Conditionally calculating the difference between max(raster) and each raster layer of raster stack

Question

I have a list of raster stacks named r.lst. I need to calculate the difference between max(r.lst[[i]]) (stacked rasters) and every raster layer of that raster stack. However, I want to do this conditionally for each pixel by considering which.max(r.lst[[i]]). So that each cell in deducted from its previous max not the next. Please see below example:

# example data-------------------------------------------
set.seed(123)
#our list of rasters
r.lst <- as.list(1:3)
# setting up list pf raster stacks
r1 <- raster(nrows = 1, ncols = 1, res = 0.5, xmn = -1.5, xmx = 1.5, ymn = -1.5, ymx = 1.5, vals = runif(36, 1, 5))
r.lst[[1]] <- stack(lapply(1:8, function(i) setValues(r1,runif(ncell(r1)))))
r.lst[[2]] <- stack(lapply(1:6, function(i) setValues(r1,runif(ncell(r1)))))
r.lst[[3]] <- stack(lapply(1:7, function(i) setValues(r1,runif(ncell(r1)))))

I can easily calculate the difference for every single layer within each three raster stack of r.lst[[i]] as below:

# compute the difference between max and min
x1 <- list()
x2 <- list()
xx <- list()

for (i in 1:length(r.lst)) {
  x1[[i]] <- max(r.lst[[i]])
  x2[[i]] <- which.max(r.lst[[i]])
  x <- list()
  for (j in 1:nlayers(r.lst[[i]])){
    x[[j]] <- x1[[i]]-r.lst[[i]][[j]]
  }
  xx[[i]] <- stack(x)
  print(i)
}

par(mfrow=c(1,4), oma=c(0,0,0,1), mai=c(0.7,0.4,0.7,0.4))
plot(r.lst[[1]][[1]], main="Input.Raster")  #Input from stack
plot(x1[[1]], main="Max")                   #Max of stack
plot(x2[[1]], main="Max.Index")             #Index of Max value in raster stack
plot(xx[[1]][[1]], main="Results")          #OUTPUT (not what I wanted)

But this is not what I want. I need to consider which.max(r.lst[[i]]) in each round of calculation so that if the index of a layer under consideration is bigger than the index of which.max(r.lst[[i]]) we calculate the difference, if not I need to calculate the difference considering earlier max layer. Simply the calculation for every raster layer is only valid compared to its earlier max not the next.

TL; DR. But my first impression is you just need one additional variable max that is initiated before the for-loop and increases in every iteration only if max(r.lst[[i]]) > max. Or is this not what you want? — symbolrush
@symbolrush thanks for trying to help. It's not what I want. The condition would be based on corresponding individual raster cells of which.max(r.lst[[i]]) and r.lst[[i]][[j]] index itself. Simple max(r.lst[[i]]) > max wont really help here. — Majid

Robert Hijmans Robert Hijmans · Accepted Answer · 2019-05-03T00:15:57

I believe you are looking for something like the below.

Example data

library(raster)
set.seed(123)
r <- raster(res = 0.5, xmn = -1.5, xmx = 1.5, ymn = -1.5, ymx = 1.5)
r.lst <- list()
r.lst[[1]] <- stack(lapply(1:8, function(i) setValues(r, runif(ncell(r)))))
r.lst[[2]] <- stack(lapply(1:6, function(i) setValues(r, runif(ncell(r)))))
r.lst[[3]] <- stack(lapply(1:7, function(i) setValues(r, runif(ncell(r)))))

Solution

xx <- list()
for (i in 1:length(r.lst)) {
    xx[[i]] <- cummax( r.lst[[i]] ) -  r.lst[[i]]
}

That is, you can use cummax to get the max value of the prior values (including the current value)

x <- c(1:3, 3:1, 4, 3)
cummax(x)
#[1] 1 2 3 3 3 3 4 4
x - cummax(x)
#[1]  0  0  0  0 -1 -2  0 -1

Conditionally calculating the difference between max(raster) and each raster layer of raster stack

1 Answers