Geoscience Reference
In-Depth Information
FIGURE 19.6
NASA's digital earth website circa January 2001. (From NASA/author.)
access. With the cost and size of memory falling at a rapid pace, it is easily possible to imagine
a situation where so much disc storage exists that it is highly redundant (sometimes as a back-up
mechanism) and suffers from the versioning problem (i.e. many versions of the same data set and
software functions must be stored, yet differences among them are minor). Retrieval costs and
time may rise, as the digital earth vision is fulfilled and new data sources and very-high-resolution
imagery become commonplace. Also critical is the handling of time. The current approach of keep-
ing a snapshot of a time-static data layer must yield to time stamps on objects, so that geographic
situations at a particular time can be reconstructed. The current generation of searching and spatial
indexing systems largely reflect research of the late twentieth century, and completely new and effi-
cient data structures and indexing may be necessary as we move forward (Samet, 2009). Generally,
as data projects become more digital and on-line, the incremental addition of creating new data will
give way to the demands for update and maintenance. Therefore, change detection and characterisa-
tion will assume new significance. Special purpose web mining tools that detect and label new roads
or settlements will be necessary or even mandatory. Cartography will enter an era when geospatial
data are compiled, identified and indexed automatically, in much that same way that web pages are
currently handled by search engines.
Fast-forwarding the issue of data fusion to 2061, the following may be possible. Images are
likely to no longer be
dumb
arrays of pixel intensities but instead embed software that continu-
ously searches for, identifies, labels and extracts geographical objects. Just as today's Microsoft's
Photosynth and the open source Bundler software (Snavely et al., 2006) mine images from Flickr
and elsewhere to extract viewing geometry and create 3D objects (Figure 19.7), a highly dis-
tributed array of webcams, geosensors, imaging systems and new data streams will be continu-
ously checking on and updating the content of digital earth, bringing interesting changes and
