I have created several service areas around a facility using different distance limits. From 100m, 200m, 300m....1000m.
I have discovered that some of the shapes have spiky, pointed outliers (looks like a strange lop-sided star) and other roads are ignored creating virtually zero distance in another direction. I am using distance and not time so I don't see why the analysis isn't treating all roads the same. I have minimal parameters because I am looking at walking distances.
It might do an outlier at 100m and then stop for 200m and redo for 300m. Thus, when I count the number of residences within each service area polygon, it is 1, 0, 1, 2 .... When I look at the 200m service area there seems to be a 'hole' in the service area. This is why the house does not have a facility counted within 200m when it was there at 100m.
A technician has infilled the 'holes' with the count result from the previous distance category, so it becomes 1, 1, 1, 2.
Is this a reasonable thing to do? Do I need to redo my service areas? If so, what should I change?
The service area is generated from the lines that are traversed within the distance you specify. So you could turn on the lines and see what portions are getting traversed. Their can be holes in the service area polygons for un-reached areas. Note that you can possibly get more holes when using the detailed method and less when using generalized method. So if you are using detailed, switch to generalized. As far as the spikes, the outer-most ring of the service area polygons is also influnced by the things that are outside the maximum break values. So in areas of dense networks, you will get less spikes but more when there are fewer edges on one side. You can control this by turning on trim and using an appropriate trim distance. The default is 100 meters. If your break values are 100 meters, then make the trim is around 25 meters. Jay Sandhu