How to map the frequency of overlapping points within an urban area?

Question

I've been asked to do a favor for someone since I'm "the mapping guy" and could use some theoretical help by way of examples since my actual cartography skills are a little rusty.

I have a series of points (~200) that represent events within a city over the course of a year. These events occur all over, but it's common that some points will overlap (e.g. events occurred at the same location).

I'd like to display all of the points, but in the case of overlapping points would like to highlight the fact that the frequency of events is higher at those locations. I can think of a few ways to do this (size, color), but could use some examples or information on techniques that I could use to turn this data into an informative map.

I'm currently using ArcGIS 10 to project and display the points.

Answer 1

Unfortunately, that other question is about web maps, which have a much more limited way to symbolize data.

With ArcMap 10, you have quite a few options. You could use graduated or proportional symbols, but that goes back to the color/size method you already mentioned. And truthfully, based on what you are describing, I think some form of chart symbology may be the most useful. You will likely need to create another field in your attribute table in order to get the results you want however.

I would probably populate this new field with integer values that represent the number of overlapping points (E.g., 3 overlapping points gets a value of 3). Then when you symbolize based on this field, your chart will give proportional weight to areas with high/low occurrence of events. A bar or stacked chart will probably display better than a pie chart.

Also, if you are concerned about multiple points trying to display charts at the same X/Y, you could just populate one of the coincidental points with a value and leave the others as zero. Then just use a definition query to not display the zeros, and that leaves all the rest that have >0 values.

By using this method, I'm assuming you aren't trying to distinguish individual events, but mostly trying to show intensity at a location?

It can be very difficult to cleanly show multiple symbols for points that occupy the same spatial location. That's why the other "duplicate" question answers mention clustering. Clustering can look really messy, though, especially if other points are nearby, but not the same X/Y.

Let me know if you need any clarification, or if I didn't answer your question well.

Answer 2

I would use an opacity level to deal with this. For example if you set the opacity (often referred to as an alpha level) to be 5; then if 5 points are completely on top of each other you would get a fully opaque dot.

An alternative approach could be to calculate the density of the points in ArcGIS (there is a function to generate a raster based on a few parameters such as raster size, interpolation type, etc). This would give you a surface which you could use, or trim to the areas where the points are.

Answer 3

I've had a need for a similar tool before too, but never took the time to put something together. So you're question made me think of ways I would do it, and the basics of what I came up with is:

1. Get a frequency of points sharing the same spatial location
2. Join the frequency back to the original point table
3. Then, as @Baltok suggested, use graduated symbols or colors to assist in the visualization.

It took a couple more steps than that, but I created an Arc10 model (works with all license levels) that accomplishes the task. My model leaves all of the original features and just adds a "Frequency" field to the end of each record that you can symbolize on. It does not take into account "clustering" of points within a distance, just purely overlapping.

Answer 4

What about using "collect events" from the spatial statistics toolbox; I think this is available at all license levels?

This will automatically count overlaps and symbolise with graduated markers, but you could use the count to symbolise any way you like.

Answer 5

A heatmap algorithm could eventually be applied to display the intensity of events. You will loose precision regarding events location but you will provide a better representation of hot areas in the city.

I'm not an ESRI guy, but I'm sure some heatmap libs are already available.

See some examples online :

• http://sethoscope.net/heatmap/
• http://www.jjguy.com/heatmap/

If heatmap can fit your needs, have a look on this Q&A.