Detecting hundreds of invalid centrelines in a single network

Todays post is a look at a centreline data validation analysis project, which had an amazing and somewhat surprising outcome!

The Datastack was commissioned by a professional services consultancy for this project, with the dataset in question relating to a local authority transport network.

What was the issue?

The client was facing an issue with invalid network centrelines, where the direction of the centreline was running in the wrong direction. This was impacting their data collection process for another project they were working on.

So what does it mean when a centreline is running in the wrong direction? Lets unpack that idea a little bit first.

For the purposes of this post, we will refer to all network sections as ‘roads’ (being the underlying terminology used in RAMM). But depending on your RAMM setup, a road could also represent things like walkways, cycleways, car parks and anything else that is assigned a road_id in the database.

A RAMM road is made up of one or more carriageway sections. Each carriageway section has information that defines, describe and classifies the section(s) that make up a road. This includes where each carriageway section starts, and where it ends.

In this example above, this carriageway section starts at Road A, which is the beginning (0m point) of the road. Then 75m later, the carriageway section ends at Road B.

In RAMM, the increasing direction of a road (i.e. where it starts and running through to where it ends) is therefore defined by these carriageway sections.

Each road will typically also have a spatial centreline, which is effectively just a polyline representation of the road on the map. This centreline will also have an increasing direction based on where the polyline starts and stops.

In this example the running direction for the ‘My New Road’ centreline (the red line) is consistent with the carriageway direction (as showing in the previous screenshot above). The centreline starts at the intersection with Road A and runs through to the intersection with Road B.

It’s important to note the carriageway section direction and spatial centreline direction will not necessarily be the same for a given road. Generally you would expect they would always align, but it is possible they will not match.

When these two directions conflict with each other, it can lead to issues such as location and data visualisation problems.

The purpose of this project was to find instances such as this, where the centreline direction was going in the opposite direction to the carriageway direction.

What was the solution?

There were no standard tools available in the RAMM software to analyse and validate this specific issue. So a custom analysis workflow was created for the project, utilising ArcGIS Pro and SQL Server.

The ArcGIS Pro app was used to perform geoprocessing and analysis of the network and centreline data. This combined carriageway section attributes with the road centreline dataset. Proximity analysis was then performed to identify other network sections within a predefined buffer threshold of each network section end point.

A custom analysis process was then built in SQL Server to take the spatial results and perform some additional analysis of the data. This generated a list of candidate roads that were identified as likely running in the wrong direction.

A review of that candidate list was then undertaken within the context of the RAMM database, to remove any ‘false positives’.

The analysis process was designed to identify as many issues as practicable within the project timeframe. It is acknowledged that it potentially may not have captured all invalid centreline instances in the database, due to factors such as the analysis thresholds supplied, unstructured network data and gaps in the network dataset.

And the better than expected result?

The outcomes from the project were somewhat surprising, as it far exceeded any expectations for the number of invalid centrelines that would be identified during the process.

In total there were over 240 centrelines identified as running in the wrong direction, even after the evaluation and removal of the false positive items.

So there was definitely great value delivered through this project, enabling the client to be able to make a substantial improvement to the centreline dataset.