I'm working on a project for a conservation plan. Part of the plan requires that areas added for conservation have less than one structure per five acres (these are pretty large, mostly undeveloped areas). I've done kernel density and and point density analyses before but always with many, many more points. In the past I've had problems if there are not enough points. Should I even be looking at density analysis options? Would Thiessen Polygons or a cost surface be a better option?

2 Answers 2


You could convert your points into a raster using the option count and perform a focal analysis on this raster. See ArcGis Help here and here. Choose a circular or square type corresponding to the size of 5 acres and use sum as statistic type. The result of this focal analysis will be a raster with the density of points per 5 acres.


I think that using point density would be fine (not Kernel). Use the buildings as input and set search radius to about 263.5 feet, which is the radius of a 5 acre circle (assuming you are using a circle-typed neighbourhood). Than if your units of choice are feet as well, than any value higher than 4.591e-06 will represnt an area (or a raster cell) in which density is higher than one building than 5 acers. You could use raster calculator to show only the areas that should be added to conservation, using the <= operator.

In a brief, point density tool counts the points within your serach radius defined neighbourhood (with no weights applied) and divide it by the neighbourhood area, namely 217800 squared feet. Hence the thershold mentioned above.

  • This actually converts the density criterion into a minimal distance criterion, which probably differs from what was intended. For instance, if all buildings within a given parcel are clustered at one end, that very well might still qualify as having a sufficiently low density for the parcel--but not according to the calculation you propose.
    – whuber
    Mar 19, 2015 at 21:55
  • 1
    @whuber, you are right, in that case - if the territory is already divided to parcels that are un-changeable, it might be better to use spatial join between parcels and points to get the counts of all buildings within a parcel and divided by its area; yet no parcel feature is presented as available data in the question above. It is also worthwhile to check zonal statistics on the output of the method suggested above with parcel data, if available.
    – dof1985
    Mar 19, 2015 at 21:59
  • Yes, that's an interesting point for the OP to clarify. I think your proposed spatial join solution might very well be what was intended by the conservation plan.
    – whuber
    Mar 19, 2015 at 22:01
  • @whuber, just added an edit for my comment, I stick with density analysis, and raster calculator to find "built" cells, than using zonal statistics as table (or just zonal) to count those cells and divide the built area by parcel area...
    – dof1985
    Mar 19, 2015 at 22:03
  • Yes, that's a good use of zonal stats. I believe one has to take some care to make sure that multiple buildings are assigned to distinct cells, though, for otherwise ArcGIS may undercount them.
    – whuber
    Mar 19, 2015 at 22:05

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.