lidaRtRee (version 4.0.8)

gap_detection: Gap detection in a Canopy Height Model

Description

Performs gaps detection on a canopy height model provided as object of class SpatRaster-class, or computed from the point cloud of objects of class LAS-class or LAScatalog-class. Function dem_filtering is first applied to the canopy height model to remove artefacts. Gaps are then extracted based on several criteria:

1. Vegetation height must be smaller than a threshold

2. Gap width must be large enough, depending on surrounding canopy height; distance to surrounding vegetation is tested with morphological closings

3. Gap must have a minimum surface

Usage

gap_detection(
  las,
  res = 1,
  ratio = 2,
  gap_max_height = 1,
  min_gap_surface = 25,
  max_gap_surface = +Inf,
  closing_height_bin = 1,
  nl_filter = "Median",
  nl_size = 3,
  gap_reconstruct = FALSE
)

Value

A SpatRaster object with three layers: gap labels, gap surface and canopy height model after filter.

Arguments

las

An object of class SpatRaster-class, LAS-class or LAScatalog-class

res

numeric. The size of a grid cell in point cloud coordinates units, used to rasterize the point cloud. In case the las argument is a SpatRaster res is not used.

ratio

numeric. maximum ratio between surrounding canopy height and gap distance (a pixel belongs to the gap only if for any vegetation pixel around it, the distance to the vegetation pixel is larger than pixel height/ratio). If ratio is set to NULL, this criterion is not taken into account

gap_max_height

numeric. maximum canopy height to be considered as gap

min_gap_surface

numeric. minimum gap surface

max_gap_surface

numeric. maximum gap surface

closing_height_bin

numeric. height bin width for morphological closing of gaps to test ratio between canopy height and gap distance

nl_filter

string. type of non-linear filter to apply to canopy height model to remove artefacts, should be an option of dem_filtering

nl_size

numeric. kernel width in pixel for non-linear filtering

gap_reconstruct

boolean. default behaviour is that areas that do not fulfill the ratio criterion are removed from gaps. If set to TRUE, in case some pixels of a gap fulfill the distance criterion, the connected pixels that fulfill the height criterion are also integrated to it.

See Also

dem_filtering, edge_detection

Examples

data(chm_chablais3)
chm_chablais3 <- terra::rast(chm_chablais3)

# fill NA values in canopy height model
chm_chablais3[is.na(chm_chablais3)] <- 0

# gap detection with distance larger than canopy height / 2
gaps <- gap_detection(chm_chablais3, ratio = 2, gap_max_height = 1, 
min_gap_surface = 0)

# gap detection with distance larger than canopy height / 2
# and reconstruction of border areas
gaps1 <- gap_detection(chm_chablais3,
  ratio = 2, gap_max_height = 1, min_gap_surface = 0,
  gap_reconstruct = TRUE
)

# gap detection without distance criterion
gaps2 <- gap_detection(chm_chablais3, ratio = NULL, gap_max_height = 1, 
min_gap_surface = 0)

# gap id and corresponding surface for third detection parameters
table(terra::values(gaps2$gap_id)) * terra::res(gaps2$gap_id)[1]^2

# plot original image
terra::plot(chm_chablais3, main = "Initial image")

# plot binary image of gaps
terra::plot(gaps$gap_id > 0, main = "Gaps", col = "green", legend = FALSE)
terra::plot(gaps1$gap_id > 0, main = "Gaps, with reconstruction", col = "green", legend = FALSE)
terra::plot(gaps2$gap_id > 0, main = "Gaps, no width criterion", col = "green", legend = FALSE)

# plot filtered CHM
terra::plot(gaps2$filled_chm, main = "Filtered CHM")