Is there an equivalent of dplyr::bind_rows (or the old plyr::rbind.fill) for spatial objects?

Question

bind_rows can append dataframes even when they do not have identical columns. It fills any non-matching columns with NAs.

Is there an equivalent for spatial point data frames in R?

I have two spdfs I want to combine (row wise) but they have almost no matching columns. I have had a look at the sf package but it doesn't have a method for this (yet?).

UPDATE:

I get an error message trying to replicate suggested solution

data(meuse)
coordinates(meuse) <- ~x+y
meuse <- st_as_sf(meuse)

m1 <- select(meuse, cadmium) 
m2 <- select(meuse, copper)

m3 <- bind_rows(m1, m2)
m3$geometry <- c(m1$geometry, m2$geometry)
m3 <- st_as_sf(m3)

> head(m3)
Error in .subset2(x, i, exact = exact) : 
 attempt to select less than one element in get1index

UPDATE 2:

Don't be an idiot like me and forget the crucial data.frames() step when using bind_rows

Answer 1

Since (I think) NA in geometry is not supported, you can bind_rows the data frames and then add on the geometry and convert.

nc1 and nc2 are two sf objects with differing columns:

> nc1
Simple feature collection with 100 features and 2 fields
geometry type:  MULTIPOLYGON
dimension:      XY
bbox:           xmin: -84.32385 ymin: 33.88199 xmax: -75.45698 ymax: 36.58965
epsg (SRID):    4267
proj4string:    +proj=longlat +datum=NAD27 +no_defs
First 20 features:
   NWBIR74 BIR79                       geometry
1       10  1364 MULTIPOLYGON(((-81.47275543...
2       10   542 MULTIPOLYGON(((-81.23989105...
3      208  3616 MULTIPOLYGON(((-80.45634460...
4      123   830 MULTIPOLYGON(((-76.00897216...
5     1066  1606 MULTIPOLYGON(((-77.21766662...
6      954  1838 MULTIPOLYGON(((-76.74506378...

> nc2
Simple feature collection with 100 features and 2 fields
geometry type:  MULTIPOLYGON
dimension:      XY
bbox:           xmin: -84.32385 ymin: 33.88199 xmax: -75.45698 ymax: 36.58965
epsg (SRID):    4267
proj4string:    +proj=longlat +datum=NAD27 +no_defs
First 20 features:
   SID79 NWBIR79                       geometry
1      0      19 MULTIPOLYGON(((-81.47275543...
2      3      12 MULTIPOLYGON(((-81.23989105...
3      6     260 MULTIPOLYGON(((-80.45634460...
4      2     145 MULTIPOLYGON(((-76.00897216...
5      3    1197 MULTIPOLYGON(((-77.21766662...

Then I can do this:

vv = bind_rows(data.frame(nc1),data.frame(nc2))
vv$geometry=c(nc1$geometry, nc2$geometry)
vv = st_as_sf(vv)

Which has the NA padding - head:

First 20 features:
   NWBIR74 BIR79 SID79 NWBIR79                       geometry
1       10  1364    NA      NA MULTIPOLYGON(((-81.47275543...
2       10   542    NA      NA MULTIPOLYGON(((-81.23989105...
3      208  3616    NA      NA MULTIPOLYGON(((-80.45634460...
4      123   830    NA      NA MULTIPOLYGON(((-76.00897216...
5     1066  1606    NA      NA MULTIPOLYGON(((-77.21766662...

tail:

proj4string:    +proj=longlat +datum=NAD27 +no_defs
    NWBIR74 BIR79 SID79 NWBIR79                       geometry
195      NA    NA     4     487 MULTIPOLYGON(((-77.14895629...
196      NA    NA     5    1023 MULTIPOLYGON(((-78.26149749...
197      NA    NA     3     763 MULTIPOLYGON(((-78.02592468...
198      NA    NA    17    1832 MULTIPOLYGON(((-78.65571594...
199      NA    NA     9    2100 MULTIPOLYGON(((-77.96073150...
200      NA    NA     6     841 MULTIPOLYGON(((-78.65571594...

Answer 2

I'm not aware of an sp method, but I think you could work around the problem in sf by creating an index column in both objects, like:

x <- mutate(x, JOINID = 1:nrow(x))
y <- mutate(y, JOINID = (nrow(x)+1):(nrowx() + nrow(y)))
left_join(x, y, by = 'JOINID')

make sure they're both in the same crs etc first, of course.

Answer 3

Similar to @Spacedman, I append dataframes and geometries separately, then bind them together into an sf object.

See here for a function that works similarly to bind_rows.

library(sf)
library(dplyr)

bind_rows_sf <- function(...){
  sf_list <- rlang::dots_values(...)[[1]]

  sfg_list_column <- lapply(sf_list, function(sf) sf$geometry[[1]]) %>% st_sfc
  df <- lapply(sf_list, function(sf) st_set_geometry(sf, NULL)) %>% bind_rows

  sf_appended <- st_sf(data.frame(df, geom=sfg_list_column))

  return(sf_appended)
}

sf_1 <- st_sf(data.frame(a=1, geom=st_sfc(st_point(0:1))))
sf_2 <- st_sf(data.frame(a=2, b=4, geom=st_sfc(st_point(1:2))))
sf_3 <- st_sf(data.frame(a=3, b=5, c=6, geom=st_sfc(st_point())))
sf_list <- list(sf_1, sf_2, sf_3)

sf_123 <- sf_list %>% bind_rows_sf
sf_123