Make a regular prediction grid from coordinate vectors

Creates an equally spaced grid covering the range of the supplied coordinates. Works with any coordinate system (UTM metres, UTM kilometres, lon/lat degrees, etc.) - resolution and the coordinates must simply be in the same units. The grid origin is snapped down to the nearest resolution multiple below each coordinate minimum. Optionally repeats the spatial grid for every element of time, adding a year column. Optionally removes grid nodes that are farther than max_dist from any observation (requires the RANN package), and fills gaps in the remaining footprint with fill_gaps.

Usage

make_survey_grid(
  x,
  y,
  resolution,
  max_dist = NULL,
  time = NULL,
  fill_gaps = NULL
)

Arguments

x

Numeric vector of X coordinates (any units).

y

Numeric vector of Y coordinates (same units as x).

resolution

Step size between grid nodes, in the same units as x and y.

max_dist

Optional distance threshold in the same units as x and y. Grid nodes whose nearest observation is farther than max_dist are dropped. Uses a k-d tree via RANN for efficiency.

time

Optional vector of time values (e.g. 1990:2000). When supplied the spatial grid is crossed with time and a year column is added.

fill_gaps

Optional distance in the same units as x and y, used together with max_dist. Holes and bays in the footprint that are narrower than about 2 * fill_gaps are filled, without extending the outer edge of the footprint. See Details.

Value

A data.frame with columns X, Y, and (if time is supplied) year.

Details

With max_dist alone, the grid keeps the nodes within max_dist of an observation. Where hauls are farther than 2 * max_dist apart, this leaves holes inside the survey area and notches along its edge, although the survey covers the area between them. fill_gaps closes these: the footprint is first grown by max_dist + fill_gaps around each observation and then shrunk by fill_gaps again (a morphological closing). Gaps narrower than about 2 * fill_gaps disappear, while the outer edge stays at max_dist from the outermost observations. Nodes kept by max_dist are always kept.

The closing is computed on the grid nodes, so it is accurate to about one resolution. Distances are measured in the units of the coordinates; in lon/lat degrees a degree of longitude is shorter than a degree of latitude, so project the coordinates first (e.g. with add_xy()) when distances should be equal in all directions.

See also

add_grid_support() to flag the grid nodes covered by the hauls of each year, add_bathymetry() to add depth.

Examples

trawls <- unique(dab[["HH"]][, c("lon", "lat")])
if (requireNamespace("RANN", quietly = TRUE)) {
  grid <- make_survey_grid(trawls$lon, trawls$lat, resolution = 0.1,
                           max_dist = 0.2)
  grid_filled <- make_survey_grid(trawls$lon, trawls$lat, resolution = 0.1,
                                  max_dist = 0.2, fill_gaps = 0.3)
  c(nrow(grid), nrow(grid_filled))
}
#> [1] 6856 7970