Geographic Information Systems Stack Exchange is a question and answer site for cartographers, geographers and GIS professionals. Join them; it only takes a minute:

Sign up
Here's how it works:
  1. Anybody can ask a question
  2. Anybody can answer
  3. The best answers are voted up and rise to the top

I'm conducting a density analysis of a series of polylines and I'm hung up on determining an appropriate search radius.

The default calculation performed by arcgis is to take the shorter dimension of the input feature extent divided by 30. However, I would prefer to use the formula set forth by Bailey and Gatrell in Interactive Spatial Data Analysis (1995).

They suggest that the bandwidth for a kernel analysis be set to 0.68n^(-0.2) where n=the number of features being analyzed.

This calculation is straightforward for points as each point is the same size. However, the polyline dataset I am working with has lines of varying length. Simply taking the number of lines as n doesn't seem appropriate.

I have considered converting the polylines to points but before I begin with this approach I am curious if there might be other valid approaches?

Thanks in advance for any help you can provide.

share|improve this question
+1 This is a great question. Bandwidths do matter, and there are indeed procedures to determine good ones as well as rules of thumb like the B&G one. But ultimately they are all in conflict, and this is because a good choice of bandwidth depends on the nature of the data and the purpose of the analysis. Could you perhaps share more information about what these polylines represent and what the density analysis aims to do? – whuber Dec 22 '12 at 17:59
Sure. The polyines represent river features. There is a separate layer of points of interest. The intent of the density analysis is to create a raster file on which the points of interest file will be overlaid. The raster value at each point of interest will then be extracted using the Extract Values to Point tool. Thanks. – scuerda Dec 22 '12 at 18:21
How will these values be interpreted? What is their meaning? What are they intended to tell us about the relationship between the points of interest and the rivers? – whuber Dec 22 '12 at 18:32
To be honest, I'm not entirely sure how they will be used. I'm doing this work under contract and only have a few details about the project. My understanding is that the intent to create a measure of how "in the creeks" each point of interest is in. My assumption is that the extracted value will ultimately be used in a regression model, but I don't know specifics. Sorry, not a very satisfying or useful answer but I've only been given limited information at this point. – scuerda Dec 22 '12 at 18:47
Ah! I see what you mean. The points in this case represent towns. It would be ideal to have some value derived from other research that would help determine a meaningful distance. But that data is not available. Thanks for clarifying. I have gone back to client to get clarification and I'll post back when I know more. – scuerda Dec 24 '12 at 20:34

Your Answer


By posting your answer, you agree to the privacy policy and terms of service.

Browse other questions tagged or ask your own question.