Which street data should be utilized to create a network dataset alongside public transit data?

[TackGuanLau]

I have two sets of network datasets: (1) a road network and (2) a walking and cycling network.

To create a GTFS transit network using the Street, should I combine these two datasets, or is using just (2) sufficient? I assume we only need the walking paths to connect to the transit stops. Once connected, travelers will board a LineVariantElements, calculate travel time, exit at the transit's end, rejoin the walking path, and proceed to their destination.

After generating the Public Transit Network dataset with GTFS, during the OD Matrix analysis, does the Total Length in the output table for Lines get calculated based on the LineVariantElements (straight line) or the Streets layer that was utilized to construct the Public Transit Network?

Much appreciated, thank you.

[MelindaMorang · Accepted solution]

If your walking and cycling paths are sufficiently comprehensive and well connected, then you can use only this in your network.  The traveler needs to be able to walk to and from transit stops and their origin/destination as well as walk between stops to make transfers.  So if your walking network can do that, it's all you need.

Problems could occur if your walking network is missing connections between places (maybe where someone has to walk along a street because there's no specific pedestrian infrastructure included) or where it's poorly connected (like if you have sidewalks looping all the way around a block but don't have a crosswalk modeled to connect from one block to the next block).

When the analysis is solved, the distance reported is fairly meaningless.  The LineVariantElements feature class is generated with arbitrary straight-line connections between connected stops, and the Shape_Length values of these features is not physically meaningful.  The reported travel time is the only meaningful value.