CODEWITHSUNDEEP
rdistancegeotelemetry
Klervi
Klervi

Reputation: 11

How to calculate distance with telemetry data

I have telemetry data with a large number of detections for each individual (ID) at several stations. I want to calculate the total distance travelled by each ID in my study.

Each station is associated with its coordinates (Longitude and Latitude), which I convert to UTM data.

I have created a data that looks like mine, with the same data type (data_detections).

library(lubridate) # ymd_hms()
library(sp)
#generate a random data
generate_random_datetime <- function(start_date, end_date, n) {
  seq(start_date, end_date, by = "min")[sample(1:(as.integer(difftime(end_date, start_date, units = "mins")) + 1), n)]
}
set.seed(123)  
n <- 100  

data_detections <- data.frame(
  Date.and.Time..UTC. = generate_random_datetime(ymd_hms("2024-01-01 00:00:00"), ymd_hms("2024-01-20 23:59:59"), n),
  Receiver = sample(1:10, n, replace = TRUE),
  Latitude = runif(n, 52.0, 53.0),
  Longitude = runif(n, 3.0, 4.0) ,
  ID = as.character(sample(1:10, n, replace = TRUE))
)

I calculated my total distances like that, but the values are not correct when i compare it with my "real" values

#convert to UTM
coord <- SpatialPoints(data_detections[, c("Longitude", "Latitude")], 
                       proj4string = CRS("+proj=longlat +datum=WGS84"))
coord.t <- spTransform(coord, CRS("+proj=utm +datum=WGS84 +zone=43"))
data_detections[, c("Longitude_UTM", "Latitude_UTM")] <- coordinates(coord.t)

head(data_detections)

#caclulate the total distance

total_distance <- function(data) {
  coords <- as.matrix(data[, c("Longitude_UTM", "Latitude_UTM")])
  distances <- sqrt(rowSums((coords[-1, ] - coords[-nrow(coords), ])^2))
  distance <- sum(distances, na.rm = TRUE)
  return(data.frame(distance = distance))
}


total_distances <- data_detections %>%
  group_by(ID) %>%
  group_modify(~ total_distance(.x))

print(total_distances)

Upvotes: 0

Views: 101

Answers (1)

margusl
margusl

Reputation: 17554

If you just need to add up distances between grouped and ordered(*) geographic coordinates, one option is to use geosphere::distGeo(), if you pass it a single 2 column data.frame or matrix, it returns a vector of sequential distances between points on an ellipsoid (default is WGS84), in meters.

* - in your example data you have created a random timestamp sequence, without sorting it first, it would simulate a random point order in your telemetry data, i.e. random distances with random sums.

library(dplyr)
library(lubridate)
library(geosphere)

dist_geosphere <- 
  data_detections |> 
  group_by(ID) |> 
  arrange(Date.and.Time..UTC., .by_group = TRUE) |> 
  mutate(dist_geo  = distGeo(pick(c("Longitude", "Latitude")))) |> 
  summarise(dist_sum_m = sum(dist_geo, na.rm = TRUE))

dist_geosphere
#> # A tibble: 10 × 2
#>    ID    dist_sum_m
#>    <chr>      <dbl>
#>  1 1        361722.
#>  2 10       556821.
#>  3 2        347572.
#>  4 3        293461.
#>  5 4        470681.
#>  6 5        303046.
#>  7 6        333351.
#>  8 7        556807.
#>  9 8        716994.
#> 10 9        408434.

Example data:

#generate a random data
generate_random_datetime <- function(start_date, end_date, n) {
  seq(start_date, end_date, by = "min")[sample(1:(as.integer(difftime(end_date, start_date, units = "mins")) + 1), n)]
}
set.seed(123)  
n <- 100  

data_detections <- data.frame(
  Date.and.Time..UTC. = generate_random_datetime(ymd_hms("2024-01-01 00:00:00"), ymd_hms("2024-01-20 23:59:59"), n),
  Receiver = sample(1:10, n, replace = TRUE),
  Latitude = runif(n, 52.0, 53.0),
  Longitude = runif(n, 3.0, 4.0) ,
  ID = as.character(sample(1:10, n, replace = TRUE))
)

Upvotes: 0

Related Questions