如何在R中编写switch语句？

Question

I am not able to write a switch statement correctly. And the examples available are not helpful. Basically, the statement should check what the value in a cell is (Choices: 1,2,4,8,16,32,64,128) and based on that increase/decrease the row and column number. Some switch statements have 2 calculations.

library(raster)
fdr<-raster("fdr.tif")
row1<-50
col1<-50
cell1<-fdr[row1,col1] #The point of origin
switch (cell1,
        4={row2 = row1 + 1
        },
        2={row2 = row1 + 1 & col2 = col1 + 1
        },
        1={col2 = col1 + 1
        },
        128={row2 = row1 - 1 & col2 = col1 + 1
        },
        64={row2 = row1 - 1
        },
        32={row2 = row1 - 1 & col2 = col1 - 1
        },
        16={col2 = col1 - 1
        },
        8={row2 = row1 + 1 & col2 = col1 - 1
        }
)

I am open to techniques other than a switch statement as well. Or any idea to make it faster. Eventually I'll wrap this into a function and do it for every cell (pixel).

Data:

1. FDR raster (small file): https://www.dropbox.com/s/7o3y8w01y6zqqwm/fdr.tif?dl=0

2. The concept is water flow direction: http://courses.washington.edu/gis250/lessons/hydrology/index.html#coded

Answer 1

This is how you write your switch and get it to return a vector of row, col:

row1 <- 50
col1 <- 50
cell1 <- 16

rowcol <- switch(as.character(cell1),
                 '4' = c(row1 + 1, col1),
                 '2'   = c(row1 + 1, col1 + 1),
                 '1'   = c(row1, col1 + 1),
                 '128' = c(row1 - 1, col1 + 1),
                 '64'  = c(row1 - 1, col1),
                 '32'  = c(row1 - 1, col1 - 1),
                 '16'  = c(row1, col1 - 1),
                 '8'   = c(row1 + 1, col1 - 1))

rowcol

> rowcol
[1] 50 49

You don't want to be doing a switch for each element of a large raster repeatedly; it will be glacial. Here is a vectorised operation that starts from any row/col and does a single update/move

## generate some dummy raster data
set.seed(1)
m <- matrix(sample(2^(0:7), 9, replace = TRUE), ncol = 3)

## collect row and column indices for raster
dd <- cbind(r = as.vector(row(m)), c = as.vector(col(m)))

## look-up matrix for row and col shifts
lu <- matrix(c( 0,  1,
                1,  1,
                1,  0,
                1, -1,
                0, -1,
               -1, -1,
               -1,  0,
               -1,  1), ncol = 2, byrow = TRUE)
## set rownames with powers of 2 to allow indexing using `m`
rownames(lu) <- 2^(0:7)

## need `m` as a vector
mc <- as.character(as.vector(m))

## which cell to move to next given values of `m`
move <- dd + lu[mc, ]
move

which gives

> move
      r c
 [1,] 2 1
 [2,] 3 1
 [3,] 3 0
 [4,] 0 3
 [5,] 3 3
 [6,] 2 3
 [7,] 0 4
 [8,] 1 2
 [9,] 2 2

Which is correct given the input data. Now you have the problem of making the next move and keeping track of things. In this example, the move would shift you to cells outside the raster for 3 of the 9 cells (the ones with a 0 in r or c and hence if you indexed back into m to get the power of two in the cell moved too , you only get 6 values returned

> m[move]
[1]   4  16  32  32 128   2

Hence you'd need to keep track of this as you walk, but this should get you started.