CODEWITHSUNDEEP
rgisgeospatialeuclidean-distance
Jade131621
Jade131621

Reputation: 389

Computationally efficient way of calculating euclidean distance between points and nearest line on a landscape in R sf

I have a large dataset of GPS locations (480,476) and a large shapefile of roads (lines; 561487) and I want to efficiently calculate the distance to the nearest road for each point.

If I tried using st_distance(points_sf, roads_sf) which calculates a distance matrix: the Euclidean distance between every point in points_sf (1:480476) and every line in roads_sf (1:561487) which was too big. Thus, I split it up into smaller chunks by ID which was more manageable. Even still it took a long time (> 1 day) and took up a lot of memory so that I couldn't run anything else on the computer or risked crashing it (even with 64 GB of RAM).

Is there a more computationally efficient way to do this sort of calculation in R?

Upvotes: 0

Views: 146

Answers (1)

Jade131621
Jade131621

Reputation: 389

What I have so far is to use st_nearest_feature() to identify the nearest line to each point and then calculate the distance between the point and its nearest line.

This was my solution:

    near_rd = st_nearest_feature(points_sf, roads_sf)
    dist_rd = sapply(1:length(near_rd), function(x)
                  st_distance(points_sf[x,], roads_ms[near_rd[x],]))

Upvotes: 0

Related Questions