bgfield

Newbie question about hot spot analysis

Discussion created by bgfield on Dec 12, 2013
My GIS experience has been mostly with digitization and a bit of web development, but a coworker asked for some assistance in an analysis project he's working on so I offered to help.  I think I'm on the right track but I wanted to run this by people with more experience who might offer some insight.

I've attached a picture of the study area we're looking at.  We're trying to determine a preferred habitat type for an invasive fish in our waters.  Biologists made multiple surveys on a boat shocking the waters to count the number of fish found in a particular tributary.  The red line in the picture represents the GPS trackline of the boat during one survey, and each point represents one of the invasive fish.  I'm using the optimized hotspot analysis tool in 10.2 to determine if there are any locations along the stream in which fish seem to be clustered.

I used two different methods so far.  First, I tried using the Count Incidents Within Aggregation Polygons method, using a 10x20m grid I created based on the GPS trackline.  My concern with that was the fact that some of the incidences would fall outside of this grid - am I correct in assuming that any incidences outside of the polygons wouldn't factor into the analysis?

In my second attempt I used the Count Incidents Within Fishnet Polygons (with no bounding polygon).  Both results returned a few hot spots more-or-less in the same location, but I'm thinking this method is ideal as every incident fell within one of the polygons.

Am I on the right track?  Is there a better tool, or more ideal preferences I should be using?  Are there any concerns I should be aware of when using this tool in stream/aquatic areas as the spatial distribution is far more confined than over a terrestrial study area?  Lastly, I've looked through some literature and videos about this tool, but if anyone has any recommendations of something I could read/watch to understand spatial statistics and clustering better, it would be greatly appreciated.

Attachments

Outcomes