Animal movement

move.aeqd_crs#

An azimuthal equidistant projection centred on the tracks (move2's mt_aeqd_crs): distances and directions from the centre are true in it, a local metric projection for tracks anywhere; and the tracks in it. The centre: the centroid of the locations (on the sphere for longitude / latitude), the middle of their longitude and latitude ranges (center), or [longitude, latitude]. move2 takes the centroid to EPSG:4236 (Hu Tzu Shan) where WGS84 is meant, about a kilometre off; here it stays in WGS84, as landschaft's.

move.as_tracks#

Tracks from points or a table of locations (move2's mt_as_move2, landschaft's as_tracks): ordered by track and time (stable), the tracks in the order they first appear, as every move node needs. The time is a column of datetimes as text (ISO 8601, UTC) or of numbers; the track one of names or numbers. A table's x and y columns become points in the coordinate system given.

move.azimuth#

The direction of each step, clockwise from north (move2's mt_azimuth): the geodesic azimuth on the coordinate system's ellipsoid in longitude / latitude (lwgeom's st_geod_azimuth, GeographicLib, the same numbers), in (-pi, pi]; the direction on the grid, atan2(dx, dy), in a projected system or without one. NaN at the last location of each track and where a location repeats the previous.

move.distance#

The distance from each location to the next of its track (move2's mt_distance, the same numbers): on the sphere of S2 (radius 6 371 010 m, as sf::st_distance with s2) in longitude / latitude, the straight line in the coordinate system's units otherwise. NaN at the last location of each track.

move.distribution#

A distribution of step lengths or turn angles (amt's make_distribution): gamma (shape, and scale or rate), exp (rate), lnorm (meanlog, sdlog), hnorm (sd) for step lengths; vonmises (kappa, mu 0) or unif (min, max) for turn angles. The parameters not given take amt's defaults (gamma 1, 1; exp 1; lnorm 0, 1; hnorm 1; unif -pi, pi; von Mises 1, 0).

move.extract_covariates#

The values of rasters at locations, or at the ends or starts of steps (amt's extract_covariates): the value of the cell holding each point (terra's extract: a point on the edge between two cells takes the cell to its right / below), none outside the raster or on no data. Connect the rasters as inputs of any names (forest, elev...): a column each, named by its input (a stack's layers input_layer). The points are taken to each raster's coordinate system.

Takes any other inputs by name (any).

move.filter_min_n_burst#

The bursts (move.resample) of at least min_n locations (amt's filter_min_n_burst).

move.filter_movebank_visible#

The tracks without the locations Movebank hides (move2's mt_filter_movebank_visible): those of the column visible when there is one; otherwise the records not marked as outliers (manually, by an algorithm or on import), and those marked valid by hand.

move.filter_per_interval#

One location per time interval of each track (move2's mt_filter_per_interval): the intervals of lubridate's floor_date, multiples counted from the start of the next larger unit (minutes within the hour, hours within the day, days within the month), weeks from Sunday; the first, the last or one at random (R's sample.int after set.seed(seed), move2's picks). The times must be datetimes.

move.filter_unique#

The tracks without duplicated times within a track (move2's mt_filter_unique and mt_unique). criterion subsets (default): of the records of a track with the same time, those that only repeat another or hold less (each value equal to the other's or missing) are dropped, with a warning when some remain; subsets_equal the same with numbers equal within R's all.equal (1.5e-8); first, last: the first or last record of each time; sample: one at random (R's sample.int after set.seed(seed), move2's picks).

move.fit_distribution#

A distribution fitted to step lengths or turn angles (amt's fit_distr, the same numbers): gamma by maximum likelihood with R's Nelder-Mead from the closed-form estimates (fitdistrplus' fitdist; steps of length 0 take the smallest other length); exp the maximum likelihood rate 1 / mean(x) (amt fits it by BFGS, which fails for rates under 0.001); lnorm by Nelder-Mead; hnorm sqrt(mean(x^2)); unif the smallest and largest value; vonmises kappa by maximum likelihood (circular's mle.vonmises), mu 0. The missing values are left out.

move.interpolate#

Locations at given times between the recorded ones (move2's mt_interpolate). Without times, the records without a location (empty points) get one between the locations before and after them. With times (a list, or a unit such as 1 hour or 15 mins for a regular sequence from the first to the last time), a record is added for each time within the span of each track that the track does not have, and located. A location is placed on the great circle between its neighbours in proportion to the time (in longitude / latitude, and through it in a projected system, as move2 does with s2), on the straight line without a coordinate system. Gaps longer than max_time_lag are left empty. Ordered by track and time.

move.nsd#

The net squared displacement: the squared distance of each location from the first of its track (amt's nsd), in the units of the coordinates (squared degrees in longitude / latitude, as amt).

move.random_numbers#

Draws of a distribution of steps (amt's random_numbers), from R's generator after set.seed(seed): amt's numbers. Turn angles of a von Mises distribution are wrapped to (-pi, pi].

move.random_steps#

Each observed step with n_control random steps from its start (amt's random_steps, the same steps): the data of a step selection function. For each step but the first of a track (or of a burst), the observed step (case_ 1) and n_control steps from the same start (case_ 0) whose lengths and turn angles are drawn from sl_distr and ta_distr (default: a gamma and a von Mises distribution fitted to each track's steps), turned from the direction of the previous step; step_id_ names the stratum. As amt, n_draws numbers are drawn from each distribution once per track and each random step takes one of them at random. Bursts of fewer than 3 steps are left out. R's generator after set.seed(seed), the tracks in turn. The steps must be in a projected coordinate system (move.aeqd_crs).

move.read_tracks#

Animal tracks from a Movebank file (move2's mt_read, landschaft's load_tracks): the CSV files Movebank exports (also .csv.gz) and the zip files of Env-DATA (the CSV inside), as points in EPSG:4326, one per record. The time is timestamp (UTC), the track the individual (individual-local-identifier; when an individual carried several tags, the individual and the tag, in a new individual-tag-local-identifier column, as move2 does); the outlier flags are numbers 1 and 0. Records without coordinates (other sensors) keep an empty location. Ordered by track and time (stable), as every move node needs.

move.resample#

Locations at a regular rate, in bursts (amt's track_resample, the same locations and bursts): from the location start of each track, the next location kept is the first at least rate - tolerance later, in the same burst when it is at most rate + tolerance later, else it starts a new burst; the locations in between are dropped. Each location kept gets its burst (1, 2... within its track) in the column burst_, which move.steps and move.random_steps follow.

move.segments#

The step from each location to the next of its track as a line (move2's mt_segments): one feature per location with its attributes, a two-point line per step and the location itself (a point) at the last location of each track.

move.simulate_brownian#

Tracks of Brownian motion (move2's mt_sim_brownian_motion): each step moves by normal draws of standard deviation sqrt(sigma^2 dt) along x and along y, from the start. R's generator after set.seed(seed): the same tracks as move2 (the tracks one after the other, x before y). Columns time and track, points without a coordinate system.

move.speed#

The speed of each step: the distance over the time lag (move2's mt_speed). Locations of a track must not share their time (move.filter_unique drops them). NaN at the last location of each track.

move.steps#

The steps between consecutive locations of each track (amt's steps and steps_by_burst, the same numbers), of each burst when the tracks have move.resample's burst_: a line per step, with its track, burst_, start and end (x1_, x2_, y1_, y2_), length sl_, direction direction_p (anticlockwise from east, in (-pi, pi]), turn angle ta_ from the previous step (NaN at the first step of each track or burst and after a step of length 0), the times t1_, t2_ and dt_ (seconds). In longitude / latitude the lengths are great-circle distances in metres (sf::st_distance) and the directions geodesic bearings turned into angles from east: amt reads the bearing, clockwise from north, as if it were that angle; here it is converted, pi / 2 - bearing.

move.time_lags#

The time from each location to the next of its track (move2's mt_time_lags), exact between datetimes (differences of nanoseconds). NaN at the last location of each track.

move.time_of_day#

Day or night at each location, or at the end or start of each step (amt's time_of_day): sunrise and sunset of the location's date (UTC) as suncalc computes them (the sun's upper edge at -0.833 degrees), and with include_crepuscule dawn and dusk (the sun at -6 degrees, civil twilight). A column of day, night (and dawn, dusk); empty where the sun does not rise or set that day. suncalc takes the noon of the machine's time zone; here, the noon of UTC.

move.track_data#

One row per track (in the order they appear) with the columns whose value does not change within a track (move2's mt_track_data): the animal's, the tag's, the study's... as features without a geometry.

move.track_lines#

Each track as a line through its locations in time order (move2's mt_track_lines): one feature per track, its id, the number of locations (n_locations) and the columns whose value does not change within a track (move.track_data's). Empty locations are left out; a track with a single location gets no line (an empty geometry).

move.track_summary#

amt's summaries of each track, a line per track (move.track_lines's, with n_locations and the track's data): cum_dist (the sum of its steps, metres in longitude / latitude), tot_dist (from the first to the last location), straightness (tot_dist / cum_dist), msd (the mean squared distance of the locations from their centre), sinuosity (Benhamou 2004, the turn angles in radians: amt takes them as degrees; NaN under two steps) and tac (the turn angle correlation, Dray et al. 2010).

move.turn_angle#

The change of direction at each location, in [-pi, pi) (move2's mt_turnangle): the azimuth of the step leaving it minus that of the step arriving, positive to the right. NaN at the first and last location of each track.