Explore a space-time cube

A space-time cube is a dataset that contains both spatial and temporal data, organized into bins. When visualized in 3D, the bins are stacked on top of each other along the z-axis to indicate time. Space-time cubes enable you to better understand and analyze spatiotemporal data.

In this tutorial, you'll use ArcGIS Pro to create and explore a space-time cube layer of homeownership rates in Florida since 2010. Your goal is to identify both where and when trends are occurring to help with regional planning efforts. You'll answer the following questions:

• How do homeownership rates change over time and space?
• Are there neighboring areas that have differing patterns of homeownership rates?
• How might other variables correlate to homeownership rates?

This tutorial was last tested on June 25, 2025, using ArcGIS Pro 3.5. If you're using a different version of ArcGIS Pro, you may encounter different functionality and results.

Requirements

• ArcGIS Pro (see options for software access)

Create a space-time cube layer

Space-time cube layers are created to visualize the contents of space-time cube netCDF (network Common Data Form) files. The space-time cube you'll use in this tutorial contains United States Census data from the American Community Survey (ACS).

Note:

Not all netCDF files contain space-time cubes. The netCDF file format is commonly used to store scientific data; these files cannot be used to create a space-time cube layer. To learn how to create a space-time cube netCDF file, try the tutorial Determine the most dangerous roads for drivers.

Create a project

First, you'll download the space-time cube netCDF file and create a project in ArcGIS Pro with a 3D scene.

1. Download Florida_Housing.zip.
2. Extract the downloaded zipped folder to a location of your choice, such as your Documents folder.

   The folder contains FloridaHousing.nc. The .ncextension is used for netCDF files, including space-time cubes. The file contains ACS housing and demographic data for Public Use Microdata Areas (PUMAs) in Florida. PUMAs are geographies defined by the United States Census Bureau that contain at least 100,000 residents.

   This space-time cube has been preprocessed in the following ways:

   • The Subset Space Time Cube tool was used to extract PUMAs in Florida from the original dataset, which covers the entire United States.
   • Analysis tools from the Space Time Pattern Analysis toolset were run on some data variables. Later in this tutorial, you'll explore the results of this analysis.
3. Start ArcGIS Pro. If prompted, sign in using your licensed ArcGIS organizational account.

   Note:

   If you don't have access to ArcGIS Pro or an ArcGIS organizational account, see options for software access.

   You'll create a project with a scene. Scenes display data in 3D. Space-time cube layers are used for 3D visualization of spatiotemporal data, so it's best to use a scene to work with them.

4. Under New Project, click Local Scene.

   

5. In the New Project window, for Name, type Florida Housing Trends. Click OK.

   The project is created and includes a scene with a default extent. Before you make the space-time cube layer, you'll turn off the scene's default elevation surface. Because space-time cube layers use the z-axis (elevation) to represent time, it's best to hide any 3D layers that may interfere with the visibility of the space-time cube layer.

6. In the Contents pane, under Elevation Surfaces, uncheck WorldElevation3D/Terrain3D or, if necessary, any other elevation surfaces.

   

Make the space-time cube layer

Next, you'll run a geoprocessing tool to create a space-time cube layer using the space-time cube netCDF file you downloaded.

1. On the ribbon, click the Map tab. In the Layer group, click the Add Data drop-down menu.

   

2. In the drop-down menu, click Space Time Cube Layer.

   

   Note:

   If you don't see the Space Time Cube Layer option, make sure that the active view is a scene, not a map.

   The Geoprocessing pane appears, showing the Make Space Time Cube Layer tool.

3. For the Input Space Time Cube parameter, click the Browse button.

   

4. In the Input Space Time Cube window, browse to the location of the Florida_Housing folder you extracted. Select FloridaHousing.nc and click OK.

   The file is added to the tool parameters.

5. For Output Feature Class (Layer Source), delete the existing text and type Florida_Housing_STC (STC is short for space-time cube).

   When you added the netCDF file, the Variables parameter populated with all variables associated with the space-time cube. Many of these variables contain census information that is not relevant to homeownership rates, so you don't need them for this tutorial. You'll select only a couple of relevant variables to include in the space-time cube layer.

6. For Variables, click the Reset button.

   

   All variables are deselected.

7. In the list of variables, check the boxes for the following variables:
   • PCTPOPULATION65YEARSANDOVER (percent of the population that is 65 years and older)
   • HOMEOWNERSHIPRATE (homeownership rate)

   

   The homeownership rate variable will be vital to your exploration. The 65 and older population variable, meanwhile, might show a potential influencing factor on homeownership rates (because homeownership generally increases with age).

   You'll leave the output geometry type as points. Points will appear as 3D bins in the scene and, in this instance, have better rendering performance.

8. Click Run.

   The tool runs. When it finishes, the space-time cube layer is added to the scene, showing cube-shaped bins across Florida. The space-time cube layer is only a visual representation of the actual space-time cube, which is in the netCDF file.

   

   Before you explore the scene, you'll investigate the details of the tool results to learn more about the space-time cube layer.

9. At the bottom of the Geoprocessing pane, click View Details.

   

   A details window appears with information about the tool results.

10. If necessary, scroll to the top of the Space Time Cube Layer Characteristics section.

    The space-time cube layer's data ranges from 2010 to 2023. The time step interval is 1 year, meaning each bin on the scene represents a year of data.

    

11. Close the details window.

Navigate the scene

Next, you'll navigate the scene to see what you can learn about the space-time cube layer.

1. Use your mouse or the navigation controls to pan, tilt, and zoom the scene until you can see an oblique view of the entire layer.

   Tip:

   To learn more about navigating a scene, read Navigation in 3D.

   

   Each column of the space-time cube layer represents a location. In this case, each location is a PUMA. Each bin (cube) represents one time step; as you saw in the details window, each time step is one year's data.

   The bin closest to the ground is the earliest time step, 2010. Each subsequent bin represents the next year, ending with the bin at the top, which represents 2023. Each column contains 14 bins, representing the 14 years between 2010 and 2023.

   The bins are symbolized by the variable value. The variable value is listed in the Contents pane, above the layer's legend.

   

   In this case, the variable value is the homeownership rate variable. Darker bins represent higher homeownership rates. Based on the legend, the bin with the lowest rate has about 13.9 percent homeownership, while the bin with the highest has about 89.1 percent homeownership.

   From this default visualization, you can already see some spatial and temporal trends in the data, helping you answer this tutorial's first question: How do homeownership rates change over time and space?

2. Pan, tilt, and zoom the scene until you can clearly see the southernmost column of bins.

   

   This column has medium-colored bins for the first 13 bins. The final bin, at the top, is darker. This pattern indicates that this area has recently experienced an increase in homeownership rate. The pattern is temporal, as all of the bins in this column represent the same location, just at different times.

3. Pan, tilt, and zoom the scene until you can see the cluster of columns around Jacksonville, in the northeastern part of Florida.

   

   Near the center of Jacksonville, there is a pattern of lighter-colored bins, indicating low homeownership. However, the areas outside of Jacksonville have darker-colored bins, indicating high homeownership. This pattern is primarily spatial; while there is some difference in rates within the same column, the biggest differences are from column to column.

4. In the Contents pane, right-click Florida_Housing_STC and choose Zoom To Layer.

   

   You return to the full extent of the layer.

5. On the Quick Access Toolbar, click the Save Project button.

   

   Tip:

   You can also save the project using the keyboard shortcut Ctrl+S.

You created a space-time cube layer and performed a basic exploration of its data. However, there is a lot of information in the scene, and it can be difficult to draw conclusions from the raw variable values. Next, you'll explore the space-time cube layer in more detail.

Visualize a space-time cube layer

For a deeper exploration of the space-time cube layer, you'll change its appearance and add a 2D visualization. You'll also focus on specific areas of interest and explore data variables that might have an influence on homeownership. Lastly, you'll explore data in the space-time cube layer using a time series chart.

Change the display theme

Space-time cube layers can contain multiple variables, as well as analysis results associated with those variables. For the purposes of this tutorial, your space-time cube layer was preprocessed and includes analysis results on the data. You'll change the display theme, which determines the way the space-time cube is symbolized, to explore these results.

1. In the Contents pane, confirm that Florida_Housing_STC is selected.

   

   When the layer is selected, the Space Time Cube contextual tab becomes available.

2. On the ribbon, click the Space Time Cube tab.

   

   This tab contains options for displaying and exploring space-time cube layers. In the Visualize group, the Variable parameter determines the variable that is symbolized in the layer. (As you saw previously, the homeownership rate variable is selected.)

   The Visualize group also includes a gallery of preset symbology options. The Variable Value option is selected by default. The options displayed in the gallery depend on the types of data in the space-time cube and any analyses performed prior to the creation of the space-time cube layer.

   During preprocessing, the Emerging Hot Spot Analysis tool was run on the data, so options to visualize the space-time cube using those analysis results are available.

3. In the Visualize group, in the Gallery, click the More button.

   

4. Under Emerging Hot Spot Analysis, click Hot Spot Type.

   

   The space-time cube layer is symbolized based on the results of the Emerging Hot Spot Analysis tool.

   

   The Emerging Hot Spot Analysis tool identifies spatial and temporal clusters of values that are either higher (hot spot) or lower (cold spot) than average. Hot spots aren't just individual bins with high values, but bins that are surrounded by neighboring bins with high values. As a result, some bins with lower values may be part of a hot spot due to their neighbors. Darker hot spots or cold spots indicate a higher level of confidence in the statistical analysis, with the darkest colors having confidence levels of 99 percent.

   Note:

   To learn more about hot spot analysis, see How Hot Spot Analysis (Getis-Ord Gi*) works. To learn how to run emerging hot spot analysis on a space-time cube, try the tutorial Determine the most dangerous roads for drivers.

5. Pan, tilt, and zoom to explore the scene.

   Based on the scene, you can see trends in homeownership rate that weren't obvious when looking at the values alone. For instance, much of central Florida is a hot spot for homeownership, indicating a cluster of relatively high homeownership values. By contrast, southeastern Florida, where Miami is located, is a cold spot of relatively low homeownership values.

   Additionally, there are some columns of bins where the earlier bins are white (not significant) but the later bins are hot spots or cold spots. Several columns on the edges of the Miami cold spot have become cold spots only in the past few years, for instance.

6. Right-click Florida_Housing_STC and choose Zoom To Layer.

Add a 2D layer

Analysis results for a space-time cube can also be displayed in 2D. You'll add a 2D layer containing emerging hot spot analysis results, which will have special symbology that summarizes temporal trends in the data.

1. On the ribbon, on the Space Time Cube tab, in the Utilities group, click Add 2D Layer.

   

   The Geoprocessing pane appears, showing the Visualize Space Time Cube in 2D tool. By default, the input space-time cube and variable are set to match your space-time cube layer.

2. For Display Theme, choose Emerging Hot Spot Analysis results.
3. For Output Features, delete the existing text and type Florida_Housing_2D.

   

4. Click Run.

   The tool runs and a 2D layer is added to the scene. It's difficult to see because of the space-time cube layer.

5. In the Contents pane, uncheck Florida_Housing_STC.

   The 2D layer shows PUMAs in Florida, symbolized based on the type of hot spot they are.

   

   Unlike the 3D results, where different shades of red or blue only indicated difference in confidence, the different types of symbols in the 2D layer refer to temporal trends in the data. Hot and cold spots can be new, consecutive, intensifying, persistent, and so on. This temporal aspect is what differentiates emerging hot spot analysis from regular hot spot analysis.

   Note:

   For a full explanation of every type of hot spot, see How Emerging Hot Spot Analysis works.

   Most of the large cluster of hot spots in central Florida and cold spots around Miami are intensifying hot and cold spots. Intensifying means that these areas have been hot or cold spots for most of the time steps, with an increasing intensity of clustering. This trend suggests that regional discrepancies between central Florida and Miami are increasing over time.

Focus on areas of interest

You now have a stronger understanding of your first analysis question: How do homeownership rates change over time and space? Next, you'll answer the next question: Are there neighboring areas that have differing patterns of homeownership rates?

Most hot spot and cold spot areas only neighbor similar areas or areas where no pattern was detected. However, there's one location where a hot spot and cold spot neighbor each other. You'll select these two areas to compare them.

1. On the ribbon, click the Map tab. In the Selection group, click the Select button.

   

2. While pressing Shift, click the two PUMAs in southwest Florida where a hot spot neighbors a cold spot.

   

   Based on the symbology, the cold spot is an intensifying cold spot, while the hot spot is a diminishing hot spot, suggesting that these two PUMAs are trending in opposite directions. You'll look at the 3D layer for these two locations to better understand these patterns.

3. In the Contents pane, check the Florida_Housing_STC box to turn the layer back on and click the layer name to select it.

   On the map, all of the 3D data reappears, making it difficult to only see the PUMAs you want to compare. Using the selection you made on the 2D layer, though, you can change the extent of the 3D layer.

4. On the ribbon, click the Space Time Cube tab. In the Spatiotemporal Extent group, click Configure Extent.

   

   Note:

   If you don't see the Space Time Cube tab, ensure Florida_Housing_STC is selected in the Contents pane.

5. In the Configure Extent drop-down menu, for Area of Interest Polygon, choose Florida_Housing_2D.

   

6. Click Apply.

   Because the area of interest polygon has a selection, the extent of the 3D space-time cube layer is changed to match the selection.

7. On the ribbon, click the Map tab. In the Navigate group, click the Explore button.

   

8. On the scene, navigate to the two columns of bins.

   

   The earlier time steps in the intensifying cold spot are cold spots with lower confidence, while the later time steps have higher confidence. By contrast, almost every time step in the diminishing hot spot has high confidence.

   Why is this area a diminishing hot spot, then? It's important to remember that confidence levels don't necessarily correspond to the intensity of clustering. Additionally, since hot and cold spots are determined by neighboring bins, the fact that the cold spot is intensifying might necessarily cause the hot spot to diminish, as its neighbors have increasingly low values of homeownership rates. If these trends continue, the diminishing hot spot might stop being a hot spot.

   You've examined two neighboring areas with different homeownership rates in detail, comparing them in 2D and 3D. Before you continue, you'll remove the area of interest and turn off the 2D layer, which you won't need for the rest of the tutorial.

9. On the ribbon, click the Space Time Cube tab. In the Spatiotemporal Extent group, click Configure Extent.
10. In the Configure Extent drop-down menu, for Area of Interest Polygon, click the reset button.

    

11. Click Apply.
12. In the Contents pane, uncheck Florida_Housing_2D. Right-click Florida_Housing_STC and choose Zoom To Layer.

Explore another variable

The last analysis question you want to answer is: How do other variables correlate to homeownership rates? According to a figure produced by the United States Census Bureau, homeownership rates tend to increase with age. Though homeownership rates can be influenced by many other factors, including race and immigration status, you'll focus on age for this tutorial.

When you created the space-time cube layer, you included a variable about the percentage of the population that is 65 and older. You'll explore this variable.

1. On the ribbon, on the Space Time Cube tab, in the Visualize group, click Variable and choose PCTPOPULATION65YEARSANDOVER.

   

   The space-time cube layer changes its symbology to reflect this variable. The display theme automatically changes back to Variable Value. Because emerging hot spot analysis was not performed on this variable, that display theme is no longer available.

   

   Darker bins indicate higher percentages of the population that is 65 and older. There appears to be more areas of older populations in central Florida, where the hot spots of homeownership rates were clustered, and fewer areas of older populations around Miami, where the cold spots were clustered.

   Changes in trends of older populations might shed insight into why homeownership rates are changing in some areas. You'll investigate trends in the data by changing the display theme.

2. On the ribbon, in the Visualize gallery, click Change Points.

   Note:

   Depending on your window size, you may need to click the More button on the gallery to see the Change Points option.

   

   This display theme highlights time steps where there has been a significant change in the trend in proportion of older residents. These time steps, called change points, are symbolized in magenta.

   

   Note:

   These change points were identified using the Change Point Detection geoprocessing tool. To learn more, see Change Point Detection (Space Time Pattern Mining).

   Change points indicate where and when changes occurred, but not whether these changes were increases or decreases. You'll investigate further using a time series pop-up, which shows time step values as a line chart.

3. On the scene, navigate closer to southwest Florida, where you compared the two neighboring hot and cold spot areas.

   There are a few change points in this area.

4. On the ribbon, in the Explore group, click Time Series Pop-ups by Location.

   

5. Click the most recent (highest) change point in the area.

   Tip:

   If you still need to navigate the scene to see the change point, you can temporarily reactivate the Explore tool by pointing to the scene and pressing the C key.

   

   A pop-up appears, showing the time series chart for the change point.

   

   The blue line represents the percentage of the population that is 65 and older for each time step. The magenta point is the change point, while the magenta lines show the overall trend in the data. At this change point, the percentage of people 65 and older was generally increasing from 2010 to 2019, then started to decrease from 2019 to 2023.

   If the age of the population is a factor in homeownership rate, this change point might help explain why this area is an intensifying cold spot.

   Tip:

   To learn more about a point on the chart, point to it.

6. Close the pop-up.
7. On the scene, click other change points and explore their time series charts.

   Not all change points reflect the changes you might expect. For instance, the other change point in southwest Florida shows the population of 65 and older decreasing until 2013, then increasing for the rest of the years.

   To better determine a statistical relationship between homeownership rates and the age of the population, you could consider performing some of the following analyses on the space-time cube:

   • You could run Time Series Cross Correlation to identify whether the 65 and older population has a correlated relationship with homeownership rates, and whether there is a delayed effect between the two variables.
   • You could run Forest-based Forecast to predict future homeownership rates based on the 65 and older population.
8. Close any open pop-ups. Save the project.

In this tutorial, you explored a space-time cube showing homeownership rates in Florida from 2010 to 2023. You learned how to create a space-time cube layer from a space-time cube netCDF file and how to visualize it in different ways using display themes, 2D layers, areas of interest, and time series charts to better understand the data.

You also answered analysis questions you had about the data, including how homeownership rates change over space and time, which neighboring areas have different trends in homeownership, and whether homeownership might be influenced by other variables. These exploratory insights revealed opportunities for further analysis that could help you understand and plan around these trends in the future.

To learn more about how to perform analysis on a space-time cube to answer specific questions, try the tutorial Determine the most dangerous roads for drivers.

You can find more tutorials in the tutorial gallery.

Acknowledgements

• The FloridaHousing.nc file was created from the United States Census Bureau's ACS data.
• World Topographic Map sources: Esri, TomTom, Garmin, FAO, NOAA, USGS, OpenStreetMap contributors, and the GIS User Community