ms_simplify produces clockwise instead of counterclockwise coordinates

[gittybobomber]

At least for R Plotly choropleth geo plots, the coordinates need to be counterclockwise, otherwise the outside area instead of the inside area gets colored. But ms_simplify produces clockwise instead of counterclockwise coordinates.

I have described the problem and a solution in detail here:
https://stackoverflow.com/questions/77289351/choropleth-plot-gets-inverted-after-simplifying-geojson-with-rmapshaper

Could you change it or offer an option to chose between clockwise and counterclockwise?

[ateucher]

Interesting, it seems to be due to the fact the D3 requires rotation different to the simple features specification. Same issue reported here in the mapshaper repo.

I've simplified your example here:

library(plotly)
library(rmapshaper)
library(readr)

geojson <- read_file('https://geodata.ucdavis.edu/gadm/gadm4.1/json/gadm41_DEU_0.json') |> 
  ms_simplify(keep = 0.01)

g <- list(
  fitbounds = "locations",
  projection = list(type = 'mercator'),
  visible = FALSE
)

fig <- plot_ly()
fig <- fig %>% add_trace(
  type="choropleth",
  geojson=jsonlite::fromJSON(geojson, simplifyDataFrame = FALSE),
  locations="Germany",
  text="Germany",
  z=0,
  colors = c('white','#ed6a12'),
  showscale=T,
  marker = list(line=list(color='grey',width=0.25)),
  featureidkey = 'properties.COUNTRY'
)

fig <- fig %>% layout(
  geo = g
)

fig

[ateucher]

I'll see about adding the -o gj2008 flag as an argument, stay tuned.

[ateucher]

@gittybobomber this works for me now. Please install from:

remotes::install_github("ateucher/rmapshaper@gj-2008")

and try again, setting gj2008 = TRUE:

# remotes::install_github("ateucher/rmapshaper@gj-2008")

library(plotly)
library(rmapshaper)
library(readr)

geojson <- read_file('https://geodata.ucdavis.edu/gadm/gadm4.1/json/gadm41_DEU_0.json') |> 
  ms_simplify(keep = 0.01, gj2008 = TRUE)

g <- list(
  fitbounds = "locations",
  projection = list(type = 'mercator'),
  visible = FALSE
)

fig <- plot_ly()
fig <- fig %>% add_trace(
  type="choropleth",
  geojson=jsonlite::fromJSON(geojson, simplifyDataFrame = FALSE),
  locations="Germany",
  text="Germany",
  z=0,
  colors = c('white','#ed6a12'),
  showscale=T,
  marker = list(line=list(color='grey',width=0.25)),
  featureidkey = 'properties.COUNTRY'
)

fig <- fig %>% layout(
  geo = g
)

fig

[gittybobomber]

Thanks a lot, but unfortunately I have problems installing it.

I uninstall/unload first (otherwise R Warning: Paket rmapshaper wird gerade benutzt und deshab nicht installiert"):
detach("package:rmapshaper", unload=TRUE)
remove.packages('rmapshaper')

Then I install from your github:
remotes::install_github("ateucher/rmapshaper@gj-2008")
library(rmapshaper)

Then run
geojson <- read_file('https://geodata.ucdavis.edu/gadm/gadm4.1/json/gadm41_DEU_0.json') |>
ms_simplify(keep = 0.01, gj2008 = TRUE, sys = TRUE)

I get error:
Error: The mapshaper node library must be installed and on your PATH.
Install node.js (https://nodejs.org/en/) and then install mapshaper with:
npm install -g mapshaper
(I installed node.js, still error)

Or if I run
geosf <- geojson_sf('https://geodata.ucdavis.edu/gadm/gadm4.1/json/gadm41_DEU_0.json')
geosf_simpl <- ms_simplify(geosf, keep = 0.01, gj2008 = TRUE, sys = TRUE)

I get error:
Error in ms_sf(input, call, ...) : unused argument (gj2008 = TRUE)

I don't get these errors when I use the "normal" install.packages('rmapshaper').
Can you advise, please?

[ateucher]

Oh, sorry! I think the installation was fine, but I accidentally pasted the wrong ms_simplify() call. You can leave out sys= TRUE and just use:

geosf_simpl <- ms_simplify(geojson, keep = 0.01, gj2008 = TRUE)

[ateucher]

When you tried it on the sf object, that error is expected as it doesn't make sense for sf objects to be wound the wrong way. But I will make that error clearer.

Edit: I did implement the gj2008 argument for sf objects, though it seems unlikely to be helpful here

[gittybobomber]

Thank you very much, it works now!

Just another question: In the webservice https://mapshaper.org/ it is possible to tranform in the console via

$ -proj from=EPSG:25833 crs=EPSG:4326

from the coordinate system used by the official German statistical office https://regionalatlas.statistikportal.de/
UTM, Zone 32N, Datum ETRS89 (EPSG: 25832)

to the GeoJSON standard coordinations
World Geodetic System EPSG 4326, WGS84

Is this possible with rmapshaper, too? Could not find it yet.

[ateucher]

Is this possible with rmapshaper, too? Could not find it yet.

Not directly, because I try not to replicate what's already available in existing tools (I.e., sf::st_transform()).

You can perform pretty much any mapshaper command on geojson objects in R with apply_mapshaper_commands(), however it's not working at the moment with -proj... not sure why