Skip to contents

Resample a move2 track to regular time intervals by geometric interpolation

Source: R/tt_regular_time.R
tt_regular_time.Rd

Resamples one or more tracks stored in a move2 object onto a regular time grid. A new grid point is placed every interval time units from the start of the track. Tracks are processed independently, so different tracks may have different temporal extents and the points from different tracks will not have matching time stamps (as the starting points are different).

Usage

tt_regular_time(x, interval, max_time_lag = NULL, snap_times = FALSE)

Arguments

x

A move2 object. Timestamps (as returned by event_time()) must be base::POSIXct.

interval

Resampling interval as a units::units object carrying time units convertible to seconds (e.g. as_units(60, "s"), as_units(1, "min"), as_units(0.5, "h")). Every track is resampled at this cadence between its own first and last observation. Passing a plain numeric value raises an error to prevent silent unit mismatches.

max_time_lag

Optional upper bound on interpolation gaps, supplied as a units::units object with time units convertible to seconds (same rules as interval). Grid points that fall strictly inside a gap between consecutive input observations that is longer than max_time_lag are silently dropped from the output. Pass NULL (the default) to interpolate across all gaps regardless of size.

snap_times

Logical (default FALSE). When TRUE, each track's grid begins at the first whole-interval boundary that is strictly after the track's first observation, rather than at the first observation itself. For example, with interval = 10 min, a track starting at 08:04 will have its first resampled point at 08:10, then 08:20, 08:30, and so on. This aligns grids across tracks that share the same epoch, so that overlapping tracks will have identical timestamps at each step.

Value

A move2 object on a regular time grid. The CRS, time-column name, track-id column name, and track-level attributes of x are all preserved.

Spatial interpolation

Positions are linearly interpolated along the observed path, with a strategy dependent on the map projection (using a strategy similar to move2::mt_interpolate()):

  • No CRS – sf::st_line_sample() is called on the Euclidean linestring connecting the observations.

  • Geographic CRS (lon/lat) – the path is treated as a spherical polyline and s2::s2_interpolate_normalized() is used, which follows great-circle arcs between consecutive points.

  • Projected CRS – the track is temporarily transformed to WGS 84 (EPSG:4326) for spherical interpolation via s2::s2_interpolate_normalized(), then the resulting points are transformed back to the original CRS.

Temporal-to-spatial mapping

Raw observations are often unevenly spaced in both time and space. tt_regular_time accounts for this by first computing the normalised arc-length fraction of every input point along the path (using s2::s2_distance() for geographic CRS and Euclidean distance otherwise), then using stats::approx() to linearly interpolate those fractions at each new time step. A stationary stretch of the track (many observations covering little distance) is therefore correctly treated as slow movement, not as a spatial shortcut.

Attribute interpolation

Numeric event-level columns are linearly interpolated at the new time steps using stats::approx(). Non-numeric columns receive the value of the nearest preceding observation via findInterval. Track-level attributes stored in mt_track_data are preserved unchanged.

See also

mt_interpolate for the move2 implementation of the same spatial strategy with flexible time targets (including interpolating to specific missing timestamps); units::as_units() for constructing the required units objects; sf::st_line_sample() for Euclidean path sampling; s2::s2_interpolate_normalized() for spherical arc sampling.

Examples

library(sf)

# Build a simple three-point track with irregular time gaps
times <- as.POSIXct("2024-01-01 00:00:00", tz = "UTC") + c(5, 90, 200)
geom <- st_sfc(
  st_point(c(-0.10, 51.50)),
  st_point(c(-0.05, 51.52)),
  st_point(c(0.00, 51.54)),
  crs = 4326
)
track <- move2::mt_as_move2(
  sf::st_sf(
    timestamp = times,
    speed_ms  = c(2.1, 3.4, 1.8),
    track_id  = "gull_01",
    geometry  = geom
  ),
  time_column = "timestamp",
  track_id_column = "track_id"
)

# Resample to one fix per minute
resampled <- tt_regular_time(track, interval = as_units(1, "min"))
resampled
#> A <move2> with `track_id_column` "track_id" and `time_column` "timestamp"
#> Containing 1 track lasting 3 mins in a
#> Simple feature collection with 4 features and 3 fields
#> Geometry type: POINT
#> Dimension:     XY
#> Bounding box:  xmin: -0.1 ymin: 51.5 xmax: -0.006820769 ymax: 51.53727
#> Geodetic CRS:  WGS 84
#>             timestamp track_id speed_ms                      geometry
#> 1 2024-01-01 00:00:05  gull_01 2.100000             POINT (-0.1 51.5)
#> 2 2024-01-01 00:01:05  gull_01 3.017647  POINT (-0.06471044 51.51412)
#> 3 2024-01-01 00:02:05  gull_01 2.890909  POINT (-0.03409567 51.52637)
#> 4 2024-01-01 00:03:05  gull_01 2.018182 POINT (-0.006820769 51.53727)
#> Track features:
#>   track_id
#> 1  gull_01

# Resample to 30-second fixes, skipping gaps > 2 minutes
resampled_gapped <- tt_regular_time(
  track,
  interval     = as_units(30, "s"),
  max_time_lag = as_units(2, "min")
)
resampled_gapped
#> A <move2> with `track_id_column` "track_id" and `time_column` "timestamp"
#> Containing 1 track lasting 3 mins in a
#> Simple feature collection with 7 features and 3 fields
#> Geometry type: POINT
#> Dimension:     XY
#> Bounding box:  xmin: -0.1 ymin: 51.5 xmax: -0.006820769 ymax: 51.53727
#> Geodetic CRS:  WGS 84
#>             timestamp track_id speed_ms                      geometry
#> 1 2024-01-01 00:00:05  gull_01 2.100000             POINT (-0.1 51.5)
#> 2 2024-01-01 00:00:35  gull_01 2.558824  POINT (-0.08235795 51.50706)
#> 3 2024-01-01 00:01:05  gull_01 3.017647  POINT (-0.06471044 51.51412)
#> 4 2024-01-01 00:01:35  gull_01 3.327273  POINT (-0.04772823 51.52091)
#> 5 2024-01-01 00:02:05  gull_01 2.890909  POINT (-0.03409567 51.52637)
#> 6 2024-01-01 00:02:35  gull_01 2.454545  POINT (-0.02045986 51.53182)
#> 7 2024-01-01 00:03:05  gull_01 2.018182 POINT (-0.006820769 51.53727)
#> Track features:
#>   track_id
#> 1  gull_01

# Resample to one fix per minute, snapping times to whole-minute boundaries
resampled_snap <- tt_regular_time(track,
  interval = as_units(1, "min"),
  snap_times = TRUE
)
resampled_snap
#> A <move2> with `track_id_column` "track_id" and `time_column` "timestamp"
#> Containing 1 track lasting 2 mins in a
#> Simple feature collection with 3 features and 3 fields
#> Geometry type: POINT
#> Dimension:     XY
#> Bounding box:  xmin: -0.06765207 ymin: 51.51294 xmax: -0.009094177 ymax: 51.53637
#> Geodetic CRS:  WGS 84
#>             timestamp track_id speed_ms                      geometry
#> 1 2024-01-01 00:01:00  gull_01 2.941176  POINT (-0.06765207 51.51294)
#> 2 2024-01-01 00:02:00  gull_01 2.963636  POINT (-0.03636799 51.52546)
#> 3 2024-01-01 00:03:00  gull_01 2.090909 POINT (-0.009094177 51.53637)
#> Track features:
#>   track_id
#> 1  gull_01