Skip to contents

Describe prediction support and extrapolation

ps_prediction_support.Rd

Classifies raster cells using the training-point convex hull, distance to the nearest observation, an optional maximum distance, mask membership, and finite prediction availability. The convex hull is a training-network geometry, not an aquifer boundary. Predictions are described, not removed.

Usage

ps_prediction_support(
  points,
  surface = NULL,
  template = NULL,
  mask = NULL,
  max_distance = NULL,
  allow_geographic = FALSE
)

Arguments

points

Training observations as a point SpatVector, sf object, or object readable by terra::vect().

surface

Optional one-layer prediction SpatRaster.

template

Optional one-layer template when surface is not supplied.

mask

Optional polygon mask used to classify cells.

max_distance

Optional positive distance threshold in projected map units.

allow_geographic

Allow longitude/latitude calculations with a classed warning. The default is FALSE.

Value

A potentiomap_support list containing rasters, a reason-code lookup table, a cell-level records table, summary, the standardized training points, and call. Stable support classes are supported, outside_training_hull, beyond_maximum_distance, outside_mask, prediction_unavailable, and multiple_limitations.

Details

Distance calculations require projected coordinates by default. An explicit override reports that longitude/latitude degrees are not linear ground units.

Examples

data("synthetic_wells")
pts <- ps_make_points(synthetic_wells, "x", "y", "gw_elevation",
                      "well_id", "EPSG:26916")
surface <- ps_interpolate(pts, methods = "IDW", grid_res = 200)$IDW
support <- ps_prediction_support(pts, surface = surface,
                                 max_distance = 1000)
support$summary
#>             support_class cells    percent
#> 1               supported   173 59.8615917
#> 2   outside_training_hull   114 39.4463668
#> 3 beyond_maximum_distance     0  0.0000000
#> 4            outside_mask     0  0.0000000
#> 5  prediction_unavailable     0  0.0000000
#> 6    multiple_limitations     2  0.6920415