raster::rasterToContour;轮廓线不连续

Question

我正在尝试使用 R 中的 raster 包从栅格 object 中提取等高线。

rasterToContour 似乎工作良好并且绘制得很好，但在调查时发现轮廓线被分解成不规则的段。来自 ?rasterToContour

的示例数据

library(raster)

f <- system.file("external/test.grd", package="raster")
r <- raster(f)
x <- rasterToContour(r)
class(x)
plot(r)
plot(x, add=TRUE)

我正在尝试提取栅格中示例站点的等高线。因此，我们选择一个随机站点，提取其高程，然后再次运行 rasterToContour()，指定等高线 level.

的高程

# our sample site - a random cell chosen on the raster
xyFromCell(r, 5000) %>%
  SpatialPoints(proj4string = crs(r)) %>%
  {. ->> site_sp} %>%
  st_as_sf %>%
  {. ->> site_sf}

# find elevation of sample site, and extract contour lines
extract(r, site_sf) %>%
  {. ->> site_elevation}

# extract contour lines
r %>%
  rasterToContour(levels = site_elevation) %>%
  {. ->> contours_sp} %>%
  st_as_sf %>%
  {. ->> contours_sf}

# plot the site and new contour lines (approx elevation 326)
plot(r)
plot(contours_sf, add = TRUE)
plot(site_sf, add = TRUE)

# plot the contour lines and sample site - using sf and ggplot
ggplot()+
  geom_sf(data = contours_sf)+
  geom_sf(data = site_sf, color = 'red')

然后我们使用 st_intersects 找到与站点相交的线（缓冲区宽度为 100 以确保它接触线）。但是，这 returns 所有的轮廓线。

contours_sf %>%
  filter(
    st_intersects(., site_sf %>% st_buffer(100), sparse = FALSE)[1,]
  ) %>%
  ggplot()+
  geom_sf()

我认为所有行都已返回，因为它们看起来是单个 MULTILINESTRING sf object.

contours_sf

# Simple feature collection with 1 feature and 1 field
# geometry type:  MULTILINESTRING
# dimension:      XY
# bbox:           xmin: 178923.1 ymin: 329720 xmax: 181460 ymax: 333412.3
# CRS:            +proj=sterea +lat_0=52.1561605555556 +lon_0=5.38763888888889 +k=0.9999079 +x_0=155000 +y_0=463000 +datum=WGS84 +units=m +no_defs
# level                       geometry
# C_1 326.849822998047 MULTILINESTRING ((179619.3 ...

因此，我使用 ngeo::st_segments() 将 contours_sf MULTILINESTRING 拆分为单独的行（我找不到任何 sf 方法来执行此操作，但我是开放的使用替代方法，特别是如果这是问题所在）。

没想到这个returns394个特征；通过查看该图，我预计大约有 15 条单独的线。

contours_sf %>%
  nngeo::st_segments()

# Simple feature collection with 394 features and 1 field
# geometry type:  LINESTRING
# dimension:      XY
# bbox:           xmin: 178923.1 ymin: 329720 xmax: 181460 ymax: 333412.3
# CRS:            +proj=sterea +lat_0=52.1561605555556 +lon_0=5.38763888888889 +k=0.9999079 +x_0=155000 +y_0=463000 +datum=WGS84 +units=m +no_defs
# First 10 features:
#   level                         result
# 1  326.849822998047 LINESTRING (179619.3 329739...
# 2  326.849822998047 LINESTRING (179580 329720.4...
# 3  326.849822998047 LINESTRING (179540 329720, ...
# 4  326.849822998047 LINESTRING (179500 329735.8...
# 5  326.849822998047 LINESTRING (179495.3 329740...
# 6  326.849822998047 LINESTRING (179460 329764, ...
# 7  326.849822998047 LINESTRING (179442.6 329780...
# 8  326.849822998047 LINESTRING (179420 329810, ...
# 9  326.849822998047 LINESTRING (179410.2 329820...
# 10 326.849822998047 LINESTRING (179380 329847.3...

然后，当我们过滤以仅保留与站点相交的线（缓冲区宽度为 100）时，仅返回预期轮廓线的一小部分（线的红色部分，我假设反映了100 缓冲区宽度）。

contours_sf %>%
  nngeo::st_segments() %>%
  filter(
    # this syntax used as recommended by this answer 
    st_intersects(., site_sf %>% st_buffer(100), sparse = FALSE)
  ) %>%

  ggplot()+
  geom_sf(colour = 'red', size = 3)+
  geom_sf(data = contours_sf)+
  geom_sf(data = site_sf, colour = 'cyan')+
  geom_sf(data = site_sf %>% st_buffer(100), colour = 'cyan', fill = NA)

任何人对以下几点有想法：

解释为什么轮廓线是'broken'
提供一种有效的方法'joining'把碎片拼起来
nngeo::st_segments() 的替代方案，如果这实际上是 394 行的来源而不是 ~15

Answer 1

将 MULTILINESTRING 转换为 LINESTRING 似乎可以满足您的需要：

contours_sf %>% st_cast("LINESTRING") %>% 
  filter(st_intersects(., st_buffer(site_sf, 100), sparse=FALSE)[,1]) %>%

ggplot()+
  geom_sf(data = contours_sf)+
  geom_sf(data = site_sf, color = 'red') +
  geom_sf(color = 'pink')

Answer 2

如果您从分解行开始，也许效果会更好

library(raster)
f <- system.file("external/test.grd", package="raster")
r <- raster(f)
x <- rasterToContour(r)
x <- disaggregate(x)

或 terra

library(terra)
r <- rast(f)
x <- as.contour(r)

x
# class       : SpatVector 
# geometry    : lines 
# dimensions  : 8, 1  (geometries, attributes)

x <- disaggregate(x)
x
# class       : SpatVector 
# geometry    : lines 
# dimensions  : 56, 1  (geometries, attributes)

你可以这样继续

y <- st_as_sf(x)

或者像这样

r <- rast(system.file("ex/meuse.tif", package="terra"))
site <- vect(xyFromCell(r, 5000), crs=crs(r))
elevation <- extract(r, site)
v <- disaggregate(as.contour(r, levels=elevation))
i <- which.min(distance(site, v))
vv <- v[i]

plot(r)
lines(v)
lines(vv, col="red", lwd=2)
points(site, col="blue", cex=2)

raster::rasterToContour;轮廓线不连续

raster::rasterToContour; contour lines are not continuous

r

raster

r-raster

sf