CODEWITHSUNDEEP
rstatisticstime-seriesraster
tibi
tibi

Reputation: 69

How to calculate percentage of change from a raster stack?

I have a raster stack (yearly time series) of land cover maps and I want to calculate the percentage of change. I have found this question, but is not clear to me. Any ideas how to do this?

Supposing that my raster stack is named r_stk, currently the only statistics that I have calculated are the sd and cv.

Any ideas on how to calculate the percentage change? 

r_stk_sd<-calc(r_stk, fun=sd)
r_stk_cv<-r_stk_sd/r_stk_Mean

Upvotes: 2

Views: 990

Answers (1)

Robert Hijmans
Robert Hijmans

Reputation: 47536

Here is some example data. From ?raster::stack!

library(raster)
s <- stack(system.file("external/rlogo.grd", package="raster"))

I understand you have done this

 mn <- mean(s)
 sd <- calc(s, sd)
 cv <- sd / mn

Now you want "% change" but you do not define it. Please edit your question, and explain better what you want to compute. Do you want local change (on a cell by cell bases); but if so, how would you compute the percentage? Or do you want it globally (one number for each layer / time period). Do you want to compare all time periods with each other? Or??

Here another possibility, compare the first layer to all other layers. Perhaps this can get you started to do what you want, or to ask a better question. 

x <- s[[1]] != s[[-1]]

Are there any changed cells? 

y <- any(x)

What is the total fraction of cells has changed relative to year 1? 

cellStats(y, "mean")
#[1] 0.7923364

Or for each "year" relative to year 1.

cellStats(x, "mean")
#    green      blue 
#0.6218336 0.7891218

If you want year by year change, you could do 

 s1 <- s[[2:nlayers(s)]]
 s2 <- s[[1:(nlayers(s)-1)]]
 z <- s1 != s2
 cellStats(z, "mean")
 #     green      blue 
 # 0.6218336 0.7672624

Upvotes: 5

Related Questions