Mapping acceleration data

In this vignette we will show how you can create an interactive map of an animal track where acceleration measurement at linked to the positions where the animal has been recorded.

Lets first get example data, here we download the tracks of two albatrosses.

library(move2Acc)
require(dplyr, quietly = TRUE)
#> 
#> Attaching package: 'dplyr'
#> The following objects are masked from 'package:stats':
#> 
#>     filter, lag
#> The following objects are masked from 'package:base':
#> 
#>     intersect, setdiff, setequal, union
require(move2, quietly = TRUE)
require(leaflet, quietly = TRUE)
albatross_data <- movebank_download_study(
  "Galapagos Albatrosses",
  individual_local_identifier = c("4264-84830852", "4266-84831108"),
  sensor_type_id = c("acceleration", "gps")
)

Next we create a acceleration column in the move2 object

albatross_data <- albatross_data %>%
  mutate(acceleration = as_acc(.))

In the move2 object accelerations are not connected to the location observations. Therefore we need to match the acceleration measurements to the location records. For these eobs tags we see that the acceleration measurements are recorded after the location has been recorded. Here we can use that with dplyr::left_join() to find the associated records. An alternative would be to do a linear interpolation of the location to the time of the acceleration measurement using move2::mt_interpolate().

merged_data <- albatross_data |>
  filter(!sf::st_is_empty(geometry)) |>
  select(-acceleration) |>
  left_join(
    albatross_data |>
      select(eobs_start_timestamp, acceleration) |>
      filter(!is.na(acceleration)) |>
      sf::st_drop_geometry(),
    join_by(
      closest(eobs_start_timestamp < eobs_start_timestamp),
      individual_local_identifier == individual_local_identifier
    ),
    suffix = c(".gps", ".acceleration")
  )
hist(
  as.numeric(merged_data$timestamp.acceleration - merged_data$timestamp.gps,
    units = "secs"
  ),
  main = "Time difference between gps acceleration",
  xlab = "Time difference [seconds]"
)

In this histogram we see that there is generally only a few seconds between the acceleration measurements and the gps record. Only on rare occasions it is more then 10 seconds.

Now we can use the merged data to plot a map:

pal <- colorFactor("Set1", mt_track_id(albatross_data))
accRange <- quantile(
  unlist(vctrs::field(merged_data$acceleration, "bursts")),
  probs = c(.0001, .9999)
)
leaflet() %>%
  addTiles() %>%
  addPolylines(
    data = mt_track_lines(albatross_data |>
      filter(!sf::st_is_empty(geometry))),
    color = ~ pal(individual_local_identifier)
  ) %>%
  addCircleMarkers(
    data = merged_data, 
    color = ~ pal(individual_local_identifier),
    popup = leafpop::popupGraph(
      mapply(
        function(x, range) {
          dygraphs::dyAxis(x, name = "y", valueRange = unname(range))
        },
        mapply(plot_time, merged_data$acceleration,
          merged_data$timestamp.acceleration,
          SIMPLIFY = FALSE
        ),
        MoreArgs = list(range = accRange),
        SIMPLIFY = FALSE
      ),
      "html", height = 350, width = 600
    )
  ) %>%
  addLegend(
    pal = pal,
    values = mt_track_id(albatross_data)
  )