How to create multiline string sf object to create a time slider in Leaflet map - mapping

Question

I'm trying to create a leaflet map which shows different movement paths for different months of the year. I.e. I've got a dataset showing multiple journeys per month and I want to display the movement paths separately for each month using the addTimeslider feature of the leaflet.extras2 package.
To do so I have been trying to adapt the code posted by SymbolixAU I found here: leaflet add multiple polylines
This code uses sf functions including st_linestring to create an object that can be supplied to a addPolylines leaflet function to show all movement paths at once.
I'm pretty sure for my purposes (showing data separately for each month) I have to use st_multilinestring, which takes a list of matrices containing the coordinates for multiple polylines per row (with one row per month) rather than a single polyline per row.
Once I have that I think I could supply that object to the addTimeslider function of leaflet.extras2 to achieve what I need. I'm quite sure of this because when I used the sf object created using sf_linestring inthe AddTimeslider feature I was able to use the time slider on the map to individual movement paths at a time.
However, I have been trying for hours and haven't been successful. Would be hugely grateful for any pointers, please and thank you.
Some example data:
#load packages
library(dplyr)
library(leaflet)
library(leaflet.extras2)
library(sf)
library(data.table)
# create the example dataset
data <- structure(list(arrival_month = structure(c(3L, 3L, 4L, 4L, 4L,
5L, 5L, 6L, 6L, 6L), .Label = c("January", "February", "March",
"April", "May", "June", "July", "August", "September", "October",
"November", "December"), class = c("ordered", "factor")), start_lat = c(33.40693,
33.64672, 33.57127, 33.42848, 33.54936, 33.53418, 33.60399, 33.49554,
33.5056, 33.61696), start_long = c(-112.0298, -111.9255, -112.049,
-112.0998, -112.0912, -112.0911, -111.9273, -111.9687, -112.0563,
-111.9866), finish_lat...4 = c(33.40687, 33.64776, 33.57125,
33.42853, 33.54893, 33.53488, 33.60401, 33.49647, 33.5056, 33.61654
), finish_lat...5 = c(-112.0343, -111.9303, -112.0481, -112.0993,
-112.0912, -112.0911, -111.931, -111.9711, -112.0541, -111.986
)), row.names = c(NA, -10L), class = c("data.table", "data.frame"
), .internal.selfref = <pointer: 0x0000026e5df41ef0>)
My attempt at the code:
# Convert the data into a list of matrices for each month
mnths <- c("May","March","April","June")
mat_list <- list()
for (i in mnths) {
month <- as.matrix(data %>% filter(arrival_month == i) %>% select(-1))
mat_list[[i]] <- month
}
# convert to an sf object
data_DT <- setDT(data)
sf <- data_DT[
, {
geometry <- sf::st_multilinestring(x = mat_list)
geometry <- sf::st_sfc(geometry)
geometry <- sf::st_sf(geometry = geometry)
}
, by = arrival_month
]
sf <- sf::st_as_sf(sf)
This yields the following result:
It's not correct because each row contains the coordinates for all the months, rather than just for the month in the respective row. I'm at a loss as to where to go from here - any help would be hugely appreciated.
Thanks

Answer 1

I would do it slightly differently today, making use of {sfheaders} to build the linestrings.
library(sfheaders)
library(sf)
library(data.table)
setDT( data )
data[, line_id := .I ] ## Assuming each row is a line
## create a long-form of the data
dt_line <- rbindlist(
list(
data[, .(arrival_month, line_id, lon = start_long, lat = start_lat, sequence = 1)]
, data[, .(arrival_month, line_id, lon = finish_lat...5, lat = finish_lat...4, sequence = 2)] ## I think 'finish_lat...5' is actually the 'long'
)
)
setorder(dt_line, line_id, sequence)
sf <- sfheaders::sf_multilinestring(
obj = dt_line
, x = "lon"
, y = "lat"
, multilinestring_id = "arrival_month"
, linestring_id = "line_id"
, keep = T
)
sf::st_crs( sf ) <- 4326 ## Assuming it's in Web Mercator
# Simple feature collection with 4 features and 2 fields
# Geometry type: MULTILINESTRING
# Dimension: XY
# Bounding box: xmin: -112.0998 ymin: 33.40687 xmax: -111.9255 ymax: 33.64776
# Geodetic CRS: WGS 84
#. arrival_month sequence geometry
# 1 3 1 MULTILINESTRING ((-112.0298...
# 2 4 1 MULTILINESTRING ((-112.049 ...
# 3 5 1 MULTILINESTRING ((-112.0911...
# 4 6 1 MULTILINESTRING ((-111.9687...
Note the arrival_month has been re-coded to it's factor levels.

Answer 2

I managed to adapt the code provided by SymbolixAU to suit my purposes, by using sf_multilinestring instead of sf_line string:
## Generate data
data <- structure(list(arrival_month = structure(c(3L, 3L, 4L, 4L, 4L,
5L, 5L, 6L, 6L, 6L), .Label = c("January", "February", "March",
"April", "May", "June", "July", "August", "September", "October",
"November", "December"), class = c("ordered", "factor")), start_lat = c(33.40693,
33.64672, 33.57127, 33.42848, 33.54936, 33.53418, 33.60399, 33.49554,
33.5056, 33.61696), start_long = c(-112.0298, -111.9255, -112.049,
-112.0998, -112.0912, -112.0911, -111.9273, -111.9687, -112.0563,
-111.9866), finish_lat...4 = c(33.40687, 33.64776, 33.57125,
33.42853, 33.54893, 33.53488, 33.60401, 33.49647, 33.5056, 33.61654
), finish_lat...5 = c(-112.0343, -111.9303, -112.0481, -112.0993,
-112.0912, -112.0911, -111.931, -111.9711, -112.0541, -111.986
)), row.names = c(NA, -10L), class = c("data.table", "data.frame"
))
## Add id column and convert to data table
data[, line_id := .I ]
setDT(data)
## create a long-form of the data
dt_line <- rbindlist(
list(
data[, .(arrival_month, line_id, lon = start_long, lat = start_lat, sequence = 1)]
, data[, .(arrival_month, line_id, lon = finish_lat...5, lat = finish_lat...4, sequence = 2)] ## I think 'finish_lat...5' is actually the 'long'
)
)
setorder(dt_line, line_id, sequence)
## Create multistring sf object
sf <- sfheaders::sf_multilinestring(
obj = dt_line
, x = "lon"
, y = "lat"
, linestring_id = "line_id"
, multilinestring_id = "arrival_month"
, keep = TRUE
)
sf::st_as_sf(sf)
## Convert arrival_month back to factor
sf$arrival_month <- as.factor(sf$arrival_month)
Might not be the most elegant but this does the job.
The key was to use sf_multilinestring and include/specify both linestring_id and multilinestring_id to distinguish between separate sets of polylines.
This is the result:
Now when this sf object is used in AddTimeslider it behaves as desired.
Credit to SymbolixAU for most of the code and bringing my attention to the sfheaders package