Calculate distance between point and nearest raster cell edge in R

I'm trying to calculate the distance in meters between a point feature and the edge of the nearest 30 x 30 meter raster cell. The data I have are aggregated by quarters for five years (2016-2020), so I need to do this calculation iteratively for each quarter within each year. The value in the cell does not matter, I just need the vector distance between each point and it's closest raster cell for a given quarter and year.

Sample data

library(raster)

# set the extent of the study area
ext <- extent(401000, 402000, 4405000, 4406000)

# create a point file with drop locations
set.seed(123) # set a random seed for reproducibility
years <- c(2016, 2017, 2018, 2019, 2020)
quarters <- c("Q1", "Q2", "Q3", "Q4")
coords <- expand.grid(year = years, quarter = quarters, 
                      x = runif(4, 401000, 402000), 
                      y = runif(4, 4405000, 4406000))
points <- st_as_sf(coords, coords = c("x", "y"), crs = st_crs(3310))
points$year <- as.factor(points$year)
points$quarter <- as.factor(points$quarter)

# create a raster file with biomass estimates
raster <- raster(ext, res = 30)
values(raster) <- NA
non_empty_cells <- sum(runif(ncell(raster)) > 0.98)
if (non_empty_cells > 0) {
  values(raster)[sample(which(is.na(values(raster))), non_empty_cells)] <- runif(non_empty_cells, 0, 10)
}

# plot the data
plot(raster, col = gray.colors(10, start = 0.8, end = 0))
plot(points, add = TRUE, col="red")

I am assuming that you want the distance to the border of the nearest cell with a value (that is not NA). If your raster is not too large you could do:

Example data

library(terra)
set.seed(123)
raster <- rast(ext=c(401000, 402000, 4405000, 4406000), res = 30, crs="local")
cells <- sample(ncell(raster), 20)
raster[cells] <- runif(length(cells))
years <- c(2016, 2017, 2018, 2019)
quarters <- c("Q1", "Q2", "Q3", "Q4")
points <- data.frame(year = years, quarter = quarters, 
          x = runif(4, 401000, 402000), y = runif(4, 4405000, 4406000))
pts <- vect(points, geom=c("x", "y"), crs=crs(raster))

Solution

lns <- as.lines(as.polygons(raster, dissolve=FALSE))
#nearest(pts, lns) |> values()
#  from_id   from_x  from_y to_id  distance
#1       1 401127.5 4405665     7 136.23261
#2       2 401753.3 4405095    19  27.12752
#3       3 401895.0 4405384    10  74.71168
#4       4 401374.5 4405274    14  57.06026

If the distance are large relatively to the resolution, you could also do

x <- distance(raster>0)
extract(x, pts)
#  ID     lyr.1
#1  1 169.70563
#2  2  42.42641
#3  3 108.16653
#4  4  84.85281

That computes the distance from cell center (that the points fall in) to cell center (of the nearest cell).