# Converting probability buffers to a probability SURFACE

I have a point feature, which I have buffered using the estimated location error of each point. Each point has a different sized buffer, representing 95% confidence that the true location is withing the bounds of the buffered area.

I would like to generate a series of individual rasters (one for each of the 208 points) where the value of each cell equals 0.95/(the number of raster cells representing the buffered area). Cells representing smaller buffers have higher values.

Essentially, I think I would like to convert a multipart feature (the buffer polygons) to a SERIES of single part features (rasters representing the buffers).

Then I would like to sum the series of rasters, so that the values of each cell in the output raster represents the degree to which the buffers overlap. The more buffers that overlap, the higher the cell value will be.

Ultimately, I am trying to produce a relative probability surface, where areas where more smaller buffers overlap will have higher values than areas where fewer, or larger buffers overlap.

Any idea how I might do this?

• You'll probably want to script (or ModelBuilder) the process. Make each buffer into a unique shapefile, so you have 208 shapefiles (one per point-buffer), then convert each of those into a raster (calculating the value based on the buffer area), then sum the 208 rasters together. It's POSSIBLE to do manually but would be tedious :) Sep 29, 2014 at 18:22
• Thanks Erica. How would you suggest I convert a multipart shapefile to a series of single part shapefiles? Sep 29, 2014 at 18:48
• I know I've seen that asked on here before and I'm looking for a link to a good Answer about it... EDIT: Here are a couple options, in ModelBuilder and in Python. Sep 29, 2014 at 19:11
• I used the USGS tool provided in the link below the model builder comment that split the multipart feature to 208 single part features (all in one gdb). However, now I am not sure how to get the model builder to automatically roll through all those polygons to convert them to rasters. Can i select the gdb containing all the polygons somehow? It sems I can only select input features that have been loaded onto my map document... Sep 29, 2014 at 20:50
• I am going to transition this to an answer and provide some more detail :) Sep 29, 2014 at 21:32

I suggest automating this process with either ModelBuilder or a Python script. The general procedure is as follows:

1. Make each of your buffers into a unique feature. (As your comment mentioned, you've done that.)

2. Convert each buffer into a raster, calculating the cell value based on the area of the buffer.

The key thing here is to iterate through each shapefile; for this, you'd need the Iterate Feature Classes tool within ModelBuilder.

Before converting, calculate what the cell value will be...

• Create a new field (e.g. `Value`)

• Use Calculate Field tool: `0.95 / (!shape.area@SQUAREMETERS! / CellSize)` (where cell size is the area of one cell)

• You can use whatever area units you need in that; refer to the help page for a list.
• Use Polygon to Raster tool, selecting your just-created field as the raster value. (Feature to Raster will also work.)

• The iterator tools produce a variable (e.g. `%Value%` or `%Name%`, you can call it whatever you want) that can be used in this step to ensure each output raster is unique (e.g. `C:\temp\out.gdb\rast%Name%` is made from the shapefile `%Name%`). The Examples of using iterators in ModelBuilder page explains that concept.

3. Sum the rasters together. Use the Collect Values tool (see example page for useful images) and feed its output into the Cell Statistics tool (you want the statistic "SUM").

Note: I think that you will need to set the processing extent to encompass all the buffers, or else it will make each rasters only as big as each buffer and then when you try to sum them it won't work quite right.

• How do I ask that each output raster has a unique name? For example, I have "test_timestamp" as my output raster, but it just keeps reproducing a raster named "test_timestamp" rather than creating a new raster each time. Sep 30, 2014 at 15:17
• The iterator tools produce a variable (e.g. `%Value%` or `%Name%`, you can call it whatever you want) that can be used in subsequent steps. The Examples of using iterators in ModelBuilder page explains it out a little more clearly with diagrams and examples :) Sep 30, 2014 at 15:20
• Howcome I can't change the name of the output from the iterate feature class tool? It automatically named it the first in my list of polygons....which then carries through the rest of my model. Sep 30, 2014 at 15:27
• That I am not sure without seeing a screenshot -- you may want to ask a new Question. Sep 30, 2014 at 15:29
• For the iterate feature class tool, I have the .gdb containing all my polygons as the input, but the green output bubble displays the name of my first polygon file. So each time I run the model, it seems to regenerate a raster based on that polygon over and over again, rather than generating rasters for each of the subsequent polygons. I hope that makes more sense. I'm not sure what I have done wrong. Sep 30, 2014 at 15:41