Apply Functions Across Spatial Tiles

For more useful params info and examples, see the Tile Processing Methods section.

Apply a function across spatial tiles to speed up processing and manage memory usage for large data operations. This is a landing page for the generic. tileApply() dispatches to different processing methods based on the tile type.

character inputs to x and y are assumed to be {terra} readable. SpatRaster and SpatVector must first be written to file. If provided, they are traced to their filepaths with terra::sources() and then processed via their filepaths for memory efficiency.

For other data types, see extending_tilework

Usage

# S4 method for class 'ANY,missing,ANY'
tileApply(x, tiles, verbose = NULL, ...)

# S4 method for class 'ANY,ANY,ANY'
tileApply(x, y, tiles, verbose = NULL, ...)

# S4 method for class 'token,ANY,ANY'
tileApply(x, y, tiles, verbose = NULL, ...)

Arguments

x

input data 1

tiles

tile* object (tilePlan, tileGroup, or tileIterator)

verbose

verbosity. TRUE, FALSE or "debug" for more info on stack tracing.

...

additional arguments passed to specific methods (one of which is the FUN function applied across the tiles.)

y

input data 2 (optional)

Tile Processing Methods

• Basic tiling: See tileApply-plan for spatialTilePlan and pixelTilePlan

• Group processing: See tileApply-group for tileGroup hierarchical processing

• Iterator processing: See tileApply-iterator for tileIterator streaming/batch processing

Boundary Inclusivity

Adjacent tiles share exact boundaries. Since {tilework} does not know the format or representation of the underlying data, it provides tile bounds as windows but does not enforce whether those boundaries are inclusive or exclusive — that is determined by the getBoundedData() method for the data type being processed.

For raster data this is generally not an issue (pixel snapping assigns each cell unambiguously). For point or tabular data, features sitting exactly on a shared tile boundary may appear in multiple tiles unless the getBoundedData() implementation applies its own inclusive/exclusive filtering. Implementing packages can use the tile grid position (available via getTile()'s get_params) to determine which edges are shared and filter accordingly.

See also

tileApply-plan, tileApply-group, tileApply-iterator

Other tile processing: getBoundedData(), getTile(), tileApply-group, tileApply-iterator, tileApply-plan

Examples

# See specific help pages for detailed examples:
# ?`tileApply-plan`     # Basic spatial/pixel tiling
# ?`tileApply-group`    # Hierarchical tile groups
# ?`tileApply-iterator` # Streaming batch processing