Generate spatio-temporal cross-validation index with spatialsample::spatial_block_cv and year-based temporal folds

This function generates spatio-temporal cross-validation indices with v spatial blocks and year-based temporal folds. The spatial blocks are generated with spatialsample::spatial_block_cv function, and the temporal folds are generated based on the years availble in data$time. Total number of folds is equal to v * length(unique(substr(data$time, 1, 4))).

Usage

generate_cv_index_spt(
  data,
  crs = NULL,
  cellsize = NULL,
  locs_id = "site_id",
  coords = c("lon", "lat"),
  v = 10L,
  time_id = "time"
)

Arguments

data

data.table or data.frame with id, coords, and time columns.

crs

The coordinate reference system (CRS) of the spatial object

cellsize

The size of each spatial block in meters. Uses the crs units

locs_id

The column name in data that represents the location identifier.

coords

The column names in the spatial object that represent the XY coordinates. Default is c("lon", "lat").

v

integer(1). The number of partitions for the resampling.

time_id

The column name in data that represents the time values.

See also

spatialsample::spatial_block_cv