The world is ever-changing, dynamic, and beautiful. For a long time, GIS did not capture this space-time dynamism, and instead was based on a static world-view.  But times have changed, and our ability to capture space-time dynamics and to analyze and model dynamic systems in GIS, has transformed.  

When you think of space-time models, you can envision different states of the system—how things are—at different points in time. At this point, you will approach something called hysteresis, which is the dependence of the state of a system on its history.

How can we undertake analysis of complex systems with hysteresis?  That’s the realm of space-time analysis.

What is Space-Time Analysis?  

First we might ask what is space-time data?  In a recent blog I discussed Big Geospatial Data – and we can define today’s space-time data as repeated observations through time on big geospatial data.  These are time series of imagery, census data, locations, health events and so on.  

Space-time analysis is the visualization, pattern recognition, and descriptive and predictive modeling of space-time data.  Space-time analysis seeks to understand the past and present in order to predict the future.  It is how we make sense of our complex and ever-changing world.  

Space-Time Analysis of Human Health

When thinking about human health, disease, environmental exposures, and genetics things get to be a bit more complex. Describing health-environment relationships in a GIS requires the 4 Ws: What, Where, When, and Why. And that is just the start. What else do we need to deal with?  

In a book chapter titled “Analyzing Cancer and Breast Cancer in Space and Time”, I identified several factors of health-environment relationships. Some are specific to cancer, but are common threads in most human diseases. In the section “Setting the Stage” I observed:

“Cancer poses a knotty problem in space-time analysis because the causes of cancer are complex, including environmental exposures, genetics, epigenetics and health-related behaviors; carcinogenesis is a multi-step process; and cancer latency, which is the period between initiating events (e.g. initial mutation) and diagnosis can be decades. This section seeks to familiarize the reader with aspects of this complexity. First, it considers cancer as a dynamic space-time system, in the context of both individual- and population-views. A conceptual model of carcinogenesis at the individual-level is presented for pancreatic cancer, and motivates the definition of cancer latency as residence times through states defined by the initiation-promotion paradigm. Next, other aspects of complexity are considered, including residential mobility and latency, biological periods of vulnerability, and temporal orientation in the space-time analysis of cancer.”

In that Chapter, I provided a series of examples illustrating how to conduct space-time analysis in SpaceStat. In a study of breast cancer accounting for residential histories over the life course, my colleagues found breast cancer clusters in Denmark in the Copenhangen-Odense area.

Analyses were conducted using adjustment for covariates (e.g. Unadjusted and adjusted analyses) and with two separate control groups.  This allowed the researchers to assess the sensitivity of the results to both known covariates (e.g. does age make much of a difference?) as well as the specification of control groups.  

The clustering persists even when covariates and different control groups are used.  In addition to true space-time analysis using Q-statistics in SpaceStat, spatial-only analyses in 1987 and 1997 (bottom row) replicated the results using a different method in the SatScan software.  This illustrates how space-time analysis can be used to localize areas of excess risk – here of breast cancer in Denmark.

Take A Deeper Dive Into Space-Time Analysis

In this blog, I’ve only touched on important considerations in space-time analysis, in an attempt to scope the problem and provide an Introduction.  Interested in more detail?  You can see a copy of this Chapter here



Jacquez, G.M. (2019). Analyzing Cancer and Breast Cancer in Space and Time. In: Berrigan, D., Berger, N. (eds) Geospatial Approaches to Energy Balance and Breast Cancer. Energy Balance and Cancer, vol 15. Springer, Cham.

Nordsborg RB, Meliker JR, Ersbøll AK, Jacquez GM, Poulsen AH, Raaschou-NielsenO.  Space-time clusters of breast cancer using residential histories: a Danish case–controlstudy. BMC Cancer. 2014;14(1):255