Найти внешний край многоугольника в R

В этом решении используется самая северная и самая южная точки, чтобы разделить границу с lwgeom. Нужный «край» нужно выбрать вручную (две линии из трех). В качестве примера я использовал meuse. Граница получается с помощью st_concave_hull(ratio=0.1) и на ней проверяются точки. В зависимости от вогнутости может потребоваться регулировка.

library(tidyverse)
library(sf)
#> Linking to GEOS 3.11.2, GDAL 3.7.2, PROJ 9.3.0; sf_use_s2() is TRUE
library(lwgeom) # for lwgeom::st_split
#> Warning: le package 'lwgeom' a été compilé avec la version R 4.3.3
#> Linking to liblwgeom 3.0.0beta1 r16016, GEOS 3.11.2, PROJ 9.3.1
#> Warning in fun(libname, pkgname): PROJ versions differ: lwgeom has 9.3.1 sf has
#> 9.3.0
#> 
#> Attachement du package : 'lwgeom'
#> L'objet suivant est masqué depuis 'package:sf':
#> 
#>     st_perimeter
library(sp) # for meuse

data(meuse) 
df_sf = meuse %>% st_as_sf(coords= c("x","y"),remove= FALSE) %>% 
  mutate(ID= 1:n())

ggplot(data=df_sf)+geom_sf()

# Boundary
st_simplify(df_sf %>% summarise()  %>% st_concave_hull(ratio=0.1) %>% st_cast(to  = "LINESTRING")) %>% 
  st_as_sfc() %>% ggplot(data=.)+geom_sf()

# Getting the points
border_point =   df_sf %>% summarise()  %>% st_concave_hull(ratio=0.1) %>% st_cast(to  = "POINT") %>% 
  st_join(df_sf, .predicate = st_equals) 

# Choosing start/end point. 
MinMax =    bind_rows(border_point %>% slice_max(y), 
                      border_point %>% slice_min(y))

# Splitting along those points
boundary_split = 
lwgeom::st_split(df_sf %>% summarise()  %>% st_concave_hull(ratio=0.1) %>% st_cast(to  = "LINESTRING"),
                 MinMax ) %>% 
  st_collection_extract(., "LINESTRING") %>% 
  mutate(id_line = 1:n())

# Three lines are generated, since the start/end of the linestring acts as a splitting point as well
# Plotting to identify the releveants id_line (left shore), 
boundary_split %>% 
  ggplot(data=.)+geom_sf( aes(col = id_line %>% as.factor()))

# manually selecting 
left_boundary = boundary_split %>% filter( id_line %in% c(1,2)) 

left_points =  st_join(df_sf,left_boundary  ) %>%  
  # Removing unjoinned point (NA), there is certainly a better way
  filter( id_line %in% c(1,2)) 

# Ploting elements

ggplot()+
  geom_sf(data=df_sf %>% summarise()  %>% st_concave_hull(ratio=0.1),fill=NA)+
  geom_sf(data=df_sf,col  = "grey")+
  geom_sf(data =MinMax, col  = "green", size =5)+
  geom_sf(data = boundary_split,aes(col = id_line %>% as.factor()) )+
  geom_sf(data = left_points, col  = "red")

^{Created on 2024-04-16 with reprex v2.1.0}