the server asynchronously in the background without affecting the display of the existing image on screen.

Examples of WMS and TMS applications

The National Library of Scotland's Web Mapping Service for the Ordnance Survey six-inch maps illustrates WMS technology and source code behind it.

National Library of Scotland Web Map Service delivering 1930s historical mapping
Screenshot: National Library of Scotland Web Map Service delivering 1930s historical mapping.

It also illustrates the problem that retrieving tiles from a Web Mapping Service can be slow, due to the software processing required to generate the tiles.

Much faster results can be achieved by pre-creating tiles using the Tile Map Service.

Some recent examples of the Tile Map Service are:

The Ordnance Survey large-scale town plan for Glasgow (1857)

This is a geo-referenced mosaic of 206 individual sheets.

This uses OSGeo's Openlayers software, which allows use of non-Google map layers.

Mosaics for over 60 other town plans of Scotland can be viewed at http://www.nls.uk/maps/townplans/overlays.html.

The OS New Popular one-inch mapping for England and Wales of the 1940s.

This is a geo-referenced mosaic of 114 separate sheets.

Detail of the OS New Popular mosaic zoomed in on Heathrow (1857)

Retrieving geo-reference mapping

Although most geographical retrieval on the web and in library catalogues is text-based, based on word matching, geographical coordinates can overcome many of the limitations of text (including inconsistencies of names, changing jurisdictions over time, and the multiple ways that people describe geographic space). Geographical coordinates are increasingly being used for searching.

Geo-Referenced Search and Retrieval Using Co-ordinates
Geo-Referenced Search and Retrieval Using Co-ordinates.

For example, within the United Kingdom, the Vision of Britain project, or EDINA's Go-Geo demonstrates simple yet powerful geographic retrieval based on geo-referenced names, administrative units, and boundaries over time. Internationally, projects such as the Alexandria Digital Library, the Electronic Cultural Atlas Initiative, and the DIGMAP Project have applied these techniques more widely. By translating textual and map-based search methods into geographic coordinates, and then retrieving geo-referenced resources based on coordinate searching, many inconsistent and complex traditional and local geographic thesauri and vocabularies can be repurposed geographically.

Libraries are steadily furthering the use of geographical coordinates. From 2006, MARC authority records have been altered to allow precise geographic coordinates to be recorded, so furthering the automatic geo-referencing of library records with textual geographical subject headings (see MARC PROPOSAL NO. 2006-06). With coordinates in a GIS, users can search using text or graphics, and provided the information is also geo-referenced or geo-referenceable, retrieval of information can be more precise.

In theory, better retrieval still can come from the metadata recorded in the geo-referenced mapping itself.

Geo-referencing of historical maps consists essentially of matching a set of control points on an old map to real world locations, and then warping or transforming the historical map. For proper accuracy, it must be based upon a known projection and datum, and this can be recorded in the metadata with the image (See Appendix 1).

These geo-referenced maps can then be made available through WMS or TMS services, which themselves can deliver useful metadata. The WMS GetCapabilities request can retrieve a range of metadata, including a name, title, abstracts and URL, source and contact information and the extent of each layer. The TMS tilemapresource.xml file stored in the root directory of the tileset will record the spatial reference system, geographic extent and zoom layers of the tiles, and the resolution of each layer (see Appendix 2)

These retrieval technologies are in an emerging state - the Tile Map Service is not yet a defined international standard, relatively few online historical maps are geo-referenced, and fewer still are available as WMS/TMS services. Nevertheless, the value of this type of metadata will steadily grow in the future, and it can be quickly and automatically created and retrieved as part of the process of making available geo-referenced maps using these services.

For this reason it is suggested that institutions working in this area are better focusing on the long-term geo-referencing of maps, which can easily be presented through multiple outputs. Geo-referencing if done well, has a long-term enduring value, in a world where interfaces and the presentation technologies have a more ephemeral nature, and the metadata for them can be then quickly and automatically generated.

View the National Library of Scotland's Historical Map Overlays page (primarily Tile Map Services) for further examples.

Appendices

Appendix 1

Driver: GTiff/GeoTIFF
Files: c:\Lomond_S_export.tif
Size is 6643, 7930
Coordinate System is:
PROJCS[" Projection Name = Transverse Mercator Units = meters GeoTIFF Units = meters",
    GEOGCS["OSGB 1936",
        DATUM["OSGB_1936",
            SPHEROID["Airy 1830",6377563.396,299.3249646000044,
                AUTHORITY["EPSG","7001"]],
            AUTHORITY["EPSG","6277"]],
        PRIMEM["Greenwich",0],
        UNIT["degree",0.0174532925199433],
        AUTHORITY["EPSG","4277"]],
    PROJECTION["Transverse_Mercator"],
    PARAMETER["latitude_of_origin",49],
    PARAMETER["central_meridian",-2],
    PARAMETER["scale_factor",0.999601272],
    PARAMETER["false_easting",400000],
    PARAMETER["false_northing",-100000],
    UNIT["metre",1,
        AUTHORITY["EPSG","9001"]]]
Origin = (228721.187000000010000,701502.853000000000000)
Pixel Size = (3.000000000000000,-3.000000000000000)
Metadata:
  AREA_OR_POINT=Area
  TIFFTAG_SOFTWARE=IMAGINE TIFF Support
Copyright 1991 - 1999 by ERDAS, Inc. All Rights Reserved
@(#)$RCSfile: etif.c $ $Revision: 1.10.1.9.1.9.2.11 $ $Date: 2004/09/15 18:42:01EDT $
  TIFFTAG_XRESOLUTION=1
  TIFFTAG_YRESOLUTION=1
  TIFFTAG_RESOLUTIONUNIT=1 (unitless)
Image Structure Metadata:
  INTERLEAVE=PIXEL
Corner Coordinates:
Upper Left  (  228721.187,  701502.853) (  4d45'34.02"W, 56d10'31.42"N)
Lower Left  (  228721.187,  677712.853) (  4d44'39.21"W, 55d57'42.74"N)
Upper Right (  248650.187,  701502.853) (  4d26'19.41"W, 56d10'55.71"N)
Lower Right (  248650.187,  677712.853) (  4d25'30.96"W, 55d58'6.83"N)
Center      (  238685.687,  689607.853) (  4d35'30.87"W, 56d 4'19.55"N)
Band 1 Block=64x64 Type=Byte, ColorInterp=Red
Band 2 Block=64x64 Type=Byte, ColorInterp=Green
Band 3 Block=64x64 Type=Byte, ColorInterp=Blue
A typical GeoTIFF header with precise geographic metadata on the coordinate system, datum, spheroid, and related geographical information.

Appendix 2

  &lt?xml version="1.0" encoding="utf-8" ?&gt 
  &ltTileMap version="1.0.0" tilemapservice="http://tms.osgeo.org/1.0.0"&gt
  &ltTitle&gt Ordnance Survey 1:1056 Town Plan of Edinburgh mosaic, 1876-7&lt/Title&gt 
  &ltAbstract /&gt 
  &ltSRS>EPSG:900913&lt/SRS&gt 
  &ltBoundingBox minx="55.92116542336204" miny="-3.24989585974644" maxx="55.99966056322133" maxy="-3.12835862947301" /&gt 
  &ltOrigin x="55.92116542336204" y="-3.24989585974644" /&gt
  &ltTileFormat width="256" height="256" mime-type="image/png" extension="png" /&gt 
  &ltTileSets profile="mercator"&gt
  &ltTileSet href="11" units-per-pixel="292.45173230329510" order="11" /&gt 
  &ltTileSet href="12" units-per-pixel="584.90346460659021" order="12" /&gt 
  &ltTileSet href="13" units-per-pixel="1169.80692921318040" order="13" /&gt 
  &ltTileSet href="14" units-per-pixel="2339.61385842636080" order="14" /&gt 
  &ltTileSet href="15" units-per-pixel="4679.22771685272160" order="15" /&gt 
  &ltTileSet href="16" units-per-pixel="9358.45543370544330" order="16" /&gt 
  &ltTileSet href="17" units-per-pixel="18716.91086741088700" order="17" /&gt 
  &ltTileSet href="18" units-per-pixel="37433.82173482177300" order="18" /&gt 
  &ltTileSet href="19" units-per-pixel="74867.64346964354600" order="19" /&gt 
  &ltTileSet href="20" units-per-pixel="149735.28693928709000" order="20" /&gt 
  &lt/TileSets&gt
  &lt/TileMap&gt
Illustration of a tilemapresource.xml file.

References

Chris Fleet

Deputy Map Collections Manager

National Library of Scotland

e: c.fleet@nls.uk

All rights reside with the author(s). WIDWISAWN recognises the author(s) as the copyright holder(s) unless specifically directed otherwise.

WIDWISAWN is delivered via the SAPIENS e-publishing service and is hosted by the Centre for Digital Library Research at the University of Strathclyde

Design and layout © CDLR 2008

Last updated: Mar 2009