Hacks 87-92

Many of the hacks we've looked at create maps of new, previously unmapped data, such as Wi-Fi usage, your GPS tracklogs, or the news. Most of these hacks are built on top of other maps that give them context. Usually they have a step like "Go to http://example.com/data/download to fetch a certain file." The web sites with the map data are all over the place, and the maps are often in different formats. GIS professionals are used to spending a large amount of time simply finding the data they need to do their actual work.

The notion of a geospatial web seeks to address this problem. Already we have seen the shift the Internet has brought toward textual information, with unheard of quantities of documents a search engine away. The vision for geospatial information has been referred to as a "digital Earth," where you could simply input a point on the globe and have instant access to all mapping data available at that pointnot just what the U.S. census gathered five years ago as TIGER street data, but also the city's zoning information, the current weather map, Rich's favorite bike routes, Jo's corrections on the TIGER street data, public Wi-Fi nodes, and Anselm's social network. Every map created with the help of this book could be available to users online, searchable by coordinates and/or information about the map, and instantly available.

The most immediate effect of this would be more customizable maps; users would be able to select which layers they wanted to display. A map of Wi-Fi nodes could be overlaid on street data, along with the coastline and a common route to work. But the geospatial web could go much further. The Internet has the potential to serve as the transport mechanism for the next leap in mapmaking. GIS as a discipline has essentially passed by most of the potential of computers in favor of traditional mapmaking. The maps currently available on the Internet are essentially paper maps with a nice zoom button. Much more can be done, but only if it is much easier to find and actually use the vast quantities of real spatial information.

The problem with current Internet maps is that you can't really do anything exciting with them because they do not let you have the data behind the maps they provide. You cannot plot your own information on top, combine one map with others, or do any analysis on it. This is because it is a mere picture of the geospatial data. All of the major Internet mapping sites pay large companies like NavTeq for their data. If they gave you real information about the location, then that data would be available to all, since you could systematically step through the U.S. and download a map database. And NavTeq would no longer be able to make money. The geospatial web seeks to make the real data about our world available, which would provide a base for the next leap in mapmaking. It is the canvas upon which new ways of mapping and seeing the world can be painted.

With the advent of the geospatial web, users could customize their display, showing only Wi-Fi nodes that are public or on a particular provider's network. They could get a map layer with all the restaurants in the city and customize that layer to only display the ones that serve Mexican food that someone in their social network has given a five-star rating to, which are within walking distance of their current location. With advances in mobile devices, users could simply ask their PDA for the walking directions to the restaurant selected.

Mapmaking has traditionally been used as a measure of control. Philip II of Spain thought his maps sufficiently subversive to keep them under lock and key as a state secret. To this day, many governments hoard their geospatial data, treating it not as a public utility but as a source of profit. And even in countries like the U.S. that have a general policy of opening their data, it is not always easy to find all the information. A geospatial web would render the commons of information about our world truly useful, making the power of that information accessible in seconds to anyone with an Internet connection. "If a picture or map is a worth a thousand words, then power in the realms of representation may end up being as important as power over the materiality of spatial organization itself." (David Harvey, The Condition of Postmodernity) The geospatial web would put that power in the hands of everyone.

Making all mapping data available to anyone with an Internet connection sounds great, but wouldn't it be a lot of work to actually make that happen? Thankfully a number of smart people have been thinking very hard about this problem for a number of years and have been working together to evolve standards to enable a spatial web. The Internet was around for 20 years before the explosion of the World Wide Web, and arguably the key that allowed that to happen was standards. The disparate programs that operated on the network agreed to operate in a standard way. HTML, HTTP, and TCP/IP are all openly specified for anyone to implement. In the geo world, the OpenGIS Consortium (now the Open Geospatial Consortium, or OGC) has emerged to write specifications for the geospatial web.

The OGC is "a non-profit, international, voluntary consensus standards organization that is leading the development of standards for geospatial and location based services" (from http://opengis.org/). It was founded in 1994, with its roots in the Open GRASS Foundation (OGF), which was created by the GRASS user community to drive a process for the management of GRASS affairs. The OGC emerged to define standards to allow geoprocessing systems to communicate on the Internet through a set of open interfaces. Since 1994, the OGC has grown from 20 member organizations to over 250 from all over the world in commercial, academic, nonprofit, and government sectors.

All standards that the OGC adopts are freely available and directly downloadable from their web site. The process of creation, revision, and adoption is open to any participating company, agency, or organization. Unfortunately membership does cost money, and thus the process is a bit more closed than some other standards organizations, but the OGC is the best hope the geospatial world has, and it does a good job of being as open and inclusive as possible, encouraging participation from varied organizations.

The two most useful OGC specifications, from an end user's perspective, are the Web Map Service (WMS) and Web Feature Service (WFS). Both are web services; instead of returning a web page, they directly return the information the user requests. Both operate over HTTP, so all web browsers, and indeed all clients that use HTTP, can issue requests. Web Map Service is the more established specification; a large number of vendors offer WMS interfaces to their products. The interfaces for WMS and WFS are similar; the main difference is what they return. A WMS returns an image, as a JPEG, a GeoTIFF file, a bitmap, SVG, and so on. It operates like most mapping sites on the Internet, but the key is that it is an interoperable interface, meaning a single client can request maps from all sorts of different servers on the Internet.

The WFS spec does not return images, but raw data in GML (Geographic Markup Language), an XML data format. This allows clients to do far more, as they have the real geographic data, not just a picture of it. A WFS also supports complex filtering operationse.g., "Give me all street names that start with S and are more than two miles long." There are extensions to WMS that allow it to do similar things, but in WFS, you can just request the information directly. The WFS specification also supports Transactions, which facilitate user-supplied changes to GML data, making WFS a generic interface for keeping data up to date. Transactions also allow users to update the maps themselves. Figure 8-1 shows a world map of users of GeoServer, the popular free WFS server. It's viewable online at http://www.moximedia.com:8080/imf-ows/imf.jsp?site=gs_users.

Figure 8-1. GeoServer users shown via a Web Map Service

The display is done with a MapServer WMS; when you query it by clicking the "i" button, a WFS command is issued, which gives more detailed information than a WMS can provide. The site also allows you to add yourself to the user map, which issues a WFS Insert request, putting your record in the backend database that the WMS also makes use of. Unfortunately the IMF client site used in this example is not open source, but the backend servers all are.

AskTheSpider (http://www.askthespider.com/) is a geospatial data-discovery service that allows you to search through a registry of different services, using the WFS and WMS protocols to publish on the Web and find different catalogs of geodata.

Right now, WMS and WFS are for early adopters. Why not join them?

Chris Holmes