Have spent some time on generalizing some datasets and got puzzled by the algorithm used by the Integrate GP tool.
I have two points in a file geodatabase of 10.3 version, those points are located on a certain distance from each other. I run the Integrate tool (tried both inside and outside the editing session). The feature class is stored in a projected coordinate system (specifically, the WGS_1984_Web_Mercator_Auxiliary_Sphere). I run the Integrate tool on this feature class with the cluster tolerance = 1000m.
I cannot comprehend how the cluster tolerance is used. I've added a scale bar while in Layout mode for illustration purposes. Here points will be integrated when the distance between them = ~2300m.
They stop collapsing into one feature in between when the distance gets larger than ~2800m. Any value smaller than that will move both points into a position that is right between them (i.e., they get integrated).
So, I could see two ways the cluster tolerance is used:
- if the distance between the two points is smaller than the cluster tolerance, they will be moved into a common central position; doesn't seem to work for me - they get integrated when distance between them > 1000m.
- if the circles around the points (with radius=cluster tolerance) intersect, they will be moved into a common central position (then they should be integrated if the distance between points < 2000m, which again is not the case.
Can anyone explain how does the Integrate tool work in terms of cluster tolerance, because from my tests it doesn't make sense.
I am on ArcGIS 10.3, but I've tested on 10.2 and got the same behaviour.