Finding loose ends of highways

I would like to present my newest addition to

As you might have already noticed the imagery coverage of Thailand and surrounding countries was recently extended.
So highways previously ending at an imagery boundary can now be extended.

This tools is helping to detect places where major highways end in the middle of nowhere.

The map shows a yellow circle at the node which is not connected. The clever part here is that only places with available imagery are reported.
Using the remote control button it’s possible to load the currently visible selection in JOSM or iD/potlatch.

The map shows areas with high resolution imagery available in green, loose ends are marked with yellow dots.

Opening the page in the editor allows to complete the tracing from the aerial. Please double-check that the aerial imagery is correctly aligned to GPS tracks before tracing features.

Find it here:

The data will be updated every four ours. As I treat it currently in beta-test I’ll do the updates manually. If you notice any problems or false positives please let me know. I’ll see if it’s possible to work around it. I have tested it using recent versions of Chrome and Firefox on Windows.


I have now updated live updates. Data is refreshed every four hours.

Currently **1469 **loose ends are reported.

Known problems:

  • Permalink of Bing layer is 1 level too low: Is a known problem of OpenLayers

  • Bing boundaries vanish on zoom level 11 and higher: I suppose this is another OpenLayers problem. I see similar effects on the OpenLayers demo page. Sometimes moving the map a greater distance made it appear again.

  • Chrome shows white background around radio for base layer selection: It is some issue with Chrome and z-order css. I think OpenLayers could work around the problem. As it’s minor I personally won’t be actively looking for a solution.


Nice tool.
In Phangnga province, I could connect two roads to #4240 - they were created with Potlach, and ended just before hitting the major highway…

Those “nearly connected” ways had not been the main reason for that tool. They would have been found by others like OSM Inspector as well.
But you’re right that they also need to be fixed. It’s things like that preventing the data being usable for routing.