Kernel Interpolation with Barriers vs Kernel Density tools

Question

Hello, I am a student working on a research project focused on estuarine dolphins. We use start and end point locational data collected while we are conducting boat surveys along a 40 km stretch of a river, starting at the mouth of the river. My goal is to get Kernel Density Estimates that include barriers in my shapefile.

I have noticed that the Kernel Density tool has an Input Barrier Features parameter. Can I use this tool instead of the Kernel Interpolation with Barriers tool, or is there a reason that I should be using the Interpolation tool? From what I have been able to gather, it seems like this is a newer feature in the Kernel Density tool.

Thank you in advance for any advice. In case it is relevant, I'd also appreciate any help in how people decide what bandwidth and cell size parameters to utilize. I was considering the hopt calculation (2/3n)^1/4*standard distance and a cell size of 150 m x 150 m but the KDEs seem large when compared to the data points.

Thanks again.

Andrea

Robert LeClair · Answer

Andrea - I do not consider myself an expert by any means, but can attempt to answer the question as best as I can.

1. The Kernel Density GP tool with the Input Barrier Features parameter was implemented at ArcGIS Pro 2.6 (2020) and uses a Quartic kernel only.  You can learn more about it here.  The summary of the tool says - "Calculates a magnitude-per-unit area from point or polyline features using a kernel function to fit a smoothly tapered surface to each point or polyline. A barrier can be used to alter the influence of a feature while calculating kernel density."

2.  The Kernel Interpolation with Barriers GP tool is a more complex GP tool and uses the following radially symmetric kernels: Exponential, Gaussian, Quartic, Epanechnikov, Polynomial of Order 5, and Constant.  From the summary of the tool - "A moving window predictor that uses the shortest distance between points so that points on either side of the line barriers are connected."

I imagine the Spatial Statistics Community will likely know more details about the differences.