Space-Time Cube
- Introduced by
- Torsten Hägerstrand (time geography); cube popularized by Menno-Jan Kraak
- Year
- 1970
- Type
- Spatiotemporal data structure and visualization framework
- Family
- Time geography / spatiotemporal binning
What does it do?
The space-time cube is a framework from time geography for representing and analyzing phenomena that move and change over both space and time.
Two horizontal axes carry geographic location and a vertical axis carries time, so each observation becomes a point in a three-dimensional x–y–t volume and a moving object traces a continuous 'space-time path' through the cube. Introduced conceptually by Torsten Hägerstrand in 1970 and turned into a practical analytic and cartographic tool by Menno-Jan Kraak, it underpins modern spatiotemporal hot-spot and trajectory analysis.
The intuition behind the method
Stack a deck of maps, one per time slice, into a transparent block: the bottom map is the earliest moment, the top the latest. A person's day appears as a line threading up through the block — sloping where they travel, vertical where they stay put — so the geometry of the path itself encodes movement, stops, and timing. Aggregating events into a grid of space-time bins turns the cube into a 3-D histogram you can slice, query, and test for emerging clusters, revealing patterns invisible on any single flat map.
History
Torsten Hägerstrand, a Swedish geographer, set out the foundations of time geography in his 1970 address 'What about people in regional science?', introducing the space-time path and the 'space-time aquarium' to put individual human activity, with all its constraints, back at the centre of spatial analysis. The idea lay largely dormant as a conceptual diagram until computing made it tractable; Menno-Jan Kraak revived and operationalized the space-time cube for geovisualization in the early 2000s, and it was subsequently implemented in mainstream GIS as the data structure behind emerging hot-spot and trajectory analysis.
When to use it
Use the space-time cube when your data carry both location and time and you want to see or test how patterns evolve — movement trajectories, the spread of events, or shifting clusters of activity. It is the natural structure for trajectory and mobility analysis, for emerging hot-spot detection, and for communicating change to an audience visually. It is less suitable when time is irrelevant or when the data are a single snapshot, when the temporal resolution is too coarse to define meaningful slices, or when a flat animated map communicates the pattern more clearly than an occluded 3-D volume; binning choices also impose the modifiable areal (and temporal) unit problem.
Limitations and considerations
- Three-dimensional displays suffer from occlusion and perspective distortion, so dense cubes become visually cluttered.
- Results depend heavily on the chosen spatial and temporal bin sizes — the modifiable areal and temporal unit problems.
- Static images of the cube are hard to read; the tool's value largely depends on interactivity.
- Large trajectory datasets produce many overlapping paths that overwhelm both the display and the analysis.
Source conditions do not replace independent scientific review or validation by an analysis engine.
Strengths
- Unifies space and time in one structure, exposing spatiotemporal patterns that flat maps and time-series plots each miss.
- Space-time paths give an exact, geometry-rich record of individual movement, dwell, and timing.
- Binned cubes feed directly into rigorous spatiotemporal statistics such as emerging hot-spot analysis.
- Powerful for visual communication of dynamic phenomena, with interactive slicing and rotation.
Applications
- Analyzing individual mobility and activity patterns from GPS trajectories in transport and behavioural geography.
- Detecting emerging, intensifying, and diminishing hot spots of crime, disease, or accidents over time.
- Visualizing the spatiotemporal diffusion of phenomena such as epidemics, wildfires, or social movements.
- Studying animal movement and migration tracks in movement ecology.
- Exploring spatiotemporal accessibility and the time-budget constraints of daily life in time geography.
How it works
How it works
Every observation is embedded as a point p with planar coordinates x and y and a time stamp t, lifting conventional two-dimensional spatial data into a three-dimensional space where the third axis is time. This embedding is the foundational move that makes simultaneous spatial and temporal reasoning possible.
Formula
p = (x,\,y,\,t) \in \mathbb{R}^2 \times TAn object that persists is represented by its space-time path — the connected sequence of its (x, y, t) positions over an interval. The slope of the path encodes speed (steeper means slower or stationary), bends encode changes of direction, and gaps encode absence, so the path's geometry is a complete record of the object's movement history.
Formula
\text{path}_i = \{(x_i(t), y_i(t), t): t \in [t_0, t_1]\}For aggregate analysis, the cube is partitioned into a regular lattice of space-time bins — a spatial grid crossed with time intervals — and each bin is given the count (or summary) of the observations that fall inside it. This 'netcdf-style' space-time binning converts irregular events into a structured 3-D array suitable for statistics.
Formula
C(b_{xy}, b_t) = \#\{p_i : (x_i,y_i)\in b_{xy},\; t_i \in b_t\}Reading the cube column by column, each spatial grid cell yields a time series of its bin values across the temporal slices. Trends and clustering can then be tested per location and across neighbours — for example combining the per-cell time trend with the Getis-Ord statistic — to classify each place as a new, intensifying, diminishing, or sporadic hot spot.
Formula
z_{b_{xy}}(b_t) = f\bigl(C(b_{xy}, b_t)\bigr),\quad b_t = 1,\dots,TCommon pitfalls
- Choosing bin sizes arbitrarily and not testing sensitivity, so the apparent pattern is an artifact of the binning.
- Mixing inconsistent time zones or irregular sampling intervals, which distorts path slopes and bin counts.
- Over-interpreting the visual cube without the supporting statistics — perceived clusters may not be significant.
- Presenting a static screenshot of an inherently interactive structure, leaving viewers unable to resolve occluded detail.
Frequently asked questions
What is the difference between a space-time path and a space-time cube?
The space-time path is the trajectory of a single moving object — a line through the cube. The space-time cube is the whole three-dimensional x–y–t container in which one or many such paths, or aggregated event bins, are represented. The cube is the framework; the path is one kind of object living inside it.
How does the space-time cube support emerging hot-spot analysis?
By binning events into a grid of space-time cells, the cube produces, for every location, a time series of counts. Applying a clustering statistic such as Getis-Ord Gi* across space and a trend test such as Mann-Kendall across each cell's time series lets each place be classified by how its clustering is changing — new, consecutive, intensifying, persistent, diminishing, or sporadic hot spots — which is exactly what emerging hot-spot analysis reports.
Is the space-time cube a visualization or an analysis method?
Both. As a visualization it renders spatiotemporal data in an interactive 3-D display; as a data structure it is the binned x–y–t array that downstream spatiotemporal statistics operate on. Hägerstrand introduced it as a conceptual visual device, but in modern GIS it is equally an analytic substrate.
Related methods
Sources
Hägerstrand, T. (1970). What about people in regional science? Papers of the Regional Science Association, 24(1), 6–21.
Open DOI referenceCopied.Kraak, M.-J., & Ormeling, F. J. (2010). Cartography: Visualization of Geospatial Data (3rd ed.). Prentice Hall.
ISBN: 9780273722793Copied.