Rasterizing using the "by" argument produce raster with possibly wrong layer names

[PMassicotte]

I am rasterizing using the by argument and found out that the resulting names in the created raster are kinda strange.

library(terra)
#> terra 1.7.71

r <- rast(xmin = 0, ncols = 18, nrows = 18)

# generate points
p <- spatSample(r, 1000, xy = TRUE, replace = TRUE)

p <- vect(p)

p$grp <- sample(c("a", "b"), nrow(p), replace = TRUE)

rasterize(p, r, fun = sum, by = "grp")
#> class       : SpatRaster 
#> dimensions  : 18, 18, 2  (nrow, ncol, nlyr)
#> resolution  : 10, 10  (x, y)
#> extent      : 0, 180, -90, 90  (xmin, xmax, ymin, ymax)
#> coord. ref. : lon/lat WGS 84 (CRS84) (OGC:CRS84) 
#> source(s)   : memory
#> names       : c("a", "a", "a"~ "a", "a", "a"), c("b", "b", "b"~ "b", "b", "b") 
#> min values  :                               1,                               1 
#> max values  :                               7,                               6

I was expecting names to be “a” and “b”. Is it expected? Any ideas on how to fix that?

^{Created on 2024-02-24 with reprex v2.1.0}

[rhijmans]

Than you, I now get:

rasterize(p, r, fun = sum, by = "grp")
#class       : SpatRaster 
#dimensions  : 18, 18, 2  (nrow, ncol, nlyr)
#resolution  : 10, 10  (x, y)
#extent      : 0, 180, -90, 90  (xmin, xmax, ymin, ymax)
#coord. ref. : lon/lat WGS 84 (CRS84) (OGC:CRS84) 
#source(s)   : memory
#names       : a, b 
#min values  : 1, 1 
#max values  : 6, 9 

[PMassicotte]

Thank you very much!

[ejlundgren]

Hi, first, thank you for the wonderful package. It's fantastic.

I am having a similar—but I would say weirder—problem. I am using terra version 1.7-78 on a macOS Sonoma 14.3.1 with R version 4.4.1. I don't think I can make a reproducible example, but here is my code and the output. This is code running on a random subsample of my data for testing purposes.

out <- terra::rasterize(x = fire_polys.vect[sample(nrow(fire_polys.vect), 50), ],
                          y = template,
                          field = "value",
                          touches = TRUE,
                          cover = TRUE,
                          by = "Fire_Year")
  out
  names(out)

[19] "qXKE>\004&\xc1l\xd7Z\x81\x85\xcb%\xc1Z\xd1\xe9\037^\xc9CAegetation/equids_wildfir \x8a\030\xb7(rCEAxW\xb4\x94\xd7CEA\xeb\x91\xf2HGDEAD2_20180916_bil/PRISM_ppTS\n65 \xc0\r\x85N\001"
[20] "egetation/equids_wildfir \x8a\030\xb7(rCEAxW\xb4\x94\xd7CEA\xeb\x91\xf2HGDEAD2_20180916_bil/PRISM_ppTS\n65 \xc0\r\x85N\001"
[21] " \x8a\030\xb7(rCEAxW\xb4\x94\xd7CEA\xeb\x91\xf2HGDEAD2_20180916_bil/PRISM_ppTS\n65 \xc0\r\x85N\001"
[22] "\x8a\030\xb7(rCEAxW\xb4\x94\xd7CEA\xeb\x91\xf2HGDEAD2_20180916_bil/PRISM_ppTS\n65 \xc0\r\x85N\001"
[23] "D2_20180916_bil/PRISM_ppTS\n65 \xc0\r\x85N\001"
[24] "TS\n65 \xc0\r\x85N\001"
[25] "\xc0\r\x85N\001"
[26] ",}w\027CHEA\xb3\x90廂HEA\xbaP\x84~\xadHEAH\x82\x88N\001"
[27] "H\x82\x88N\001"
[28] "E\x9b\xd9\xfb\xf9IEA\xda\xebЊ,JEAo\031\xaf\xcctJEA"
[29] ""
[30] "\b\x95\xe5W\003"
[31] "\b\037\xe2W\003"
[32] "4kmM3_201408_bil.hdr"
[33] ""
[34] "\xc0\xee\xdc\n\003"
[35] "\002"
[36] ""

These appear to be bits and pieces of file paths in my hard drive (though I'm confused by the '廂' symbol in output 26). When I try to run the same code on the entire dataset (without the sample() subset) I get the following error:

Error: basic_string

Which I suspect is related to some issue with these layer names.

If there's any way I can make this more reproducible let me know!

[rhijmans]

Hi, it is not a good idea to post more in a closed issue.
Have you inspected

unique(fire_polys.vect$Fire_Year)

I suspect it has these strange values.

[ejlundgren]

Hi, sorry. I guess I should have opened a new issue.

Fire_Year is a numeric variable as expected...

head(unique(fire_polys.vect$Fire_Year))

[1] 2021 2022 2023 2024 1860 1870

class(fire_polys.vect$Fire_Year)

[1] "numeric"

Very much a mystery. An easy work around is just putting the rasterization in a loop by Fire_Year, which is what I've done. Anyways, just wanted to flag this.