[R-sig-Geo] Spatial correlation between two 'sf' kriging objects

[Nicola Gambaro]

I have created two ‘sf’ kriging objects (point vectors), one for temperature and another for agricultural yields. To make the grid and carry out the point interpolation, I have remained within the ‘sf’ package.

I would now like to create a spatial local correlation ‘raster’ between these two variables, as shown on this webpage https://statnmap.com/2018-01-27-spatial-correlation-between-rasters/ <https://statnmap.com/2018-01-27-spatial-correlation-between-rasters/>. However, in that example, they use the ‘raster’ package and the ‘focal’ function. I was wondering if there was a way of doing this within ‘sf’, i.e. without having to change classes? If not, what is the best way to convert those objects into raster classes?

Here is an excerpt of my kriging code for reference:
library(sf)
sf_data <- st_as_sf(x = data, coords = c("longitude", "latitude"), crs = 4326)
library(gstat)
vgm_utci <- variogram(UTCI~1, sf_data)
utci_fit <- fit.variogram(vgm_utci, vgm("Gau"), fit.kappa = TRUE)
plot(vgm_utci, utci_fit)
istria <- read_sf(“./Istria_Boundary.shp")
istria <- istria$geometry
istria.grid <- istria %>% 
  st_make_grid(cellsize = 0.05, what = "centers") %>% 
  st_intersection(istria) 
library(ggplot2)
ggplot() + geom_sf(data = istria) + geom_sf(data = istria.grid)
library(stars)
utci_krig <- krige(formula = sf_data$UTCI ~ 1, locations = sf_data, 
                   newdata = istria.grid, model = utci_fit)


Thank you very much in advance,

Nicola
	[[alternative HTML version deleted]]