Generate Service Area tool skipping polygon generation when run with Arcpy command

Question

I created a tool that uses service areas to group points that are within a certain distance from one another along a network (in my case, a river system). Overall, the tool works. However, when I run it, the Generate Service Areas geoprocessing step does not produce polygons around all input point 'facilities' when run with the Python code.

I tried running the same datasets manually with the Network Analyst toolbar and window and all polygons were successfully created. I also tried running the same dataset through the Generate Service Areas geoprocessing tool and it also generated all polygons as expected.

However, when I run the Python code snippet from the Python window or from within my custom tool, it randomly skips polygon creation for several of the point 'facilities'. This is my code snippet for the generate service areas step:

species_pts = 'species_pts'
sep_distance = 500
scratchGeodatabase = r'C:\Users\mmoore\Documents\ArcGIS\Default.gdb'
arcpy.GenerateServiceAreas_na(species_pts,sep_dist,"Meters",network,os.path.join(scratchGeodatabase,"service_area"),Polygons_for_Multiple_Facilities="MERGE",Detailed_Polygons="DETAILED_POLYS")

Anyone have an idea as to why this may be somewhat randomly skipping polygon creation around several points?

Answer 1

This sounds to me as if some of the points to get properly located when you run the GP tool. There is a nasty (but often times handy) option "Exclude restricted portions of the network" which is something you define when creating a Service Area NA layer. It can make it possible to locate the source points on the next allowed edge, but when you run the tool from Python, it may locate it on a restricted edge (say, oneway=N) thus making a point non-routable.

Try using the GP tool Make Service Area Layer, then Add Locations, and then finally Solve. If you are able to create service area polygons around all of your points by doing this, then you could copy Python snippets and run the code from Python window or as a stand-alone script (with minor modifications).

Just to be clear, you do not use the Network Analyst toolbar at all when testing this.

If it does work when you run the GP tools, but when running the Python code some of the points do not get polygons created, I'd start looking into the cut-off and default break value settings. There have been a bug in Network Analyst when a different cut-off and break values are used when you specify not a single cut-off value but rather when each point has its own one (read from an attribute).

I had to work around this by setting a ridiculously large value for default_break_values to be able to use different break values for each point.