described as “the democratization of GIS [geographic information systems]” (Butler, 2006) because it has allowed millions of users to become familiar with some of GIS’s very basic operations.
However, although virtual globes allow us to see a rough approximation of how the world looks (at the time the base imagery was acquired), Gore goes on to describe how a Digital Earth could be used to explore both the past and the future of the planet. Our ability to simulate the future relies on our understanding of the processes, both social and environmental, that shape the planet, and to do so at apparently increasing rates. Google Earth now has the ability to display a sequence of snapshots representing the development of some phenomena through time, and many mash-ups have been created to show historical data, but the ability to visualize future scenarios by exploiting our understanding of processes remains essentially futuristic. Enhancing forward-looking visualizations will require research in many disciplines (Craglia et al., 2008), including the geographical sciences, on topics such as the following:
Methods to use the spatial structure of virtual globes (discrete global grids) as the basis for a range of simulation models;
Methods of visualization that go beyond the current emphasis on rendering a realistic approximation of Earth’s surface appearance, and include properties that are abstract or nonvisual in nature, such as personal income, gross domestic product, or rainfall;
Standards and mechanisms that allow a user to search not only for data, but also for simulation models, and to implement them in a virtual globe environment;
Tools that allow models of a wide range of processes, from environmental to social, to be represented using a common and reusable set of software primitives;
Methods for downscaling predictions to the local level, so they can be made meaningful in a local context; and
Ways of enhancing understanding of the uncertainty associated with simulations of process, and of how that uncertainty can be displayed on a virtual globe and communicated to the user.
Today, detailed forecasts of global climate change are readily available in the scientific community, but the high level of technical and domain expertise needed to access them limits their value. Policy makers and individual citizens are far more likely to respond to such forecasts when they can see their implications locally. The idea of bringing the global message home, of localizing the global, was very much part of Gore’s original vision for Digital Earth—and far beyond what we can do today.
Because of the legacy of mapping relatively static phenomena, our methods of analysis are similarly geared to so-called cross-sectional data, or data of spatial distributions at one point in time. This limitation is exacerbated by the tendency for many social data-gathering exercises, such as the U.S. Census, to take place at fixed intervals. Remotely sensed images have also provided timed snapshots, although the effective frequency of overpasses has been improving recently as more satellites are launched, allowing the recovery effort following the Wenchuan earthquake in China in May 2008, for example, to make use of images collected from dozens of satellite sensors. Another reason for the paucity of lengthy time series has been the difficulty of maintaining the flow of public resources needed to keep large-scale, expensive government data-gathering programs in operation decade after decade.
As change accelerates and as sensor networks begin to provide densely sampled data in both space and time, we will need to add rapidly to our collection of spatio-temporal techniques of analysis. At this time, we know little about how to analyze and mine the increasing supply of data resulting from the tracking of vehicles, people, and animals (Miller and Han, 2001). We know little about how to assess the significance of an apparent change on Earth’s surface detected by remote sensing. We need a comprehensive battery of easy-to-use models to simulate a range of social and environmental processes, and to investigate the footprints they leave on Earth’s surface.
Even more urgent is the need for methods that can continuously monitor the stream of data coming from our acquisition systems, searching constantly for