routing with r5r

[temospena]

From our recent findings, a disagregation level of 500 trips might be enough (do not need to go for 200 or 100?) Also, the routing results with r5r, using Minetti algorythm for slopes (use Copernicus with 25m spatial resolution, projected to WGS84), and LTS = 2 proves to bring better results when applying to Lisbon municipality.

Nevertheless, the LTS2 and LTS3 does not route in bicycle lanes (not bicycle tracks) when they are in a structural avenue (e.g. Almirante Reis). Some tweaks to code should be made to be more realistic. This is how LTS is being used: https://docs.conveyal.com/learn-more/traffic-stress and https://github.com/conveyal/analysis-backend/blob/v5.9.0/src/main/java/com/conveyal/r5/labeling/LevelOfTrafficStressLabeler.java

Using 2 gtfs files in the r5r folder, with both fluvial operators may also bring up the connections that we want, in that case maybe set a Infinite waiting time when routing.

Next step:

• setup r5r to use both gtfs, and use elevation = "MINETTI".
• re-run for 500 disagregation threshold (although should we wait to have different jittering weights? #24 )
• "use the dev version of r5r allows one to extract segment-level estimates of LTS using the street_network_to_sf{r5r} function" before using rnet() ?

[temospena]

computed with od disagregation of 100 pairs, r5r with Minetti elevation, plust method to get osm tags at segment level for each route.