I wish to plot Kernel Density Map for geo-referenced rural road crashes in ArcMap 10 (using Spatial Analyst Tools->Density->Kernel Density-Dialogue box pops up to enter data). I found that the "Output Cell size" and "Search Radius" by default the ArcGIS software is taking for plotting Kernel density Map for road crashes is very small say 0.001 metre(m) and 0.01 m respectively, which is small to conclude inferences from the plotted kernel density maps. When I change"output cell size" and "radius" in the Dialogue box to larger values say 5 m and 50 m respectively, ArcGIS does NOT plot the kernel density map and show error. Could anyone help me to diagnose the problem in plotting kernel density maps for road crashes. Secondly I also request the esteemed members to write steps or procedure to carry out Network Density Estimation using ArcGIS as Kernel Density estimation is planar based and not popular in analysing highway crashes.
Thanking you in advance.
Reply to the cited query may be posted at my email account: firstname.lastname@example.org
H. M. Naqvi
What coordinate system is your data in? Sounds like there is a mismatch between the coordinate system (geographic degrees perhaps) and you distance. 50 in degrees is a long way.
Probably need to get the data into a projected coordinate system first.
Thanks for prompt reply. My crash data (latitude and longitude) are in degree and decimal. I used GCS->World->WGS43.Similarly for projected coordinate system is also taken as Northern hemisphere->WGS43. While making kernel density one is required to give search radius and Output cell size, these two items are to be given in metres or kilometres etc. May I request you to send your valuable inputs on the same.
From the Help: default cell size is the shorter of the width or height of the output extent in the output spatial reference, divided by 250.
If ArcMap suggest the size 0.001, then the height or width of you data extent (smallest of the two) will be 0.25, which most certainly is expressed in decimal degrees.
...and it also states "Units are based on the linear unit of the projection of the output spatial reference". For these types of tools it is always best to work with a projected coordinate system as suggested by Neil Ayres and Dan Patterson .
Thanks for furnishing reply on cell size of kernel density map. My next question is as follows:
I have to analyse traffic accident data in order to investigate the density of crashes(point events). A standard (planar) Kernel Density Estimation (KDE) produce a smooth density surface of spatial point events over a 2-D geographic space. However, the planar KDE may not be suited for analysis road crashes (events), which usually occur inside a linear space.
Does anyone have suggestion how to plot network kernel density estimation for traffic accident data using ArcGIS tools?
Thanks in advance.
Kernel Density is an interpolation that does not have the option to include barriers. So in case of two parallel roads or two highways that cross at different Z altitudes, points located on one road will contribute to the density of the other. In this case it might be better to split up your network in sections and crossings and aggregate the number of accidents per part.