3D footprint gridding (lon, lat, time) — obs_grid

Bins particles into a 3D grid: longitude, latitude, and time. Supports two methods: "simple" (bin-and-sum) and "kernel" (Gaussian smoothing).

Usage

obs_grid_cube(
  x,
  lon_min,
  lat_min,
  lon_max,
  lat_max,
  res = 0.1,
  nt = 240L,
  t0 = 0,
  dt = 60,
  method = c("simple", "kernel"),
  bandwidth = res,
  use_haversine = FALSE,
  n_threads = 1L,
  npar = 1L
)

Arguments

x: A data.table or data.frame with columns lat, lon, time (minutes), and foot.
lon_min, lat_min, lon_max, lat_max: Grid extent (degrees).
res: Spatial resolution (degrees).
nt: Number of time layers in the output cube.
t0: Reference time for the first layer (minutes). Default 0.
dt: Time step per layer (minutes). Default 60 (1 hour).
method: Character. Either "simple" for bin-and-sum or "kernel" for Gaussian smoothing. Default "simple".
bandwidth: Kernel standard deviation for method = "kernel". In degrees if use_haversine = FALSE, or metres if TRUE. Default equals res.
use_haversine: Logical. If TRUE, uses great-circle distances and bandwidth in metres for kernel smoothing.
n_threads: Number of OpenMP threads.
npar: Normalization factor (total particles released).

Value

A named list:

grid: Numeric 3D array of dimension [ny x nx x nt].
lon, lat: Cell-centre coordinates.
time: Start time of each bin (minutes back from t0).