Skip to contents

Distance to the nearest shoreline

Source: R/shore.R
distance_to_shore.Rd

Distance from each point to the nearest coast, in kilometres. Static — it does not vary in time — so it is computed once for the unique locations and shared across every time step.

Usage

distance_to_shore(
  env_dat,
  resolution = c("medium", "large", "small"),
  margin = 5,
  name = "shore_dist"
)

Source

Coastlines are Natural Earth, via rnaturalearth (https://www.naturalearthdata.com/). Public domain, and the resolution argument selects between their 1:110m, 1:50m, and 1:10m physical coastline layers.

The resolution is a real choice rather than a detail. A deeply indented coast like Maine's is shortened by a coarse layer, so distances near shore come out too large. "large" is the 1:10m layer and needs rnaturalearthhires.

Arguments

env_dat

an sf POINT object with one row per location and time step, as datamatch's access functions return

resolution

Natural Earth coastline resolution: "medium", "large", or "small"

margin

degrees of padding around the data when cropping the coastline. Land just outside the study area can still be the nearest shore, so cropping tightly to the bounding box would overstate distances at the edges.

name

name for the new column

Value

env_dat with a distance-to-shore column, in km

Details

Distance from shore is a broad proxy for several things at once: depth, terrestrial nutrient input, tidal mixing, and larval retention all covary with it. That makes it a useful covariate and a poor explanation — a model leaning on it is telling you where, not why. This is the covariate the older pipeline had as dist.

Choosing a coastline

resolution selects the Natural Earth coastline detail:

  • "medium" (the default, 1:50m) suits shelf-scale work.

  • "large" (1:10m) resolves individual islands and inlets, which matters in a place like the Gulf of Maine where the coast is deeply indented, but is slower and needs rnaturalearthhires.

  • "small" (1:110m) is too coarse for coastal work; it smooths bays away entirely.

Distances are computed against true geodesic distance on the ellipsoid rather than a projected approximation, so they stay correct across a wide domain.

Examples

if (FALSE) { # \dontrun{
env <- distance_to_shore(env)
# Finer coastline, for a deeply indented shore:
env <- distance_to_shore(env, resolution = "large")
} # }