Create a VRT execution plan from STAC items

vrt_plan() creates a lightweight metadata structure from rstac items without creating VRT files. This plan can be passed directly to vrt_warp() for efficient processing, skipping the intermediate VRT creation step that vrt_collect() performs.

Usage

vrt_plan(x, vsi_prefix = "", driver = "", ...)

# S3 method for class 'doc_items'
vrt_plan(x, vsi_prefix = "", driver = "", ...)

Arguments

x: A doc_items object from rstac.
vsi_prefix: Character string specifying the GDAL virtual file system prefix (e.g., "/vsicurl/", "/vsis3/"). Default is empty string which lets rstac determine the appropriate prefix.
driver: Character string specifying a driver prefix to prepend to source paths.
...: Additional arguments (unused).

Value

A vrt_plan object containing:

sources: List of items, each containing source URLs per asset
assets: Character vector of asset/band names
date_time: Character vector of datetimes (one per item)
n_items: Number of items in the plan
vsi_prefix: VSI prefix used
driver: Driver prefix used
config_options: GDAL configuration options

Details

Unlike vrt_collect(), vrt_plan() does not create any VRT files or make GDAL calls to read spatial metadata. This makes it faster when you intend to immediately warp the data to a common grid.

The trade-off is that spatial metadata (CRS, bounding box, resolution) is not available until warping occurs. Use vrt_collect() if you need to inspect or use the source spatial properties before warping.

Examples

if (FALSE) { # \dontrun{
# Query STAC items
items <- sentinel2_stac_query(
  bbox = c(-12.5, -37.5, -12.0, -37.0),
  start_date = "2023-01-01",
  end_date = "2023-01-31",
  max_cloud_cover = 10,
  assets = c("B02", "B03", "B04", "SCL")
)

# Create plan (fast, no VRT creation)
plan <- vrt_plan(items)

# Warp directly from plan
warped <- vrt_warp(
  plan,
  t_srs = "EPSG:32724",
  te = c(700000, 5850000, 710000, 5860000),
  tr = c(10, 10)
)
} # }