Implicit radius in matrix v2?

openrouteservice matrix
1

Hello, new to ORS but getting along fine. I noticed that a matrix v2 request snaps to the nearest graph node, presumably using snap v2 internally. However, since I cannot set the snap radius, I guess matrix uses the default 350 m (I can figure this out by using snap v2 with different radius values).
Would it be possible to pass radius also as a post parameter to matrix v2?

Keep up the good stuff!

Guido

2

Hey,

yup, that is exactly what happens.
Currently, radius is not planned as a parameter to /matrix.
Introducing the /snap-endpoint was done among others to enable you to check the results of snapping and alter your input as desired.

Best regards

3

Thanks for this. As a follow up, I run into the following:

  • For a (large) set of (medium sized) administrative units, I try to get matrix response for the units’ centroids to a fixed (national) location. This is good enough for our current modeling set up;
  • The direct matrix call runs into the problem described above, i.e. for many admin units, no snapped node is found within the default snap radius applied inside matrix;
  • I now first do a snap call with a 10 kilometer radius. This leaves 8 (out of > 600) admin units without a snapped node, and, therefore, without a matrix response;
  • When I inspect OSM for the centroids of the missing 8 admin units, I find roads well within the 10 km radius. I am a bit puzzled, therefore, why these are not found as valid nodes.

(NB. I also used a 100 km radius, but the same 8 units centroids will not be snapped to anything).

May this be linked to (topological) issues in OSM?

G.

4

It might, and it might also be linked to this issue of differing snapping implementations.
I am also not sure whether there doesn’t exist a global (and probably undocumented) limit to snapping…

8 units without a snapped node sounds like something one could manually fix, by setting the corresponding centroids to something valid?