使用jsonlite读取GeoJSON后Leaflet无法渲染MultiPolygons的问题

Let's break down why your first approach doesn't work, and how to fix it so you can add your custom fields for a choropleth map.

The Root Cause #

When you use jsonlite::fromJSON() on a GeoJSON URL, it parses the JSON into a flattened R list/data frame structure by default. The problem is that leaflet's addGeoJSON() expects either raw GeoJSON text or a properly structured GeoJSON object—not the simplified, de-nested output from jsonlite. This flattening breaks the hierarchical structure that leaflet needs to recognize MultiPolygon features.

In contrast, readr::read_lines() just pulls the raw text of the GeoJSON file, preserving its original structure, so addGeoJSON() can interpret it correctly.

Fix 1: Preserve GeoJSON Structure with jsonlite #

If you want to stick with jsonlite, you need to avoid simplifying the GeoJSON structure during parsing, then convert it back to valid GeoJSON text before passing it to leaflet:

library(leaflet)
library(jsonlite)
library(geojsonlint)

url <- paste0("https://geodata.nationaalgeoregister.nl/cbsgebiedsindelingen/wfs?", 
              "request=GetFeature&service=WFS&version=1.1.0&", 
              "typeName=cbsgebiedsindelingen:cbs_arrondissementsgebied_2019_gegeneraliseerd&", 
              "outputFormat=application/json&srsName=EPSG:4326&propertyName=geom,statcode")

# Parse without simplifying the nested structure
boundaries <- jsonlite::fromJSON(url, simplifyVector = FALSE)
# Convert back to valid GeoJSON text
boundaries_geojson <- jsonlite::toJSON(boundaries, auto_unbox = TRUE)

# Now it will render correctly
leaflet() %>% 
  addTiles() %>% 
  addGeoJSON(boundaries_geojson) %>% 
  setView(5.387740, 52.155499, 7)

The simplifyVector = FALSE argument tells jsonlite to keep the original nested list structure of the GeoJSON, so converting it back with toJSON() produces a valid format that leaflet understands.

Fix 2: Use sf (Recommended for Choropleths) #

For adding custom fields and building choropleths, the sf package is far more flexible—it's the standard tool for spatial data in R. It reads GeoJSON directly into a spatial data frame, which you can manipulate just like a regular data frame (e.g., joining with other datasets):

library(leaflet)
library(sf)
library(dplyr)

url <- paste0("https://geodata.nationaalgeoregister.nl/cbsgebiedsindelingen/wfs?", 
              "request=GetFeature&service=WFS&version=1.1.0&", 
              "typeName=cbsgebiedsindelingen:cbs_arrondissementsgebied_2019_gegeneraliseerd&", 
              "outputFormat=application/json&srsName=EPSG:4326&propertyName=geom,statcode")

# Read GeoJSON directly into an sf spatial data frame
boundaries_sf <- st_read(url, quiet = TRUE)

# Example: Join with your custom dataset (replace with your actual data)
# custom_data <- data.frame(statcode = c("GM0014", "GM0034"), your_field = c(10, 25))
# boundaries_merged <- boundaries_sf %>% left_join(custom_data, by = "statcode")

# Build your choropleth map
leaflet(boundaries_merged) %>%
  addTiles() %>%
  addPolygons(
    fillColor = ~colorNumeric("viridis", your_field)(your_field),
    fillOpacity = 0.7,
    color = "#BDBDC3",
    weight = 1,
    popup = ~paste("Statcode:", statcode, "<br>Value:", your_field)
  ) %>%
  setView(5.387740, 52.155499, 7)

This approach makes it trivial to merge your additional fields and customize the choropleth styling directly in leaflet.

内容的提问来源于stack exchange，提问作者Willy