Assign raster values to graph nodes

The function assignByRaster takes a gGraph object and a SpatRaster, assigns each raster cell to the nearest graph node, and collapses the values to a single scalar per node using a summary function.

Usage

assignByRaster(
  graph,
  raster,
  layer.name = "raster_points",
  fun = "mean",
  na.rm = TRUE,
  ...
)

Arguments

graph

a gGraph object.

raster

a SpatRaster object (from the terra package).

layer.name

a character string giving the name of the new node attribute. Defaults to "raster_points".

fun

a function or character string specifying how to summarize raster values within each node. Built-in options are "mean", "max", "min", "median", "sd", "any", and "all". Alternatively, pass any function that takes a numeric vector and returns a scalar. Defaults to "mean".

na.rm

logical. Whether to remove NA values before summarizing. Defaults to TRUE.

...

additional arguments passed to fun.

Value

A gGraph object with a new scalar node attribute named layer.name.

Details

This function is memory-intensive for high-resolution rasters or large graphs. Once values are assigned, further node attributes can be derived using setNodesAttr.

See also

setNodesAttr to set node attributes manually. assignByPolygon to assign attributes from GIS shapefiles.

Examples

## Make a new gGraph without any nodes attribute
geo.box <- c(xmin = -10, xmax = 30, ymin = 35, ymax = 60)
ggraph <- makeHexGrid(geo.box, spacing = 1000)
#> Resolution: 4, Area (km^2): 629710.644103813, Spacing (km): 783.739159045648, CLS (km): 895.60184164835

## Create a synthetic raster of random elevation values over the same region.
set.seed(42)
r <- terra::rast(
  xmin = geo.box["xmin"], xmax = geo.box["xmax"],
  ymin = geo.box["ymin"], ymax = geo.box["ymax"],
  resolution = 5, crs = "EPSG:4326"
)
terra::values(r) <- runif(terra::ncell(r))

## assign mean raster value per node
ggraph <- assignByRaster(ggraph, r, layer.name = "elevation", fun = "mean")
#> Spherical geometry (s2) switched off
#> although coordinates are longitude/latitude, st_nearest_feature assumes that
#> they are planar
#> Spherical geometry (s2) switched on

## assign standard deviation per node (useful for ruggedness)
ggraph <- assignByRaster(ggraph, r, layer.name = "ruggedness", fun = "sd")
#> Spherical geometry (s2) switched off
#> although coordinates are longitude/latitude, st_nearest_feature assumes that
#> they are planar
#> Spherical geometry (s2) switched on