I am currently trying to obtain two rasters that are the Principal component axes 1 and 2 for a list of bioclimatic variables that I am using. I am using the rasterPCA() function that exists in the RStoolbox in R, but I am unable to write it to a raster. Below is the code and the error that I am getting:


##Loading the variables and then creating a raster stack, followed by a PCA
bio3 <- raster("C:\\Users\\rameshv\\bio3")
bio4 <- raster("C:\\Users\\rameshv\\bio4")
bio5 <- raster("C:\\Users\\rameshv\\bio5")
bio6 <- raster("C:\\Users\\rameshv\\bio6")
bio14 <- raster("C:\\Users\\rameshv\\bio_14")

pres_stack <- stack(bio3, bio4,bio5,bio6,bio14)

pre_pca <-  rasterPCA(pres_stack, nComp = 2)

writeRaster(pre_pca, "C:\\Users\\rameshv\\PC.tif")

Error in (function (classes, fdef, mtable) : unable to find an inherited method for function ‘writeRaster’ for signature ‘"rasterPCA", "character"’

I even tried a modification of the above code

writeRaster(pre_pca,"C:\\Users\\rameshv\\4_PCAforR\\PC.asc", format="ascii", bylayer=T)

Getting the same error as above: Error in (function (classes, fdef, mtable) : unable to find an inherited method for function ‘writeRaster’ for signature ‘"rasterPCA", "character"’

Any suggestions?

  • The out put of the function is a list object that contains the rasters. You can have the function result in a raster brick my using pre_pca <- rasterPCA(pres_stack, nComp = 2)$map Mar 7 '17 at 20:03
  • Thanks @JeffreyEvans. That is definitely a quicker workaround. Failed to notice that it is a list object. Mar 7 '17 at 20:05
  • @JeffreyEvans Could I nudge you towards this? This query was based off of one of your suggestions: gis.stackexchange.com/questions/230970/… Mar 7 '17 at 20:26

After digging in to the RSToolbox package, I realized that the rasterPCA() command writes in a "large rasterPCA" that consists of 3 elements and I failed to call the right element.

Shown below is the output from using the rasterPCA() command:

> pre_pca
  rasterPCA(img = pres_stack, nComp = 2)

  princomp(cor = spca, covmat = covMat[[1]])

  Standard deviations:
  Comp.1       Comp.2       Comp.3       Comp.4       Comp.5       Comp.6  
  113.53111875  22.39603549   8.08710815   4.11482299   2.21497438   0.79337709    

  6  variables and  83300 observations.

  class       : RasterBrick 
  dimensions  :   (nrow, ncol, ncell, nlayers)
  resolution  : 0.04166667, 0.04166667  (x, y)
  extent      :  (xmin, xmax, ymin, ymax)
  coord. ref. : +proj=longlat +ellps=WGS84 +towgs84=0,0,0,0,0,0,0 +no_defs 
  data source : in memory
  names       :        PC1,        PC2 
  min values  : -870.80335,  -49.97994 
  max values  :  141.77369,   75.40234 

 [1] "rasterPCA" "RStoolbox"

By calling $map when using the writeRaster command does the trick.

writeRaster(pre_pca$map,"C:\\Users\\rameshv\\4_PCAforR\\PC.asc", format="ascii", bylayer=T)

Now I have an output of two rasters with a .asc extension that can be visualized in ArcMap now.


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.