Kernel Density

Open Table of Contents ToolBox Implementation
Uses Kernel Density Estimation to interpolate a surface from the input Point or Polyline features. The function fits a symmetrical surface over each  input point using Gaussian kernel.
The surface has maximum value in the input point, decreases as the distance from the input point increases and has 0 values at distance equal to the search radius.
The volume of the created surface is equal to the value of the input point.
The density of each cell of the output raster is calculated by adding the values of all individual surfaces for that cell.

The search radius does not influence the volume of the surface, but has a major impact on the generalization of the data. The larger the search tolerance, the smoother and more generalized surface will be interpolated. As you can see from the image below the selection of the search radius is very important and influences greatly the surface. You should evaluate your data and have always in mind what is the goal of the task when deciding on what search tolerance to use.
 

Inputs:

  • A Point or Polyline feature layer or feature class.
  • Output raster name and format
  • Cell Size of the output raster
  • Search radius.
  • Value field -  a field from the attribute table to be used as value for each point. The value might be the population at this point, the number of incidents at the location etc. If you do not have such a field in the attribute table, just create a new field and calculate the values of all records equal to 1.
  • Area Units - The value of each cell of the output raster will actually have value measured in Value per square unit. The default unit is the unit of the spatial reference of the input dataset, but you might want to change this to a different area unit (for example incidents per square kilometer)

Output:

  • A floating point raster.

Example - the surface created from the 6 points in the profile above:

Search Radius = 2000 meters - smoother more generalized appearance
Search Radius = 1000 meters - more pronounced influence of the individual points
Both surfaces superimposed - two very different surfaces. The volume of both however is the same and is equal to the sum of the values of the input points.

Notes:

  • Initially the name of the output raster defines the raster format
    • no extension specified - ESRI binary GRID
    • .img extension (for example raster1.img) - ERDAS IMAGINE image.
    • .tif extension (for example raster1.tif - Tagged Image File Format (TIFF) image.
  • The initial output raster format can be changed by selecting the desired output in the dialog.
  • Currently only file based rasters are supported. Rasters cannot be stored in a GeoDatabase. After you get the desired result, you can export the raster to a GeoDatabase using the standard ArcGIS tools.
  • The input feature class must be in a projected coordinate system

ToolBox implementation

Command line syntax

ETS_GPDensity <Input Dataset>  <Out Raster> <Elevation Field> < Cell Size> <Interpolate Radius>   <Area Units>

Parameters

Expression Explanation
<Input Dataset> A Point, Polyline or Polygon layer or feature class
<Out Raster> A String - the full name of the output raster (A raster with the same full name should not exist). The output raster type depends on the extension of the output file(see Notes above)
<Value  Field> A String representing the name of the field which values are going to be used for interpolation.
<Cell Size> A Double representing the cell size of the output raster.
<Interpolate Radius> A Number - see main description above
<Area Units> A String - see description above

Scripting syntax

ETS_GPDensity (Input Dataset, Out Raster, Value Field, Cell Size, Interpolate Radius, Area Units)

See the explanations above:
<> - required parameter
{} - optional parameter

All ESRI products mentioned are trademarks of Environmental Systems Research Institute, Inc.
Copyright: Ianko Tchoukanski