rrasterr-rastersatellite-image

Sentinel-1 & 2: create coincident points in images of different tiles

I have a Sentinel-1 image and Sentinel-2 images and I want to make the pixels in Sentinel-1 coincident to Sentinel-2. The more simple solution is put both images in the same geometry (spatial extent and resolution) using S2 <- resample(S2, S1), but I'm worry about the spectral information integrity. Is better resample the Sentinel-1 or Sentinel-2 and bilinear is better for radar(S1) or nearest neighbor?

In my example:

# Packages
library(raster)

# Sentinel 1 and 2 samples 
S1_sample <- stack("https://github.com/Leprechault/trash/raw/main/AguaSumida_PR001_1_2016-06-29_20160615_6_97EE.tif")
S1_sample
class      : RasterStack 
dimensions : 200, 200, 40000, 5  (nrow, ncol, ncell, nlayers)
resolution : 10, 10  (x, y)
extent     : 445099.9, 447099.9, 7779496, 7781496  (xmin, xmax, ymin, ymax)
crs        : +proj=utm +zone=22 +south +datum=WGS84 +units=m +no_defs 
names      :          vv,          vh,         RVI,       RATIO,        RFDI 
min values : -22.5303173, -31.3865096,   1.7272604,   0.7599905,  -0.6713753 
max values :  -2.9259523,  -8.6576518,   3.3427506,   5.0859694,   0.1363698

S2_sample <-stack("https://github.com/Leprechault/trash/raw/main/AguaSumida_PR001_1_2019-08-06_20190801_T22KDC.tif")
S2_sample
class      : RasterStack 
dimensions : 201, 200, 40200, 4  (nrow, ncol, ncell, nlayers)
resolution : 10, 10  (x, y)
extent     : 445100, 447100, 7779490, 7781500  (xmin, xmax, ymin, ymax)
crs        : +proj=utm +zone=22 +south +datum=WGS84 +units=m +no_defs 
names      : B2, B3, B4, B8 

# Plotting
par(mfrow=c(1,2))
plotRGB(S1_sample,  r = 3, g = 2, b = 1, stretch = "lin")
plotRGB(S2_sample,  r = 3, g = 2, b = 1, stretch = "lin")
#

s1&2

The final objective is extract the values of Sentinel-1 in the same pixels of Sentinel-2, including NAs (some times my images are crop in one of the sensors). Please, help me. Thanks in advance.

Solution

  • Example data

    library(terra)
S1 <- rast("https://github.com/Leprechault/trash/raw/main/AguaSumida_PR001_1_2016-06-29_20160615_6_97EE.tif")
S2 <- rast("https://github.com/Leprechault/trash/raw/main/AguaSumida_PR001_1_2019-08-06_20190801_T22KDC.tif")

plot(S2, 1, legend=FALSE)
plot(S1, 1, add=TRUE, legend=FALSE, alpha=.5)

    Your question is not very clear, but I think that you are looking for something like this:

    S2 <- resample(S2, S1)

    Now S1 and S2 have the same geometry (spatial extent and resolution).