Visualisation of spatio-temporal data set?

Geography, and later GISc have been struggling to incorporate 4th dimension since Torsten Hägerstrand brought time into geographic research.

Couple of things from the top of my head:

One of the solutions is to use 'space time aquarium' where in 3D space you can use X and Y to represent location in space and Z to represent time.

Two names that would be helpful exploring this approach would be

• Mei-Po Kwan - see for example paper here and some more figures here
• Menno-Jan Kraak - he calls it 'cube'; see for example here

From more recent stuff Fabian Neuhaus is writing his PhD @ UCL and his blog UrbanTick showcases interesting developments in this domain.

For more visualization inspiration you can check some projects from wizards of senseable city lab @ MIT, for example here, here or here.

One more eye candy I stumbled upon is Urban Mobs.

If you want to try building something on your own you might want to have a look at:

• timemap
• this ArcGIS extension
• GeoTime [commercial!]
• VisualEyes

You might be interested in my slides from a SXSW panel on geotemporal visualization. While they don't cover every single approach, they do a pretty good job of offering examples for the most common approaches (note that many of these examples require a browser with SVG or Canvas support, so not IE<9):

• Showing time as a line on a map
• Showing time as map data, e.g. to classify points or polygons
• Small multiples
• Allowing users to filter the map by time period
• Showing time via animation
• Geography as categorical data in a traditional time-series graph
• Linked map and time representations
• Time as the third dimension

My main points in the talk were:

• You generally have a trade-off between a good display of geography and a good display of trends over time. You need to figure out what's more important to you.

• "Flashy" approaches like animation and 3D are likely to be more engaging for the general public, but less useful for the specialist.

• Many good visualizations combine a number of these techniques (but use too many and it gets overwhelming)

If you're looking for a video output, commercial tools like EONfusion can make nice 3D environments with temporal information. Similarly, I agree Google Earth (and its plugin) are a simple tool for visualizing interactive temporal data. Visual Complexity keeps a database of network visualizations, many of these are spatio-temporal visualizations, such as the Gapminder World site, or the growth of Walmart.

From a technology perspective, the emerging use of HTML5, <canvas> and javascript may provide a rich interactive environment for visualizations which works directly on many devices, here's an example of global dependency using JS & canvas.

From a best practices perspective, I like much of Tufte's advice, here's an article examining his advice as applied to mapping.