AISanalyze: User guide

Introduction

AISanalyze provides a workflow to analyse Automatic Identification System (AIS) data, including:

estimating vessel travel distance, time and speed;
correcting GPS errors and delays;
identifying AIS stations and aircraft;
interpolating vessel positions;
extracting vessels around target locations;
estimating vessel characteristics.

This vignette illustrates a typical workflow.

Example data

data("ais")
data("point_to_extract")

Convert timestamps to Unix time.

ais$timestamp <- as.numeric(lubridate::ymd_hms(ais$datetime))
point_to_extract$timestamp <- as.numeric(lubridate::ymd_hm(point_to_extract$datetime))

Estimate travelled distance and speed

ais <- AIStravel(ais_data = ais)

Three variables are added:

distance_travelled
time_travelled
speed_kmh

Identify stations and aircraft

ais <- AISidentify_stations_aircraft(ais_data = ais)
#> Stations and aircraft are identified from speed, distance and time only. Other criteria (e.g. MMSIs with fewer than 9 digits) are not considered.

Two logical variables are added:

station
high_speed

Correct GPS errors

ais <- AIScorrect_speed(ais_data = ais)
#> For consecutive GPS errors, only the first point is removed to avoid overcorrection.
#> High-speed craft are not corrected.

This step corrects unrealistic speeds caused by GPS errors or transmission delays.

Interpolate vessel positions

The example below interpolates vessel positions every 60 seconds.

ais_interpolated_60sec <- AISinterpolate(
  ais_data = ais,
  type_interpolation = "maximum_time_interval",
  maximum_gap_seconds = 60
)

Alternatively, interpolation can be performed at exact timestamps. Target locations and a search radius (m) can be specified to limit interpolation to the area of interest and reduce computation time. The datetime column can then be updated from the new timestamp.

ais_interpolated_exact_timestamps <- AISinterpolate(
  ais_data = ais,
  type_interpolation = "exact_timestamp",
  exact_timestamp = list(
    timestamp_to_interpolate = point_to_extract$timestamp,
    locations_of_interest = point_to_extract[c("lon", "lat")],
    radius = 200000
  )
)

The datetime column can be updated from the interpolated timestamps:

ais_interpolated_60sec$datetime <- lubridate::as_datetime(ais_interpolated_60sec$timestamp)

ais_interpolated_exact_timestamps$datetime <- lubridate::as_datetime(ais_interpolated_exact_timestamps$timestamp)

Extract nearby vessels

Extract all vessel positions within 50 km and ±5 minutes of the target locations and timestamps (point_to_extract).

AISextract(
  ais_data = ais_interpolated_60sec,
  data = point_to_extract,
  return_all_vessel_locations = TRUE,
  search_into_radius_m = 50000,
  interval_time_before = 300,
  interval_time_after = 300
)

Set return_all_vessel_locations = FALSE to return only the vessel position at the target timestamps:

AISextract(
  ais_data = ais_interpolated_exact_timestamps,
  data = point_to_extract,
  return_all_vessel_locations = FALSE,
  search_into_radius_m = 50000,
  interval_time_before = 300,
  interval_time_after = 300
)

Alternatively, you can pass the centroids of a square grid to data and set search_shape = "square" to extract vessel positions within square grid cells:

AISextract(
  ais_data = ais_interpolated_exact_timestamps,
  data = point_to_extract,
  return_all_vessel_locations = FALSE, # or TRUE
  search_into_radius_m = 50000,
  search_shape = "square",
  interval_time_before = 300,
  interval_time_after = 300
)

Estimate vessel characteristics

infos <- AISinfos(ais)

summary_values <- infos$summary
estimated_values <- infos$estimated_values

This function estimates the most likely vessel characteristics for each MMSI, including ship type, dimensions, draught, IMO number, and name. summary_values summarises all values found in the AIS data, whereas estimated_values contains the estimated characteristic for each vessel.

Workflow summary

The recommended workflow is:

AIS data
│
▼
AIStravel()
│
▼
AISidentify_stations_aircraft() (optional)
│
▼
AIScorrect_speed() (optional)
│
▼
AISinterpolate()   (optional)
│
▼
AISextract()
│
▼
AISinfos()         (optional)

Rémi Pigeault