Problem for extract mean values between same coordinates in different csr

[Leprechault]

I have a data set extracted from S2 satellite rasters (10 x 10 meters) with 12 values (ras.df.ll), but 6 is in a tile (T21JYG) and the same 6 coordinates but in a second tile (T21JYG). I'd like to calculate the mean in the same (x,y coordinates) between tiles without success. In my example:

library(sfheaders)
library(dplyr)
library(sf)

# Raster (10x10 meters) info in data frame 1 - crs +proj=utm +zone=21 +south +datum=WGS84 +units=m +no_defs
x <-c(789385,789395,789405,789415,789425,789435)
y <-c(6626865,6626865,6626865,6626865,6626865,6626865)
tile <- rep("T21JYG",6)
values <-c(321,249,234,238,224,244)
ras.ds1<-data.frame(x,y,tile,values)
ras.ds1.sf <- st_as_sf(ras.ds1, coords = c("x", "y"), crs = 32721, agr = "constant")
ras.ds1.sf.ll <- st_transform(ras.ds1.sf, crs=4326)
ras.ds1.sf.ll


# Raster (10x10 meters) info in data frame - crs +proj=utm +zone=22 +south +datum=WGS84 +units=m +no_defs 
x <-c(213285,213295,213305,213315,213325,213335)
y <-c(6626955,6626955,6626955,6626955,6626955,6626955)
tile <- rep("T22JBM",6)
values <-c(336,355,363,426,341,308)
ras.ds2 <-data.frame(x,y,tile,values)
ras.ds2.sf <- st_as_sf(ras.ds2, coords = c("x", "y"), crs = 32722, agr = "constant")
ras.ds2.sf.ll <- st_transform(ras.ds2.sf, crs=4326)
ras.ds2.sf.ll


# Join information
ras.ds.sf.ll <- rbind(ras.ds1.sf.ll, ras.ds2.sf.ll)
ras.df.ll <- sf_to_df(ras.ds.sf.ll, fill = TRUE, unlist = NULL)

# Mean values by same coordinate 
ras.df.ll %>% 
  group_by(x,y) %>% dplyr::summarise(values=mean(values))

# # A tibble: 12 x 3
# # Groups:   x [12]
#        x     y values
#    <dbl> <dbl>  <dbl>
#  1 -54.0 -30.5    321
#  2 -54.0 -30.5    249
#  3 -54.0 -30.5    234
#  4 -54.0 -30.5    238
#  5 -54.0 -30.5    224
#  6 -54.0 -30.5    244
#  7 -54.0 -30.5    336
#  8 -54.0 -30.5    355
#  9 -54.0 -30.5    363
# 10 -54.0 -30.5    426
# 11 -54.0 -30.5    341
# 12 -54.0 -30.5    308

This output is not correct, because the group_by(x,y) does not recognise 1st row in the first tile is the same coordinate of 1st row in the second tile but originally in another CRS in UTM. Means for me a transformation problem between projections.

The correct output for the variable values is:

#  x         y      values
# -53.986	-30.456	328.5
# -53.986	-30.456	302
# -53.986	-30.456	298.5
# -53.986	-30.456	332
# -53.986	-30.456	282.5
# -53.986	-30.456	276

Please, any help using some sf solution for it?

Thanks in advance,

Alexandre

[markolipka]

Hi Alexandre,

you seem to assume, that the two point sets you define (ras.ds1 and ras.ds1) are referencing the same locations. What makes you so sure about that?