Mutate when function output has data in two columns (geosphere) - r

Question

I have animal survey data from transects. Transects are divided into sections. There are lat/lon data for the start/endpoints of some sections but not others, and I want to calculate the start/endpoints for sections where these values are missing. Missing start/endpoints should be calculated using the section bearing (degrees), section length (m).
Example data:
Section
StartLon
StartLat
EndLon
EndLat
Bearing
Length
1
-132.4053
53.00704
-132.4053
53.00714
360
5
2
-132.4053
53.00714
NA
NA
360
10
I'm trying to use destPoint (geosphere) to calculate the missing start/endpoints (NAs). The output of destPoint looks like:
lon lat
[1,] -132.4053 53.00701
My code:
data %>%
mutate(EndLon = if_else(is.na(EndLon), destPoint(c(StartLon, StartLat), Bearing, Length), EndLon))
data %>%
mutate(EndLat = if_else(is.na(EndLat), destPoint(c(StartLon, StartLat), Bearing, Length), EndLat))
My code gives this error:
Error: Problem with `mutate()` input `test`.
x Wrong length for a vector, should be 2
i Input `test` is `if_else(...)`
I think the error is because the output of destPoint is two values (lon and lat), and the mutated column can only hold one value. Maybe there's a way to use select() so that only lon or lat goes into the mutated column?
Hoping for a dplyr solution.

Answer 1

We may use rowwise
library(dplyr)
library(geosphere)
data %>%
rowwise %>%
mutate(EndLon = if(is.na(EndLon))
destPoint(c(StartLon, StartLat), Bearing, Length)[, 'lon'] else EndLon) %>%
ungroup
-output
# A tibble: 2 x 7
# Section StartLon StartLat EndLon EndLat Bearing Length
# <int> <dbl> <dbl> <dbl> <dbl> <int> <int>
#1 1 -132. 53.0 -132. 53.0 360 5
#2 2 -132. 53.0 -132. NA 360 10
data
data <- structure(list(Section = 1:2, StartLon = c(-132.4053, -132.4053
), StartLat = c(53.00704, 53.00714), EndLon = c(-132.4053, NA
), EndLat = c(53.00714, NA), Bearing = c(360L, 360L), Length = c(5L,
10L)), class = "data.frame", row.names = c(NA, -2L))
The issue would be that c(StartLon, StartLat) would concatenate the whole column values from both of those column, and thereby the length of one of the arguments for if_else becomes different in length than the rest. If we do the rowwise, it is grouped by row and we can use if/else (which requires a input logical expression of length 1)