Optional extent argument in grid_canopy and grid_terrain

[jmmonnet]

Hi, when creating gridded data from las files, it is sometimes useful to be able to specify the extent in order to make analyses with other existing rasters easier (e.g. optical images). Could it be possible to add extent or (xmin, xmax, ymin, ymax) as optional arguments in grid_canopy and grid_terrain functions ? Thank you ! JM

[Jean-Romain]

Hi, thank you for the suggestion. 4 additional parameters sounds to much to me. But we can imagine to be able use a raster as input a return results only within the non empty cells of the given raster using an extent parameter. This solution sounds flexible and convenient to me and fulfill your requirements.

[Jean-Romain]

Warning: the following is expected to work but is currently in development (it should work but it is not documented and subject to modifications).

---

In grid_metric functions the parameter res can be a RasterLayer. In that case metrics will be computed only in non empty cells of the raster (all other parameters are not considered)

LASfile <- system.file("extdata", "Megaplot.laz", package="lidR")
las = readLAS(LASfile)
ext = extent(las)

# make a raster smaller than the data and with NAs
r = raster(ext-40, res = 20)
n = ncell(r)
x = runif(n)
x[sample(1:n, 10)] = NA
r[] <- x

# Display to check
plot(ext, asp = 1)
plot(r, add = T)

# Compute using a raster as reference
X = grid_metrics(las, mean(Z), res = r)
plot(X)

Also it works with a LAScatalog

ctg = catalog("~/Documents/Haliburton dataset/Landscape LiDAR/")
cores(ctg) <- 1

r = raster(ext + 500, res = 20)
n = ncell(r)
x = runif(n)
r[] <- x

plot(ctg)
plot(extent(r), add = T, col = "red")

X = grid_metrics(ctg, mean(Z), res = r)
plot(X)

---

To use this, install rlas v1.2.0 currently in the development branch of rlas and lidR v1.5.0in the development branch of lidR.

About grid_terrain , grid_canopy and grid_tincanopy. They are driven by other mechanisms. I will update this thread later.

[Jean-Romain]

Works also with grid_terrain

LASfile <- system.file("extdata", "Topography.laz", package="lidR")
las = readLAS(LASfile)

ext = extent(las)

r = raster(ext-100, res = 1)
n = ncell(r)
x = runif(n)
x[sample(1:n, 10)] = NA
r[] <- x

plot(ext)
plot(extent(r), add = T, col = "red")

dtm = grid_terrain(las, "delaunay", res = r)

ctg = catalog("~/Documents/Montmorency dataset/las/")
cores(ctg) <- 1

plot(ctg)
ext = extent(330000, 330800, 5230000, 5230800)

r = raster(ext, res = 1)
n = ncell(r)
x = runif(n)
x[sample(1:n, 10)] = NA
r[] <- x

dtm = grid_terrain(ctg, method = "delaunay", res = r)

[jmmonnet]

Thank you very much, I will install the devel. versions and tests it !

JM

[Jean-Romain]

I'm not done yet. Still missing grid_canopy which is more difficult to manage because it is written in C++

[zmpeg]

This looks awesome! Any chance it works for grid_tincanopy and catalog?

[Jean-Romain]

Sure it will. But I don't know when. It requires some free time. I guess I can do it relatively quickly for grid_tincanopy. As you can see I planned to do it for the release of 1.6.0.

[Jean-Romain]

6 months later (following the Debian mantra when it's ready), this suggestion is fully implemented in the draft of lidR v2.0.

• Officially documented
• I use the entiere raster as overlay. Not only the non NA cells as I did in previous examples.