Thoughts on Temporal GIS

Time has been and continues to be a mysterious entity for philosophers and physics. It doesn’t make any sense to talk about things in space with time, since no entity can exist without also having duration as one of its attributes. We also know from special relativity that space is intrinsically tied to time through the Lorentz Transformation, but this really only comes into play when traveling at extremely high velocities. However, there are many ways in which time is considered different from space. One can move in all directions in space but only in forward in time. I really liked Table 2 where Langran and Chrisman (1988) presented the analogies of GIS concepts between space and time. It helped me to clarify my thinking about temporality and how these two concepts can be combined.


While reading the article, I found myself asking the similar questions as Outdoor Addict. What constitutes as a change? And at what temporal scale should these changes be observed? For me, the first question is closely related to cartographic scale and must be considered in tandem with the specific research question at hand. If the research is concerned with the location of maple trees then perhaps a new map tree will constitute a change worth recording. Conversely, if the research question is concerned with the expansion or reduction in the size of a maple tree forest then the growth of one additional maple tree may not be enough to count as a mutation. An “event” in the latter example would be related to a percentage change in the forest boundaries. The cartographic scale one chooses will have a direct influence on the frequency of mutations and thus, on the appropriate temporal scale. This also leads to questions about precision for temporal data. When an event occurs, how precise should the temporal records be? To the minute? to the second? To the hundredth of a millisecond? It is likely that different kinds of events will have different temporal requirements.


Although I liked the map/overlay model the authors proposed, I imagine it is not the best way to visualize the data, especially when many polygons are undergoing mutations and data is collected at very fine scales (i.e. every 5 minutes).  For me, spatial-temporal visualization must involve some sort of animation/video that enables the user to select the speed it is played. This reminds me of the Agent-Based Models we saw in class.


Comments are closed.