One of the domains of GIS where still remains a lot to be done is the field of temporal analysis . While almost all the available tools focus on the spatial evolution of a phenomenon, we are left helpless when trying to visualize or analyse a phenomenon that changes as a function of time as well.
For some time, we have the support of animation tools that allow us to see a sequence of maps and perceive the change over time. Regardless of how useful they are for communication, they are useless to perform a serious data analysis. In this series of articles, we will discuss new ArcMap and ArcGis Pro tools available in the form of a Toolkit Tools model for the spatial-temporal exploration (Time Pattern Mining). An example has been selected in order to process, at first with ArcGis Pro, and, later, with ArcMap.
What is a Spatial-Temporal Cube ?
The following image summarizes the answer to this question.
On the left we have our points. They are distributed in the XY space, exactly the distribution we usually study and analyse by using the ArcGis tools. But we have several points located in the same place that correspond to different moments. This is the reason why the points are also distributed along the Z axis, the time axis. The space-time cube will allow us to analyse the distribution of our points on these three axes.
First important thing to remember: The tools that we are going to be using to analyse the spatial-temporal cube are not apt to perform the analysis of a given phenomenon based on its attributes. They are limited to the analysis of the occurrence or non-occurrence of a phenomenon (the point exists or does not exist).
The process is simple; we will determine the size of a box (bin), such as pixels or raster cells, but in three dimensions. And we will determine the number of points contained in each box. It is this number that we will visualize and analyse.
Let’s use an example in order to follow the process in detail. We have the weather forecast containing, that includes the gusty winds forecast. These forecasts are for a total of 8 days, every 3 hours, for an area that includes the Brittany.
A look at the attribute table let us see the attribute ” gust “, the speed of the wind gusts in knots, and, another field named Datehour, indicating the date and hour of each forecast.
Maybe one day it will be possible to analyse the distribution of the gust values as a function of time. But for the time being, we are limited to study the occurrence of the points, we can, for example, define a threshold of speed that we consider as risky and analyse the distribution of the points which exceed this threshold value.
If we define the threshold value as 20 knots we will create a new layer of data containing only the points exceeding this value. The points that do not exceed this value will not be present. We do that by selection of attributes, gust> = 20, and we export the selected points.
That’s it; we have now a layer to manufacture our cube.
Creation of the space-time cube
To create the cube, which will be stored under a NetCDF format ( .nc ), we will use the tool Create a space-time cube Toolbox Tools models for exploration spatial-temporal .
Once issued the order, the settings window appears.
The input entities correspond to our layer of selected gusts ( > 20 knots).
The space-time cube output will be our cube.
The time field will serve to define the Z axis and it must absolutely be it a Date type field. For those who are always with shapefiles, remember that the date fields can only contain the date but never the time. If you are going to work with time lapses of less than 24 hours (a day) you must have a class of geodatabase entities mandatorily .
For the time being we do not have count with a model cube to recover the others cubes information. If we make a series of cubes, and the first serves as model , we can be certain that the steps and the location of the boxes will be the same for all the cubes produced .
The time interval is the size of the cube boxes on the Z axis . In others terms, the time step that we are going to apply to our analysis. In our example , the time step of the forecasts is 3 hours . A size 12 hour box will produce 16 time intervals in Z. A box size of 24 hours, will produce 8.
One of the constraints of the cube is it must contain, at least 10 time lapses. Therefore we will choose so a time lapse of 12 hours .
The distance interval is the cell size in the XY plane.
If you do not select one or both intervals, the program will calculate one by default .
Visualization of the space-time cube
To visualize the cube you must Download a Toolkit containing two tools: one that allows to visualize the cube in 2D and another that allows to visualize it in 3D.
The Toolbox East downloadable at http://esriurl.com/SpaceTimeCubeUtilities.
Once downloaded and decompressed, you can add it in the geoprocessor by clicking on the ribbon Create -> Toolbox
We will see the visualization in 3D.
You must have a 3D window (Scene) in your project . If you do not, please make sure to, click Insert -> New Scene
You must remove the reference to a layer of elevation. If you do not, the columns of the cube will be displayed relative to the ground level and you will loose visibility of the correspondence of the different time lapses.
In order to do this, go to panel Content, click on your scene -> properties -> Elevation surface, deploy Ground and delete all services present.
Execute the visualization command of the spatial-temporal cube in 3D:
Point on the cube that you just created. Submit a name to your viewing layer.
In the variables to display, you will not have, at this stage anything but Count . In our next article we will discuss how to introduce other variables after having used the command analysis of emerging hotspots.
After some calculations and display process display , depending on your graphics card , you will have the result :
You can navigate in 3D on the stage to see where the time lapses with larger gusts (over the fixed threshold). But remember that it is only a visual Evaluation.
To determine statistically the number of overruns, the order Analysis of emerging hotspots allows a standard deviation ranking of the results.
We will discuss this topic in the next article.
Hello Atilio,
Do you say this .nc cube file created by GIS Pro is the ‘real’ netCDF file? I have successfully created space-time cube from locations, and I need to process this cube file in R together with other GCM netCDF files. I’m not sure if this will work. Thank you.
As I say int the post :”First important thing to remember: The tools that we are going to be using to analyse the spatial-temporal cube are not apt to perform the analysis of a given phenomenon based on its attributes. They are limited to the analysis of the occurrence or non-occurrence of a phenomenon (the point exists or does not exist).”