Add new node to SpatialLinesNetwork in stplanr

Question

How can one add a new node to a SpatialLinesNetwork?

context of my problem: I have a shapefile of a bus route and another shapefile of bus stops. I want to calculate the distance between stops along the bus route. Ideally, each stop would be a node and I would use stplanr::sum_network_routes() to calculate the distance between them. The problem is that when I convert the bus route into a SpatialLinesNetwork the network only has a few nodes that are far from each other and unrelated to bus stops locations.

reproducible dataset:

# load library and data
  library(stplanr)
  library(sf)

# get road data
  data(routes_fast)
  rnet <- overline(routes_fast, attrib = "length")

# convert to sf obj
  rnet <- st_as_sf(rnet)

# convert SpatialLinesDataFrame into  SpatialLinesNetwork
  sln <- SpatialLinesNetwork(rnet)

# identify nodes
 sln_nodes = sln2points(sln)

# Here is a bus stop which should be added as a node
 new_point_coordinates = c(-1.535, 53.809)
 p = sf::st_sf(geometry = sf::st_sfc(sf::st_point(new_point_coordinates)), crs = st_crs(rnet))


# plot
plot(sln, col = "gray")                 # network
plot(sln_nodes, col="red", add = TRUE)  # nodes
plot(p, add=T, col="blue")      # stop to be added as a new node

Answer 1

Thanks for the question, thanks to this question and subsequent collaboration with Andrea Gilardi, I'm happy to announce that it is now possible to add new nodes to sfNetwork objects with a new function, sln_add_node().

See below and please try to test reproducible code that demonstrates how it works:

devtools::install_github("ropensci/stplanr")
#> Skipping install of 'stplanr' from a github remote, the SHA1 (33158a5b) has not changed since last install.
#>   Use `force = TRUE` to force installation
library(stplanr)
#> Registered S3 method overwritten by 'R.oo':
#>   method        from       
#>   throw.default R.methodsS3
#> Warning in fun(libname, pkgname): rgeos: versions of GEOS runtime 3.7.1-CAPI-1.11.1
#> and GEOS at installation 3.7.0-CAPI-1.11.0differ

sample_routes <- routes_fast_sf[2:6, NULL]
sample_routes$value <- rep(1:3, length.out = 5)
rnet <- overline2(sample_routes, attrib = "value")
#> 2019-09-26 16:06:18 constructing segments
#> 2019-09-26 16:06:18 building geometry
#> 2019-09-26 16:06:18 simplifying geometry
#> 2019-09-26 16:06:18 aggregating flows
#> 2019-09-26 16:06:18 rejoining segments into linestrings
plot(sample_routes["value"], lwd = sample_routes$value, main = "Routes")

plot(rnet["value"], lwd = rnet$value, main = "Route network")

sln <- SpatialLinesNetwork(rnet)
#> Linking to GEOS 3.7.1, GDAL 2.4.0, PROJ 5.2.0

new_point_coordinates <- c(-1.540, 53.826)
crs <- sf::st_crs(rnet)
p <- sf::st_sf(geometry = sf::st_sfc(sf::st_point(new_point_coordinates)), crs = crs)
p_dest <- sln2points(sln)[9, ]

# We can identify the nearest point on the network at this point
# and use that to split the associated linestring:

sln_new <- sln_add_node(sln = sln, p = p)
#> although coordinates are longitude/latitude, st_nearest_feature assumes that they are planar
route_new <- route_local(sln = sln_new, from = p, to = p_dest)
plot(sln_new)
plot(p, add = TRUE)
plot(route_new, lwd = 5, add = TRUE)
#> Warning in plot.sf(route_new, lwd = 5, add = TRUE): ignoring all but the
#> first attribute

^{Created on 2019-09-26 by the reprex package (v0.3.0)}

In case it's of use/interest, see the source code of the new small family of functions that support this new functionality here: https://github.com/ropensci/stplanr/blob/master/R/node-funs.R

Answer 2

This doesn't answer your question at the outset, but I believe it does resolve your "Context" by showing how your desired network distances can be calculated. This can be done with dodgr (latest dev version) like this:

library (dodgr)
library (stplanr)
library (sf)
library (sp)
dat <- st_as_sf (routes_fast)
net <- weight_streetnet (dat, wt_profile = 1)

The net object is a simple data.frame containing all edges and vertices of the network. Then adapt your code above to get the routing points as a simple matrix

rnet rnet <- overline(routes_fast, attrib = "length")
SLN <- SpatialLinesNetwork(rnet)
sln_nodes = sln2points(SLN)
xy <- coordinates (sln_nodes)
colnames (xy) <- c ("x", "y")

Node that sln2points simply returns "nodes" (in stplanr terminology), which are junction points. You can instead replace with coordinates of bus stops, or simply add those to this matrix. The following three lines convert those coordinates to unique (nearest) vertex IDs of the dodgr net object:

v <- dodgr_vertices (net)
pts <- match_pts_to_graph (v, xy)
pts <- v$id [pts]

To calculate distances between those pts on the network, just

d <- dodgr_dists (net, from = pts, to = pts)