Invalid geometry for Amazônia in read_biomes

[OlivazShai]

When running:
read_biomes( year = 2019, simplified = TRUE )

The geometry for the Amazon is invalid, which makes it impossible to perform intersections and other operations. When running:
st_is_valid(reason=TRUE)

We get following result, for Amazon:
"Loop 114: Edge 168413 crosses edge 168418"

Applying st_make_valid doesn't solve the issue, for some reason...

[rafapereirabr]

Try using simplified = FALSE

read_biomes( year = 2019, simplified = FALSE)

[OlivazShai]

Ah sorry! I copy-pasted the wrong version, lack of attention there. It's actually the opposite. I had switched to simplified = TRUE earlier, because then I get no issue and can perform intersections normally (just not accurately, I reckon).

The problem I mentioned only appears when I use simplified = FALSE.

I basically want which biomes are in each municipality, so I imagine using simplified = TRUE for now is probably okay, right? Do you know any alternative sources or ways to get that info? Oh, and thx a lot for the quick reply!

[rafapereirabr]

Could you please share the code to reproduce the error ?

[OlivazShai]

Sure! I'm sending a PDF knit of the code, in R.

Essentially, I'm running:

biomes <- read_biomes(
  year = 2019, 
  simplified = FALSE,
  showProgress = FALSE
  )

mun <- read_municipality(
  code_muni = "all", 
  year = 2017,
  simplified = FALSE,
  showProgress = FALSE
  )

And doing:

st_intersects(
    mun,
    biomes,
    sparse = FALSE
  )

Which yields this error:

Error in wk_handle.wk_wkb(wkb, s2_geography_writer(oriented = oriented, :
Loop 114 is not valid: Edge 168413 crosses edge 168418

I think more details can be seen in the file
biome_issue.pdf
.

[rafapereirabr]

Hi @OlivazShai , thanks for reporting this issue. This is a problem with the original data set made publicly available by IBGE on the http: 'https://geoftp.ibge.gov.br/informacoes_ambientais/estudos_ambientais/biomas/vetores/Biomas_250mil.zip'

I haven't found a way to fix this topology error yet. I wanntry dissolving borders a fast approach like this one but I haven't found the time to test it yet.

[OlivazShai]

Thanks a lot for the help! Let me know if you find a solution, or if IBGE updates their data. I'm currently using the simplified polygons, which seem to suffice for my analysis by municipality. If/when I switch to a more granular approach I'll probably have to look for some workarounds, I'll let you know if I reach any usable solution too.

By the way, IBGE issued recently (like, two weeks ago) a list of municipalities by predominant biome, which is kind of what I was looking for. And there also is an older list, from 2019, with a slighly different approach. They're available at https://geoftp.ibge.gov.br/informacoes_ambientais/estudos_ambientais/biomas/documentos/

It's a simple table and very tractable as it is, but seems like it could be a nice addition to the geobr package :)

[OlivazShai]

As an update, I checked turning off s2 following this suggestion, with sf_use_s2(FALSE).

After doing so, running:

st_is_valid(biomes)

Retuns TRUE for all biomes.

Not ideal, though...