如何从 st_coordinates 制作的坐标矩阵重建一个 sf 对象？

Question

st_coordinates 为简单特征列中的所有环提供坐标矩阵（对于每个简单特征几何、所有多边形的外环和孔）。有没有办法进行相反的操作，即从与st_coordinates 的输出格式相同的矩阵构建 sfc？

提醒一下，这里是st_coordinates 输出的结构：

        X   Y L1 L2 L3
 [1,]   0   0  1  1  1
 [2,] -10   0  1  1  1
 [3,] -10 -10  1  1  1
 [4,]   0 -10  1  1  1
 [5,]   0   0  1  1  1
 [6,]   1   1  1  1  2
 [7,]  11   1  1  1  2
 [8,]  11  11  1  1  2
 [9,]   1  11  1  1  2
[10,]   1   1  1  1  2
[11,]   2   2  2  1  2
[12,]   3   2  2  1  2
[13,]   3   3  2  1  2
[14,]   2   3  2  1  2
[15,]   2   2  2  1  2
[16,]   5  -2  1  2  2
[17,]  10  -2  1  2  2
[18,]  10  -1  1  2  2
[19,]   5  -1  1  2  2
[20,]   5  -2  1  2  2

对应这张图片：

如何制作sf对象？

Answer 1

通过使用 data.frame 可以做到：

library(sf)
library(tidyverse)
d <- 
  structure(list(X = c(0L, -10L, -10L, 0L, 0L, 1L, 11L, 11L, 1L, 
                       1L, 2L, 3L, 3L, 2L, 2L, 5L, 10L, 10L, 5L, 5L), 
                 Y = c(0L, 0L, -10L, -10L, 0L, 1L, 1L, 11L, 11L, 1L, 2L, 2L, 3L, 
                       3L, 2L, -2L, -2L, -1L, -1L, -2L),
                 L1 = c(1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 2L, 2L, 2L, 2L,
                        2L, 1L, 1L, 1L, 1L, 1L), 
                 L2 = c(1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L,
                        1L, 2L, 2L, 2L, 2L, 2L), 
                 L3 = c(1L, 1L, 1L, 1L, 1L, 2L, 2L, 2L, 2L, 2L, 2L, 2L, 2L, 2L, 
                        2L, 2L, 2L, 2L, 2L, 2L)), 
            class = "data.frame", row.names = c(NA, -20L))   %>% 
  st_as_sf(coords = c("X", "Y")) %>% 
  group_by(L1, L2, L3) %>% 
  summarise(do_union = FALSE) %>% 
  st_cast("LINESTRING")
# to visualise
d %>% 
  ggplot() +
  geom_sf()
# to generate polygons
d %>% 
  st_cast("POLYGON") %>% 
  ggplot() +
  geom_sf()

附录

对 Jean-Luc Dupouey 关于洞的正义评论的回应，我想这段代码应该解决这个问题：

d.po <-
  d %>% 
  group_by(L2, L3) %>% 
  summarise(do_union = FALSE) %>% 
  st_cast("MULTILINESTRING") %>% 
  st_cast("MULTIPOLYGON") %>% 
  ungroup()
# visualise
d.po %>% 
  unite("group", L2:L3) %>% 
  ggplot() +
  geom_sf(aes(fill = group))

Answer 2

是的，你可以。一种可能的方法是使用sf::st_polygon() 调用。

调用需要列表格式的参数 - 第一项是外边界，然后是所需的孔。

以这个代码为例，它建立在广为人知且深受喜爱的 NC shapefile 之上。

它所做的是将梅克伦堡县分解为边界坐标矩阵，然后将其构建为sfg >> sfc >> sf 对象。

library(sf)
library(dplyr)
library(ggplot2)

shape <- st_read(system.file("shape/nc.shp", package="sf")) %>%  
  filter(CNTY_ID == 2041) # Mecklenburg, as in Charlotte of Mecklenburg-Strelitz

#coordinates of Mecklenburg cnty
coords <- st_coordinates(shape)

class(coords)
[1] "matrix" "array" 

head(coords)
             X        Y L1 L2 L3
[1,] -81.04930 35.15153  1  1  1
[2,] -81.02396 35.14903  1  1  1
[3,] -81.00728 35.16325  1  1  1
[4,] -81.00152 35.19599  1  1  1
[5,] -81.01405 35.24990  1  1  1
[6,] -80.97964 35.33327  1  1  1

#a polygon from coords
meck_cnty <- st_polygon(x = list(coords[, 1:2])) %>%  # just X and Y please
  st_sfc() %>%  # from sfg to sfc
  st_sf() # from sfc to sf

class(meck_cnty)
[1] "sf"         "data.frame"

plot(meck_cnty)