Find Cotravelers shows tracks that cross but which are not true co-travelers

DonBarker · ‎01-08-2021

I'm testing the Find Cotravelers tool with AIS maritime tracking data from MarineCadastre.gov - 500K events off the coast of Miami.

The tool lets us set a time window and distance/proximity parameters.

What's missing is a parameter for the minimum duration during which the distance/proximity conditions must be met. For example, "They must be moving close together in time and space for a minimum 60 minutes."

Without this duration parameter, the tool also shows vessels that merely crossed paths within 100 meters and 10 minutes of each other [the tool defaults]. This can generate hundreds of false hits.

Here is the results layer - supposedly co-travelers:

This is the real picture:

[I've discussed this with @JamesJones4 but wanted to document it here for others who run into this.]

JamesJones4 · ‎01-19-2021

Hi Don - Thanks for the post. I've been trying a few different methods to get this done. And by and large it is very much doable.

Follow up question, in the image that you have attached above, we have two distinct sets of tracks that could be identified as cotraveling. The one on the left (red/yellow) sort of crosses each other, while the one on the right (yellow/yellow) appears to be diverging. Would you think that it is useful to include the how those bearings are related to one another? Because if something is converging/diverging may be of relevance as well.

Thanks for the feedback!

DonBarker · ‎01-29-2021

James,

Thanks for continuing to work with me on this. You raise an interesting question.

I suppose I would think of it differently. Any time two tracks meet the input criteria (space distance, time distance, and the new "mininum time for converged tracks" parameter *), it's the converged tracks alone that interest us. Because every set of cotraveling (converged) tracks will have a segment before when they are converging, and a segment afterward when they are diverging. Both converging and diverging will always be present before and after the valid "cotraveling (converged) segment. It seems to me that they don't need to be characterized.

Another way to put this might be: In the condition where the tracks meet the desired criteria, they are neither converging or diverging.

I hope this helps.

( * I see this has been added to the module you sent me to test. I'm still working on that. See my other post reply about problems updating Pro for this task. Thanks.)