At present, I am looking at mapping permafrost in mountains, which is modelled as a function of Mean Annaul Air Temperature. To predict statistically the probability of permafrost, you use an equation with the erfc function to predict the probablity of permafrost occuring at any given point (0-1). So far I am having no such luck with this. Converting one number through erfc is simple in Excel but I cannot seem to do it for an entire raster dataset in R, whereas this is simple to do when using simple mathematical commands (adding, multiplying etc), for the whole dataset.

Any tips?


You can update any raster with any function in R by doing:

r[] = anything(r[])

as long as the anything function takes a vector and returns a vector of the same length.

The error function is (I think):

erfc = function(x){2 * pnorm(x * sqrt(2), lower.tail = FALSE)}

Make a test 3x4 raster with 0 to 1:

> r = raster(matrix(seq(0,1,len=12),3,4))

Calling the function fails:

> erfc(r)
Error in pnorm(x * sqrt(2), lower.tail = FALSE) : 
  Non-numeric argument to mathematical function  

But replacing the values like this works:

> r[] = erfc(r[])
> r[]
 [1] 1.0000000 0.6997229 0.4404763 0.2472381 0.8977020 0.6070706 0.3681447
 [8] 0.1985657 0.7970786 0.5203381 0.3037058 0.1572992

If you want to keep the original raster, make a copy, or you can use calc which returns a new raster:

> r2 = calc(r,erfc)
> r2[]
 [1] 0.1572992 0.3223904 0.5333322 0.7266033 0.2042477 0.3906013 0.6026209
 [8] 0.7788528 0.2596413 0.4618103 0.6675559 0.8239600
| improve this answer | |

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.