Talk:Tag:surface=laterite

Routing consideration

for at least Valhalla, which can respect time windows, it might make sense to publish a conditional restriction like "motor_vehicle:conditional = no @ (Dec-Jun)". right now, I'm quite uncomfortable even opening it for routing, if the majority of the year the road is not accessible with a non-4x4 car. what do you think? Nilsnolde (talk) 07:47, 4 July 2026 (UTC)

uh, pls don't read this as "tag for valhalla"! IMO it makes sense to add this for everything else as well. Nilsnolde (talk) 07:52, 4 July 2026 (UTC)

last thing: I realize those "months" are not "stable" over the years or well predictable probably. but a good guess is better than none at all. people can actually if we route them the wrong way, see https://edition.cnn.com/2012/12/10/tech/apple-maps-australia-flaw. that was a huge news at the time, I remember. Nilsnolde (talk) 07:58, 4 July 2026 (UTC)

Good point on the CNN example. On the month range: monsoon months do reliably mean rain, so Dec-Jun is a reasonable coarse filter, not a bad guess. The nuance is inside that window: a given day's passability still depends on how recently it rained and how much traffic has churned the ruts since (near-frictionless within minutes of rain, then plastic for days after), so "no" for the whole window is pessimistic on drier stretches within it, and "yes" outside it can miss an occasional out-of-season storm. If Valhalla (or others) can layer recent-precipitation data on top of a motor_vehicle:conditional=* range, that would sharpen it further.

julcnx (talk) 11:35, 4 July 2026 (UTC)

thanks @Julcnx for your thoughts! Hehe, Dec-Jun was really just an example, same as "no" restriction for that time range. the exact range is not only monsoon is it? it's south american/african countries etc which get hit by other phenomena in other months I guess. I'm not so familiar with the outside-of-routing OSM tags, but I'd imagine there's smth signaling seasonal uncertainty with a month range? gotta be the same for some "winter" roads in norther canada/russia, which can only be driven when it's frozen. the end game is what you're suggesting, integrating weather forecast into our costing.

for now, we'll add a warning in the routing response if any such way is in the final path. maybe that's even enough for us. I just don't want people (esp tourists/foreigners) to travel super risky with our output. Nilsnolde (talk) 16:21, 5 July 2026 (UTC)

one last thought: maybe it's worth it to add a table to the wiki with the most common regions and their most likely rainy season? on second/third thought, maybe we can spare OSM the bloat for those added tags I proposed, they're far from exact or accountable anyways. however, if there's such an approx table in the wiki, people could look it up themselves. we (routers) just have to return those things in the response, so people can be aware. and the routing apps must include the wiki in their UI/UX as link for those type of surfaces.

what do you think? Nilsnolde (talk) 09:03, 6 July 2026 (UTC)

Rethinking the season/table angle: expecting mappers to reliably add and maintain a seasonal conditional tag on top of `surface=laterite`, across thousands of ways, in regions that are already undermapped and where plenty of mappers don't even bother with `surface=*` and just leave `highway=track`, isn't honestly realistic. Even someone who can identify laterite in the field might not know the wet-season window precisely, since it varies with local terrain and whether the road sits under jungle canopy or full sun, a wiki table would only ever be a rough approximation. The warning-in-response idea you mentioned is the right shape though. What would actually help more: a routing-profile-level avoidance option for soft/wet-prone surfaces (`mud`/`clay`/`laterite`), similar to how some profiles already expose avoid-unpaved toggles, on by default for something like a city-car profile and off by default for mountain-bike/4x4 profiles. That puts the judgment call where it belongs, on the router/user, without asking mappers to maintain data they can't reliably keep accurate. julcnx (talk) 13:08, 6 July 2026 (UTC)

The Wikipedia articles on wet season and Earth rainfall climatology use the following detailed animation showing long-term average precipitation for each month of the year. I believe implementing support for this information is not a trivial task, but it would be fantastic if it were possible. Ideally, the user would have the option to specify a particular time for the route.

--Fernando Trebien (talk) 14:14, 6 July 2026 (UTC)