I am creating a network dataset using the roads polylines contained in the OS Meridian 2 data set, in order to eventually calculate road journey distances. I have been considering connectivity and elevations, and have been using the sDNA manual - network preparation module as guidance. This suggests that the 'link-node rule' applies to the Meridian 2 data set whereby "Intersections are not allowed in the data at all; ALL lines where they touch MUST have coincident endpoints. Where links don’t join at their endpoints this is shown by providing elevation or grade separation data – endpoints are assumed not to join unless their elevation and grade separation match. This data is provided in a separate layer of points, cross-referenced to the lines whose elevations they represent (OS Meridian does this)." - I have the separate layer of points (link nodes) which contains a 'levels' field. For example, if a road over a bridge can be accessed then a level value of 0 is used; a value of 1 is used when there is no access from one road to the other at an intersection.

It is then recommended to convert from this format to a network with a 'coincident endpoint connectivity rule'. However, the guidance seems to stop there. In order to do this for 'link-node rule' data, the manual simply suggests that I get in touch with the sDNA via this website, as there are tools available for doing this conversion.

Can anyone offer advice on this issue? I'm not sure what I can do next.

For reference, here is the full article provided by sDNA: http://www.cardiff.ac.uk/sdna/wp-content/downloads/documentation/manual/sDNA_manual_v3_3_alpha0/network_preparation.html


I wrote that, and I did indeed write a script to transfer meridian node information onto the links, though it's a tool for internal use so I didn't publish it. The easiest answer for you may be don't use OS Meridian: it has been superseded by OS OpenRoads which is also free to use and I think comes with elevation data attached to the links rather than nodes.

Somebody else here may know of an existing tool to convert network data, but if not, and you decide Meridian is essential to you, do get in touch via the website.

  • Thanks very much for your suggestions. I would have considered using other data but my study looks at changes in public access between 3 time points (1995, 2000 and 2011) and Meridian seems to be the only suitable roads data set going back to the nineties, according to OS. Any further advice you could give would be much appreciated. – Jen Jun 13 '16 at 11:16
  • Hi Jen, please drop me a line at sdna@cardiff.ac.uk – Sideshow Bob Jun 13 '16 at 11:24

Meridian is not very accurate to represent street network and it has a lot of problems. I suggest to inquire OS or Digimap to get what was then called land line it goes as far back than 1998. At that time the layer with the road center line was called OSCAR.

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.