Computers, Society, and Nature

In their post on temporality in GIS, Outdoor Addict brings up the date of the Langran and Chrisman paper–1988.  That was a long TIME ago!  So this brought to mind for me, the same as it appears to have for many others, that there must be many innovations since 1988 allowing us to better represent time in GIS datasets.  The sliders on internet graphics, or like in agent-based modelling, where the slider moves, changing the time step and the data displayed; the same thing in Google Earth, you can witness both historical changes and the visualization as day turns to night–all are examples of how temporality is displayed today: digitally.

But these are ways of displaying data, and as someone noted, not necessarily the best way of analyzing data.  This made me think further, though.  At each time step, the “event” is static in that time.  The process is fluid across time, but the events are solidly placed within time.  So my question is: why must we do time differently?  Couldn’t we have one map, where red dots are 1998, blue dots are 2000, and yellow dots 2012?  We could see where time factors in, and the data could be in one attribute table.

I am analyzing landscapes for another project, and we are comparing 1998 to 2004.  We have two maps with essentially the same parcels, and are trying to compare the land use.  We have one attribute table with all the parcels, and then have the time steps as individual fields, listing the land use at that time.  It can be displayed at whatever time step we like.  I can see where the authors suggest this is loading a lot of additional data, however.  If you have upwards of 20 000 unique attributes, say, but 75% of them weren’t changing, you would still have to store each time step of land use where nothing was changing.  But as the authors note, it seems the space-time composite is the best way to go forward, as combining all the temporal events in one space/data set minimizes the chances for error.

So at this point I’m not sure… where do we go from here?

–sah

Temporal GIS absolutely fascinated me once I found out what it was through this paper. The idea that spatial principles can be applied to time interests me as it signals to me that my spatial information knowledge has an additional use. The descriptions of each “image of cartographic time” were extremely helpful in visualizing precisely what the authors were trying to explain.

However, for each method of thinking about geographic temporality, events or mutations are needed. Langran and Chrisman describe a mutation as “an event that causes a new map state” and “a point that terminates [a] condition and begins the next”. It theory this makes sense. In the real world what qualifies as a mutation or event? Take for instance a map of a suburb’s development. The first version may only have a few houses. The next might have new houses, new streets and a new school and the following one might show a new fire hydrant as the only change. At what point in time does the map need to be updated? What event is considered significant enough to warrant making an update to the database? Additionally, who decides this? Perhaps it might be similar to the argument on ontologies as it could be a subject specific database where particular changes are more closely followed than others. A fire department may be far more interested in updates concerning each fire hydrant than a family which may be more concerned about where the nearest park is located. Furthermore, is technology advanced sufficiently to be able to determine this on its own once parameters are set or is this a manual job? (For example, could a satellite constantly taking pictures of the suburb be programmed to recognize when 5 new houses are completed and automatically update the database to which it is connected?)

On a slightly different note I would like to emphasize the importance of going digital for temporal GIS. The authors only point out that their work focuses on “digital methods of storing and manipulating sequent states of geographic information” but neglect to explain why this is so important. Much like geolibraries, the concepts and theory to operate and organize them may have been present may years ago (this paper dates to 1988 while geolibraries date to 1998) but the technology did not exist to bring them to the digital world and make them practical, useful tools and studies. For the many reasons discussed for promoting digital libraries in addition to the nature of spatiotemporal information, digital is the only way to move forward.

-Outdoor Addict