coregistration: Tree inventory and canopy height model coregistration

Description

Computes the correlation between the canopy height model and a virtual canopy height model simulated from tree locations, for different translations of tree inventory positions, and outputs the translation corresponding to best estimated co-registration.

Usage

coregistration(chm, trees, mask, buffer = 19, step = 0.5, dm = 2, plot = TRUE)

Value

A list with two elements : first the correlation SpatRaster returned by rasters_moving_cor, second a data.frame returned by raster_local_max

Arguments

chm: raster. canopy height model
trees: data.frame. the first two columns contain xy coordinates, and the third is the value to correlate to the chm (e.g. tree heights or diameters)
mask: raster. raster mask of tree inventory area
buffer: numeric. radius of the circular buffer area of possible translations
step: numeric. increment step of translations within buffer area to compute correlation values, should be a multiple of raster resolution
dm: numeric. minimum distance between two local maxima in meters
plot: boolean. whether to display the results or not

References

Monnet, J.-M. and Mermin, E. 2014. Cross-Correlation of Diameter Measures for the Co-Registration of Forest Inventory Plots with Airborne Laser Scanning Data. Forests 2014, 5(9), 2307-2326, tools:::Rd_expr_doi("10.3390/f5092307")

Examples

Run this code

# tree inventory
trees <- data.frame(x = c(22.2, 18.3, 18.1), y = c(22.1, 22.7, 18.4), 
z = c(15, 10, 15))

# mask of inventory area
# empty raster with extent
tree_mask <- circle2Raster(20, 20, 9, resolution = 1)
# fill binary mask
tree_mask <- raster_xy_mask(rbind(c(20, 20), c(20, 20)), c(9, 9), tree_mask, 
binary = TRUE)

# simulate chm raster
chm <- terra::rast(extent = c(0, 40, 0, 40), resolution = 1, crs = NA)
xy <- terra::xyFromCell(chm, 1:(ncol(chm) * nrow(chm)))

# add Gaussian surfaces to simulate tree crowns
z1 <- trees$z[1] * exp(-((xy[, 1] - trees$x[1])^2 + (xy[, 2] - trees$y[1])^2 / 2) * trees$z[1] / 50)
z2 <- trees$z[2] * exp(-((xy[, 1] - trees$x[2])^2 + (xy[, 2] - trees$y[2])^2 / 2) * trees$z[2] / 50)
z3 <- trees$z[3] * exp(-((xy[, 1] - trees$x[3])^2 + (xy[, 2] - trees$y[3])^2 / 2) * trees$z[3] / 50)
chm <- terra::rast(cbind(xy, pmax(z1, z2, z3)), type = "xyz") #+rnorm(length(z1),0,1)))

# translate trees
trees$x <- trees$x + 1
trees$y <- trees$y + 2

coreg <- coregistration(chm, trees, mask = tree_mask, buffer = 5, step = 1, dm = 1, plot = FALSE)
coreg$local_max[, c("dx1", "dy1")]

# plot raster
terra::plot(coreg$correlation_raster)
abline(h = 0, lty = 2)
abline(v = 0, lty = 2)
# add location of two local maxima
graphics::points(coreg$local_max[1, c("dx1", "dx2")],
  coreg$local_max[1, c("dy1", "dy2")],
  cex = c(1, 0.5), pch = 3, col = "red"
)