Eonfusion Tech Blog August 2011
We continue the Eonfusion technical blog by looking at how Eonfusion deals with time varying datasets. In the previous article, we saw how attribute values can freely vary over features, so in this article we consider what happens when time values vary across a feature.
In the video below, we show a complex 3D environment which plays host to a predator-prey fish behavior model. The environment is represented by a complex surface that varies over X, Y and Z and defines the limits of where the modeled fish can swim. The fish themselves (with predators shown as large purple objects and prey shown as small white objects) have time-varying behavior which means that each fish feature varies in X, Y, Z and time.
The video shows the fish frozen at a moment in time, but also the track of where each fish has been in the recent past. Under the hood, Eonfusion is representing each fish as a line that has X, Y, Z and time values varying along its length, which is why we can show a continuous track of its history.
The traditional approach to representing this type of time varying data in a GIS package would involve taking numerous “snapshots” of the data at discrete time intervals. The disadvantage to this approach is that we have no information about the whereabouts of each fish between these time intervals. By representing each fish as a line feature, Eonfusion can determine where the fish is located at any point in time.
There are a couple of big advantages to adopting this approach to time-varying datasets. Firstly, information about individuals is often collected from different instruments at different time intervals. This is a real headache for the “snapshot” approach to data storage, since the snapshots from each instrument will be unaligned. In Eonfusion, the disparate data about a single individual can all be fused onto the one track and interpolated appropriately so that any attribute value can be determined at any point in time.
Secondly, having a continuous representation of an object’s track through time (as opposed to discrete samples), allows us to determine the time that critical interactions occur more accurately. For example, the precise time at which a fish comes within range of a predator can be determined (via an intersection operator). By contrast, the snapshot approach to representing our fish track can only tell us whether the fish is inside or outside of the predator’s range at each of the time snapshots, it can’t tell us the time that the transition occurred or whether the fish passed right through the predator’s range and out the other side within a single snapshot interval.
The video below also shows a dynamic view of the fish model as it progresses through time. Note how the fish move continuously through time rather than by fixed time steps and how the fish tracks avoid passing through the solid environment. This video also introduces one of the upcoming features in version 2.3 of Eonfusion - the time instant visualizer for lines – allowing lines that vary through time to be visualized as a single point positioned at the leading edge of the time slider.
