# Performing neighborhood analysis?

I have a point map of trees along with their size (diameter) as an attribute and I want to create a raster that with the output of a neighborhood analysis

Basically for every point in a raster of specified resolution, I'd like to make the following calculation

For every tree within radius r, sum up for all trees.

``````(tree size)^2 / (distance from center)^2
``````

Note that this is getting the quotient for each tree individually, then summing the up and assigning the middle pixel that value.

I can't picture this working with with the raster calculator or focal analysis tools, so I think I'd have to program something, but I'm not very experienced using Python.

Any suggestions?

I am using ArcGIS 10. And have tried various ways of looking at both Focal Statistics and Point Statistics. The problem with both of these is that they summarize on a single attribute. That is, I can get point densities for a given area, or I can sum up tree sizes for a given area... Dividing those two layers dones't quite give me what I need.

If I were to do this by hand, I would repeat these steps in a grid every 5 meters

1. Identify all trees within radius
2. Get size of each tree
3. Divide each tree by the distance^2 from the center
4. Add up the quotients for all trees within radius
5. Assign that sum to a raster pixel at the point center

I've been able to do this analysis in R using a loop that goes through each point, but I'm not familiar with Python, so I hoped I could do this out of the box

• Welcome to GIS SE! You say you can't picture this working with certain tools. What have you tried? – Barbarossa Mar 17 '16 at 19:36
• You're on the right track with a focal statistic. However, you should show due diligence before we can help you. Show us you what you have tried, or at least what research have you done along with specific questions about what is still unclear to you. – Barbarossa Mar 18 '16 at 15:28
• I've edited my question and hopefully clarifed. Will that 'bump' the question to the top or has it been buried? – user69494 Mar 18 '16 at 16:55
• You are asking for a weighted focal mean of the squared tree size, with the weights equal to the inverse squared distance. Assuming the cellsize of the raster is small enough to assign all trees to distinct cells, this is simple and straightforward to do, requiring no programming. – whuber Mar 18 '16 at 18:59