3

I am attempting to clip sf lines where they overlap with polygons in R - the opposite of the clip in the image below:

enter image description here

I've created a reprex below, where the blue lines outside of the circle should be what is left after the clip.

library(sf)
#> Linking to GEOS 3.10.2, GDAL 3.4.2, PROJ 8.2.1; sf_use_s2() is TRUE
library(tmap)

begin.coords = data.frame(lon = -78,
                          lat = 43.45)
end.coords = data.frame(lon = -78,
                        lat = 43.55)

l_sf2 <- vector("list", nrow(begin.coords))
for (i in seq_along(l_sf2)){
  l_sf2[[i]] <- st_linestring(as.matrix(rbind(begin.coords[i, ], end.coords[i,])))
}

line_df = st_sfc(l_sf2) %>%
  st_as_sf() %>%
  st_set_crs(.,4326) %>%
  st_transform(.,crs=32618)

poly_df = data.frame(lon = -78,
                     lat = 43.5) %>%
  st_as_sf(coords=c('lon','lat')) %>%
  st_set_crs(.,4326) %>%
  st_transform(.,crs=32618) %>%
  st_buffer(.,dist=4000)

tm_shape(line_df)+
  tm_lines(col='blue')+
tm_shape(poly_df)+
  tm_polygons(col='red',alpha=.5)

Created on 2023-07-12 with reprex v2.0.2

I understand this process is likely not an easy one, given that line geometry is defined by starting and ending points, thus, such a clip would have to create separate lines with new starting and ending points.

EDIT:

After incorporating Spacedman's answer, it seems that the real issue occurs when there are multiple linestrings that change direction. In the reprex below, it appears as though st_difference() looks to clip lines only when they are overlapping locations where all of the polygons overlap:

library(sf)
library(tmap)
library(dplyr)

df1 = data.frame(lon=c(-78,-77.95,-77.9,-77.85,-77.8),
                lat=c(43.45,43.5,43.55,43.5,43.45),
                id = c(1:5)) %>%
  st_as_sf(coords=c('lon','lat'),crs=4326) %>%
  st_transform(crs=32618)

df2 = data.frame(lon=c(-78,-77.95,-77.9,-77.85,-77.8),
                 lat=c(43.5,43.45,43.5,43.45,43.5),
                 id=c(1:5)) %>%
  st_as_sf(coords=c('lon','lat'),crs=4326) %>%
  st_transform(crs=32618)

df1 %>%
  summarise(do_union = F) %>%
  st_cast("LINESTRING") -> sf_line

df2 %>%
  summarise(do_union = F) %>%
  st_cast("LINESTRING") -> sf_line2

sf_bothlines = rbind(sf_line,sf_line2)

poly_df = data.frame(lon = c(-77.9,-77.95),
                     lat = c(43.5,43.5)) %>%
  st_as_sf(coords=c('lon','lat')) %>%
  st_set_crs(.,4326) %>%
  st_transform(.,crs=32618) %>%
  st_buffer(.,dist=4000)

tm_shape(sf_bothlines)+
  tm_lines(col='blue')+
  tm_shape(poly_df)+
  tm_polygons(col='red',alpha=.5)


clipped=st_difference(sf_bothlines,poly_df)

tm_shape(clipped)+
  tm_lines(col='blue')+
  tm_shape(poly_df)+
  tm_polygons(col='red',alpha=.5)

Created on 2023-07-12 with reprex v2.0.2

0

1 Answer 1

3

Use st_difference:

> dd = st_difference(line_df, poly_df)
> plot(dd)

enter image description here

Note that for this to work as a "clip" to the polygon you need the polygon as a single feature, otherwise the lines will get clipped to each feature in turn, possibly resulting in many lines inside feature A because they were the bits resulting from the difference with feature B. To fix this, unite all the polygons into a single feature using st_union.

So given your object with two features:

> poly_df
Simple feature collection with 2 features and 0 fields
Geometry type: POLYGON
Dimension:     XY
Bounding box:  xmin: 257503.6 ymin: 4816427 xmax: 269546.1 ymax: 4824570
Projected CRS: WGS 84 / UTM zone 18N
                        geometry
1 POLYGON ((269546.1 4820427,...
2 POLYGON ((265503.6 4820570,...

The union creates a single combined feature:

> st_union(poly_df)
Geometry set for 1 feature 
Geometry type: POLYGON
Dimension:     XY
Bounding box:  xmin: 257503.6 ymin: 4816427 xmax: 269546.1 ymax: 4824570
Projected CRS: WGS 84 / UTM zone 18N
POLYGON ((269546.1 4820427, 269540.6 4820218, 2...

Then the difference of the lines with the union is:

> plot(st_difference(sf_bothlines, st_union(poly_df)))
> plot(poly_df, add=TRUE, col="#80808080")

enter image description here

6
  • Weird. I swear this wasn't working when I tried earlier. ¯_(ツ)_/¯ Jul 12, 2023 at 19:56
  • I do get some weird behavior when doing this for curved lines though - 3,130 lines becomes over 466,000 lines. Is there a reason for this or a way around it? Jul 12, 2023 at 20:04
  • It could happen, depends how often your lines cut the polygon, they could could get chopped up into hundreds of pieces. See what happens to a small sample of those 3130.
    – Spacedman
    Jul 12, 2023 at 21:02
  • It's behaving very strangely. It's still creating large numbers - 31 lines is turning into 4,619, but they are still overlapping and aren't being clipped. I'll try to create another reprex. Jul 13, 2023 at 0:13
  • I made a major edit after figuring out what the problem is. Since this answered the original question, I will leave it as accepted until another answer that addresses the edit is posted. Jul 13, 2023 at 0:48

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.