Last week I continued in my quest to identify suitable data providers for my own application use. Because of the limited capacity of OpenStreetMap (OSM) and other free data servers to accommodate high request volumes I decided to investigate commercial vendors. What I discovered was that geographic data sellers essentially breakdown between providers of static, published maps and suppliers of raw data. Working with each has its advantages and disadvantages and you may well need both. This post will:

1. Define the differences between tile map providers and data vendors.
2. Describe the relative advantages between each kind of product.
3. How to choose whether you need access to published maps, data or both.

It is important to note that while most companies lean heavily in one direction or another there are no hard rules. Many map providers infuse additional data or make corrections prior to publishing; data vendors may also offer tile services.

Tile Map Providers

Static map providers create maps in the form of images in well known formats such as .png, .jpg and .tiff. Usually the images are a standard size (such as 256 X 256 pixels) and each image is edge-mapped with other images on its north, south, east and west borders (thus they can be said to be “tiled”). Often there is a pyramidal hierarchy of images to accommodate different zoom levels; therefore when you zoom in or out what you are viewing may be an entirely different image from what you started with (although styles across the hierarchy are usually consistent and therefore you may remain unaware of the switch).

Tiled maps are usually delivered live to customers in the form of a hosted web service. The customer’s website includes javascript code to requests the service and provides parameters such as zoom level and centre coordinates for the map to be displayed; the service returns all map tiles required to display the current geographic extent.

Often there is a cache of the most frequently requested tile extents to reduce server load; otherwise these exports are produced at need. A vendor API may extend capabilities to include services such as geocoding, routing or identifying locations within a search radius distance.

Easy Button – What You Can Do With Tile Maps

Map tiles are an easy way to add professional looking maps to your website. They offer the following advantages over a simple hard-coded map image:

1. The ability to zoom to different scales for greater or lesser detail.
2. The ability to pan the map to view different parts of the map extent while retaining current zoom level.
3. Maintain resolution across different zoom levels (maybe, if vector image formats are used or pyramiding across all zoom levels implemented).

The First One is Free

The “freemium” business model is often used by image tile providers in the form of offering a certain number of requests for free and then charging a fee for any overage. Other forms of revenue generation include:

• Offering tile maps for free butting charging for API services.
• Creating custom tile maps (e.g. custom styles, additional data layers) for a user.
• Premium accounts for faster access or higher maximum quotas.
• Hosting customer data (in the form of their own geographic data or results from services such as geocoding).
• Developer services to write the code that will be used on the customers website.

Data Vendors

Data vendors are the actual source of raw data that gets turned into maps and other geographic information products. What they sell are either photographs (aerial or satellite) or datasets of geospatial information that a computerized system can interpret as a map layer. Map layer datasets represent one homogeneous feature type (such as water bodies, types of vegetation cover or electoral boundaries).

For data at the national and sub-national geographic extent there many data vendors of various sizes. In many cases these are value added resellers – they obtain an original source of data and either enhance it (e.g. merge two datasets into one) or facilitate delivery (e.g. by adding metadata search or hosting on high capacity servers to support volume downloads). There are also a large number of free data providers – government agencies distributing open data or non-profits fulfilling a mandate.

Everything in the World is Exactly the Same

If you require data covering entire regions of the globe or the world your options are more limited. This is because anyone assembling a single collection of data applicable to the world has to merge data from 200+ national standards. This includes such tasks as transforming co-ordinates to a common map projection; ensuring features continuous across borders line up; layers share the same definition and have common attributes; data is all in the same format. The implication to do all this is the availability of significant resources and over time only a few organizations have managed to complete the task and still remain in the business. These entities are:

• NavTeq – Acquired by Nokia in 2008; the original base map supplier for Google.
• Tele Atlas – Acquired by Tom Tom in 2008; briefly took over as base map supplier for Google.
• OpenStreetMap (OSM) – Not a commercial entity but an open data not-for-profit that uses volunteers and donations to maintain a world map dataset.
• Natural Earth – Another not-for-profit that hosts data with a global extent that has entered the public domain.

Notably absent from the list is Google. While Google is a major player in the online published map industry until recently most of its layer datasets and images came from either NavTeq or Tele Atlas plus supplemental purchases from local providers. Only recently has Google migrated to become its own data provider and with the exception of aerial (but not satellite) photos does not resell its raw data. For the most part Google remains in the category of API services and tile map provider.

Horrible Freedom – What You Can Do With Dataset Layers

Because dataset layers are in a structure that combines feature coordinates with attribute information they offer more flexibility then the same data represented as a static image. What you can do with a dataset layer is only limited by the technology you are using and the expertise at your disposal. Some capabilities exposed not available from an image tile are:

1. Custom styling.
2. Dynamic styling (e.g. respond to a mouseover).
3. Filtering features shown (e.g. only show major highways).
4. Edit feature co-ordinates or attribute values (e.g. move a misplaced road segment; rename a road).
5. Query against an attribute (e.g. combine with filtering and dynamic styling to highlight major highways while keeping other road types visible).
6. The full range of spatial analysis offered by a GIS (Geographic Information System) – which gives you the power to derive entirely new layer datasets.

Note that tile maps are essentially products of layer datasets and images that have been overlain in an editing environment and then exported to image format. It is entirely possible to create your own set of tile maps if you have the raw data and appropriate software – whether it is worthwhile to do so (and then setup your own tile server) is up to you.

What Do You Need ?

The characteristics of each type of product described above already provide a hint at what each is best suited for. To make the list explicit, refer to the table below. The short answer is that tile maps provide the base layer that add geographic context to other data. You only need to mess around with satellite images and dataset layers if custom styling is important, querying is a requirement or new layers need to be derived.

Use a tile map service for ...	Use vendor data for ...
A professionally styled basemap underlying all displayed data.	Custom or dynamic styling required.
Support zooming and panning.	Filtering what geographic features are made visible.
Speed and responsiveness are a priority.	Query geographic features by attribute value or spatial relationship.
Minimize custom coding and complexity of technology.	Edit geographic feature attribute value or co-ordinates defining its position or shape.
Have a short discrete list of required spatial analysis services that can be fully accommodated by a vendor or third party API.	Derive a new layer more suitable as a basemap through processing raw images or other dataset layers.