r gis：用sf识别多边形之间的内边界

Question

这是my previous similar question的更新，我需要在sf框架内执行相同的任务。

我需要识别此地图中多边形，红线之间的内部边界。

在sp框架内，我曾经使用自编的函数来包装@Spasdman的answer。这是：

identify_borders <- function(SPolyDF){
    require(rgeos)
    require(sp)
    borders <- gDifference(
            as(SPolyDF,"SpatialLines"),
            as(gUnaryUnion(SPolyDF),"SpatialLines"),
            byid=TRUE)

    df <- data.frame(len = sapply(1:length(borders), 
                                  function(i) gLength(borders[i, ])))
    rownames(df) <- sapply(1:length(borders), 
                           function(i) borders@lines[[i]]@ID)

    SLDF <- SpatialLinesDataFrame(borders, data = df)
    return(SLDF)
}

或者，可以使用raster::boundaries()。

获取空间数据并复制地图的代码

# dev version of ggplot2 for geom_sf()
devtools::install_github("tidyverse/ggplot2")

library(tidyverse)
library(sf)

load(url("https://ikashnitsky.github.io/share/1712-so-q-identify-borders/geodata.Rdata"))

ggplot() + 
        geom_sf(data = gd_nuts0) + 
        geom_sf(data = gd_borders, color = "red") + 
        coord_sf(datum = NA) + 
        theme_void()

Answer 1

rmapshaper使用javascript，即作弊！我试过了：

i = st_intersection(gd_nuts0, gd_nuts0)
i2 <- i[i$nuts_id != i$nuts_id.1,]
plot(gd_nuts0[1])
plot(i2, add = TRUE, col  ='red', lwd = 2)

Answer 2

事实证明rmapshaper具有正确的功能，可以很好地使用sf个对象 - ms_innerlines()。唯一的困难（可能是错误）是ms_innerlines()返回列表而不是sf对象。但这种奇怪的行为很容易解决。以下是解决方案代码。请注意，我简化了初始多边形以查看差异。从sf对象创建的新内边框以海军蓝色绘制。

# dev version of ggplot2 for geom_sf()
devtools::install_github("tidyverse/ggplot2")

library(tidyverse)
library(sf)
library(rmapshaper)

load(url("https://ikashnitsky.github.io/misc/171211-so-question-identify-borders/geodata.Rdata"))

sf_poly_simp <- gd_nuts0 %>% 
        ms_simplify(keep = .2, keep_shapes = TRUE) 

sf_bord_simp <- sf_poly_simp %>% 
        ms_innerlines() %>% 
        as_tibble() %>% 
        st_as_sf()

ggplot() + 
        geom_sf(data = sf_poly_simp) + 
        geom_sf(data = sf_bord_simp, color = "navy", size = 1) + 
        geom_sf(data = gd_borders, color = "red", size = .1) + 
        coord_sf(datum = NA) + 
        theme_void()