Density Surfaces
Many types of event or count data are collected at points
- population in census tracts
- earthquakes
- bird counts
- rainfall and snowfall
- pest traps (gypsy moths, for example)
- air quality (e.g., PM10)
- well samples (for water quantity or quality)
- houses with asbestos
- hazardous waste sites
- volcanoes
It is often desirable to know how many whatevers occur per area, like this example of landslides triggered by a major earthquake in New Zealand. A Density analysis counts the number of points within a search area, and is one of two ways that you can go from points to a surfaces. The others are to interpolate a surface from the values of the points themselves, or use Thiessen polygons or allocation. Which of the above are appropriate for analyzing “density” and which would you use another?
For the ArcGIS density analysis, you have the option of choosing the search radius, and whether the summation used in “count” is just the number of points or the value of some field in the attribute table associated with the point. Open the demo\density\density project. First, try it by analyzing this week’s earthquakes in the US. Then we’ll look at the population data for Rockbridge County, where each point has a value associated with it..
In this example, I’ve tallied the number of residents from the Census point file called “Population_1990” that contains a value for each point, instead of simply counting the number of points. The output is persons per square mile as a grid, with the following inputs to the “Point Density” tool.
Make sure you control the output coordinate system! Notice the impact of the search radius on the output. You can see the effect easily. What happens if you make the search radius bigger, or smaller?
Try the streams in a “Line Density” analysis as well. Make sure your search radius is much larger than the average spacing between streams (well…. try it both ways, smaller AND larger; which is of more use?)
We can collect some earthquakes from the USGS in real time and see where they are most dense. Or we can use the earthquakes I collected some years back. What radius do you use? Note what units and projection the earthquake map uses.
Kernels
The density analysis above uses a uniform search area to count or average the point/line density. But the search area can also have weights, or a kernel, that is applied throughout the search area. The Kernel Density tool does that. It emphasizes the close points and gradually the influence of further points is diminished in a bell-shaped curve. Try that with the same radius circle as the Point Density tool.