Linear Referencing

04-07-2011 05:48 PM
New Contributor III
I have an issue with a Linear Referencing process that I need to perform so I can do a Find Closest Facility later on.

I have a road network and a table identifying the start and end points (using distances in metres) along certain roads. The original road network was spaghetti digitised in a CAD and lacks topology.
The Linear Referencing process works fine on the original spaghetti road network and locates the sections of gravel roads. However to then do a Find Closest Facility to work out the distance of the gravel roads to the closest gravel pit, the connectivity in the road network needs to be fixed. Also the connectivity needs to be correct in the road network for the future.

To fix the connectivity I have opened an edit session and used the Planarize Lines tool in the Topology toolbar. This has worked well until now.

I have returned to redo the analysis but I have begun with the corrected (planarized) road network and when I go through the Linear Referencing process I now get a slightly different and fragmented result and I can't figure out why. What happens is that the analysis doesn't recognise the start point and then splits the route event in to multiple pieces and distributes them along the road. The sum of the pieces adds up to the correct total, but the distribution is all over the place, i.e. random!!!

My thought is that maybe I should have used a different tool to rebuild the connectivity in the road network.

Although, at the begining of the Linear Referencing process, i.e. Find Route, I get the same result regardless if I use the original spaghetti roads or the connectivity corrected roads. The issue then becomes apparent in the result of the Make Route Event Layer. I enter all of the same information the same way and I get a different result.

This is the problem as I then need to use the Make Route Event Layer to determine the start locations of the gravel roads to then run the Find Closest Facility analysis.

I keep looking at this an then having to come back and edit the statement. what I have just notices is that when I corrected the connectivity of the original road network, it of course created multiple segments out of a road that was a single feature (line) at intersections. What "Make Route Event Layer" has done is split the gravel road section (length - from start and end measurement fields in a table) into a number of line segments commensurate with the number of segments in the road, and distributed them proportionally, but at random, one to each line segment making up the road.

Does anyone have any thoughts?
Tags (2)
0 Kudos
2 Replies
Occasional Contributor
I have worked with linear referencing problems before, but am just getting into network analyst. One simple thing you could do to start troubleshooting the problem is select a really small number of roads (maybe 5) from your original data and then run them through your process.  The whole time keeping an eye on topology, node m-values, lengths, etc.

It sounds like you want to:
1. Planarize your original roads
2. Make a Gravel Road Event Layer (based on an external table you have?)
3. Buffer the Gravel Road Event layer and use the buffered layers as a polygon cost barrier for the Network analysis

Linear Referencing and Network Analyst are related, and it is confusing because both freely use the term 'Route'.  But 'Route' in Linear Referencing parlance is more for 'Route IDs', Node M-Values, and transforming event data between them.  Where as 'Route' in Network Analyst is more of a least cost and direction based analysis.

From what you wrote it seems that you have gravel road data that will be used as a route event layer (Linear Referencing) and that route event layers will be used as an input for your Find Route Analysis (Network Analyst).
0 Kudos
New Contributor III
Yeah, I have pretty much done all of that. the Planarize Lines has done something to the dataset which makes it divide the linear features I'm trying to extract and then distribute them along a line.

For example, I have one road with f intersecting roads, making up six segments in my first road. I have m values to show sections of gravel road along the first road, i.e. only 1 gravel section on the road. When I Make Route Event Layer, the 1 gravel section is divided and distributed proportionally among the six road segments, as six separate segments.

ESRI has it at the moment so we'll have to see aht happens.
0 Kudos