Everything happens at a place and occurs at some point in time. Space time analysis seeks to understand when and where (and sometimes why) things occur. With the evolution of ArcGIS Pro we now have the opportunity to not only successfully analyse patterns of time and space, but also immerse in a deep 3D visual experience.
ArcGIS allows you to exploit the space and time aspects of your data, allowing you to answer questions like:
- Is there an emerging hot spot?
- Are there any anomalies?
- Where your decisions or resource allocations effective?
Several tools, including Create Space Time Cubes, Cluster and Outlier Analysis, Emerging Hot Sport Analysis and Grouping Analysis assist with answering these difficult questions. So, let’s explore how we can utilise some these tools to analyse earthquakes that have occurred across South Australia. The quake data was sourced online from Geoscience Australia as a csv, recording the last 58 years of earthquakes across the State. Which I then subsequently created a feature class.
One of the easiest first steps you can do with a point layer in ArcGIS Pro is add it to a map and change the symbology to a heat map. This very quickly generates a dynamic heat map that focuses on the density of the points. You can also adjust the search radius and optionally specify a weight field (for example magnitude).
Let’s now take a look at some of the geoprocessing tools we can use to explore this space time data. The first step is to run the ‘Create Space Time Cube’ tool. This tool will summarise the earthquakes into a netCDF data structure by aggregating them into space-time bins. For example, this will sum up the number of earthquakes that occur with a specified distance tolerance at the same specified time interval.
The default parameters for the tool are fairly self-explanatory and are explained here. The hardest part is deciding what or if you need to specify anything for the optional parameters that can alter the cube dimensions. The first optional parameter – Time Step Interval, will often need to be considered and specified. As my data spans 58 years, I may want to consider aggregating the bin time slice to be 4 years. Another important parameter to think about is the Time Step Alignment when aggregating your data, because it determines where the aggregation will begin and end. I am going to specify mine to be End Time. The final optional parameter I need to consider is the Distance Interval. The distance interval is basically how large the space time bins should be. I am going to specify 100 km for mine. In most cases you will not know how to define the cube, the strong recommendation is to think about what is the appropriate dimensions should be for the particular question you are trying to answer. As my data covers a large area and earthquakes don’t exactly occur right at the same location I need to consider adjusting my bin dimensions to be relatively large. If you are looking at crime events across Adelaide for example, you may want to decide to aggregate points into 400 meter bins because that is about a city block. If the data spans a year, you may want to look at trends in terms of monthly or weekly. You will want to select a Distance Interval that makes sense for your analysis. Try to find a balance between making your distance interval too large and loosing the underlying patterns or too small so you end up with heaps of cubes filled with zero.
After running the Create Space Time Cube tool, it is important to check the results of the tool. Right clicking on the Message entry in the Results window and selecting View will display the results in a message box. The next step I will now want to complete is to visualise my space time cube in 3D. To do this, I will need to run the Visualize Space Time Cube in 3D. However, prior to this I need to ensure I have inserted a scene into my Project.
After running the tool, the output will add to your display. Allowing you to explore in 3D the space time cube. Viewing the space time cube this way allows you to easily see bins that have a higher count of events than other bins. The 3D space time cube can be further explored using the time slider options via the layer properties.
Let’s now take a look another tool, consuming the space time cube that I already created as the input. The Emerging Hot Spot Analysis tool will identify trends in the clustering of earthquakes, setting my Analysis Variable parameter as Count. To get a measure of intensity of feature clustering, this tool uses a space time implementation that considers the value for each bin within the context of the values for neighbouring bins. For Neighbourhood Distance I am going to specify 200 km and Neighbourhood Time Step of 2 (this will analyse the centroids of bins and consider them to be neighbours if they fall within the same neighbourhood distance and neighbourhood time step interval). The resulting layer now depicts trends in the data. The tool finds new, intensifying, diminishing and sporadic hot and cold spots. You can see an explanation of the how to interpret the results further here.
ArcGIS Pro, can assist with analysing patterns of time and space with the added bonus of visualising results in 3D. This is by no way implying that we can predict earthquakes, it just another perspective on how to review the data in ArcGIS. I strongly encourage you explore some of the great aspects of incorporating space time analysis as a potential method for exploring your data in ArcGIS Pro. Whether it be looking at trends in customer complaints, utility or equipment faults or even the spread of a disease; there could be many applications for this type of analysis. If you would also like to learn more about this topic we do also have a training course on offer – Exploratory Spatial Data Analysis and Geostatistical Interpolation.