I have a dataset that looks like this:
site lat long
bras2 41.21 -115.11
tex4 45.3 -112.31
bras2 41.15 -115.15
bras2 41.12 -115.19
For samples with the same site
name, I want to calculate their centre point and then add it as a column to the dataset. Some site
names are duplicated twice, other three times, other four times.
Like this:
site lat long centre_lat centre_long
bras2 41.21 -115.11 value here value here
tex4 45.3 -112.31 45.3 -112.31
bras2 41.15 -115.15 value here value here
bras2 41.12 -115.19 value here value here
How can I do this?
If you're using spatial data, you should look into using the sf
package. It handles geometries and functions for operating on them well.
Code below shows using both sf::st_centroid
and geosphere::centroid
. I prefer sf
's way of doing things.
df <- read.table(header=TRUE, text= "site lat long
bras2 41.21 -115.11
tex4 45.3 -112.31
bras2 41.15 -115.15
bras2 41.12 -115.19")
library(dplyr)
library(geosphere)
library(sf)
# Using sf's st_centroid
df_sf <- st_as_sf(df, coords = c('long', 'lat'))
centroids_sf <- df_sf %>%
group_by(site) %>%
summarize(geometry = st_union(geometry)) %>%
st_centroid
# Using geosphere::centroid
centroids_geoshpere <- df_sf %>%
group_by(site) %>%
filter(n() >2) %>% ## geosphere needs polygons therefore 3+ points
st_union() %>%
st_cast('POLYGON') %>%
as('Spatial') %>% # geoshpere expects SpatialPolygons objects
centroid()
centroids_geoshpere
#> [,1] [,2]
#> [1,] -115.15 41.16001
centroids_sf
#> Simple feature collection with 2 features and 1 field
#> geometry type: POINT
#> dimension: XY
#> bbox: xmin: -115.15 ymin: 41.16 xmax: -112.31 ymax: 45.3
#> CRS: NA
#> # A tibble: 2 x 2
#> site geometry
#> * <chr> <POINT>
#> 1 bras2 (-115.15 41.16)
#> 2 tex4 (-112.31 45.3)
Looks like thery're close enough to the same point. I don't think geosphere::centroid
can give a centroid for a single point, but may be wrong. sf::st_centroid
has no problem with 1,2, or more points.
Created on 2020-12-20 by the reprex package (v0.3.0)