Railway population

Michaël — Sun, 28 Dec 2025 19:00:00 GMT

Eurostar 3213 and 3214 in Marseille-Saint-Charles in 2017 – CC-BY by Roel Hemkes

The following map from 2024 by Simon C. Scherrer (via @freakonometrics) indicates that one-third of the Swiss population lives in a five-kilometer-wide strip on either side of the Intercity 1 railway line.

Figure 1: A third of the Switzerland population lives along the Geneva–St. Gallen railway line – 2024-04-24 by Simon C. Scherrer – data: Kontur/data.sbb.ch/Natural Earth

What is the situation in France?

I didn’t easily find the statistics per line, but I guess a similar configuration —main line (whatever it means), main cities, many passengers, crossing the country— would be the discontinued Eurostar (London)–Calais–Lille–Marne-la-Vallée–Lyon–Marseille or the current shorter TGV inOui Lille-Lyon-Marseille. However they don’t stop at Paris intra muros, so we miss a good part of the population.

Config

library(sf)
library(glue)
library(dplyr)
library(osmdata)
library(purrr)
library(httr2)
library(rnaturalearth)
library(stringr)
library(ggplot2)

sf_use_s2(FALSE)

Note

A 7z system binary is also required

Data

Railway

We can get the railway geometry from the OSM relation with {osmdata} and an Overpass API query.

osm <- r"([out:xml][timeout:6000];
relation(5951791);
(._;>;);
out body;)" |> 
  osmdata_sf() 

# the line
eurostar <- osm |> 
  pluck("osm_multilines")

# main stations
stations <- osm$osm_points |> 
  filter(railway == "stop", 
         uic_ref %in% c("8775100", "8772319", "8711184", "8722326"))

# base map from Natural Earth
fr <- ne_countries(scale = "large") |> 
  st_intersection(eurostar |> 
                    st_bbox() |> 
                    st_as_sfc() |> 
                    st_buffer(4.5, joinStyle = "MITRE", mitreLimit = 10))

eurostar |> 
  ggplot() +
  geom_sf(data = fr, color = "darkgrey") +
  geom_sf(color = "red", linewidth = 0.5, alpha = 0.8) +
  geom_sf(data = stations) +
  geom_sf_text(data = stations, 
               aes(label = str_wrap(name, width = 15,
                                    whitespace_only = FALSE)), 
               size = 3,
               hjust = 1.1) +
  labs(title = "London-Marseille",
       caption = glue("data: OpenStreetMap contributors
                      Natural Earth
                      https://r.iresmi.net - {Sys.Date()}")) +
  theme_void() +
  theme(plot.caption = element_text(size = 7, color = "grey40"),
        plot.background = element_rect(fill = "white"),
        plot.margin = unit(c(.2, .2, .2, .2), units = "cm"))

Figure 2: London-Marseille railway

Population

Population data comes from INSEE 2015 200 m grid.

Caution

The 7z file is itself zipped!

if (!file.exists("carreaux_200m_met.gpkg")) {
  pop_file <- "Filosofi2019_carreaux_200m_gpkg.zip"
  if (!file.exists(pop_file)) {
    request(glue("https://www.insee.fr/fr/statistiques/fichier/7655475/\\
                 {pop_file}")) |> 
      req_perform(pop_file)
  }
  unzip(pop_file)
  system("7z e Filosofi2019_carreaux_200m_gpkg.7z carreaux_200m_met.gpkg")
  system("rm Filosofi2019_carreaux_200m_gpkg.*")
}

pop <- read_sf("carreaux_200m_met.gpkg")

Results

Within 5 km

pop_5km <- pop |> 
  st_filter(eurostar |> 
              st_transform("EPSG:2154") |> 
              st_buffer(5000)) |> 
  st_drop_geometry() |> 
  summarise(pop_tot = sum(ind, na.rm = TRUE)) |> 
  pull(pop_tot)

pop_fr <- pop |> 
  st_drop_geometry() |> 
  summarise(pop_tot = sum(ind, na.rm = TRUE)) |> 
  pull(pop_tot)

Total population: 62,971,073
Population within 5 km: 4,163,426

So only 6.6 % of the french (metropolitan) population is within 5 km of this railway line.

Within 50 km

If we extend to 50 km, we’ll capture most part of Paris…

eurostar |> 
  ggplot() +
  geom_sf(data = fr, color = "darkgrey") +
  geom_sf(color = "red", linewidth = 0.5, alpha = 0.8) +
  geom_sf(data = eurostar |> 
            st_transform("EPSG:2154") |> 
            st_buffer(50000),
          fill = "red", alpha = 0.2) +
  geom_sf(data = stations) +
  geom_sf_text(data = stations, 
               aes(label = str_wrap(name, width = 15,
                                    whitespace_only = FALSE)), 
               size = 3,
               hjust = 1.1) +
  labs(title = "London-Marseille",
       caption = glue("data: OpenStreetMap contributors
                      Natural Earth
                      https://r.iresmi.net - {Sys.Date()}")) +
  theme_void() +
  theme(plot.caption = element_text(size = 7, color = "grey40"),
        plot.background = element_rect(fill = "white"),
        plot.margin = unit(c(.2, .2, .2, .2), units = "cm"))

Figure 3: London-Marseille railway in France and a 50 km buffer

pop_50km <- pop |> 
  st_filter(eurostar |> 
              st_transform("EPSG:2154") |> 
              st_buffer(50000)) |> 
  st_drop_geometry() |> 
  summarise(pop_tot = sum(ind, na.rm = TRUE)) |> 
  pull(pop_tot)

Population within 50 km: 25,696,130

So 40.8 % of the french (metropolitan) population is within 50 km of this railway line. Not bad, but the question could be now: does the train stop at a station near you? This is left as an exercise to the reader (spoiler: less and less…).

Oil spill

Michaël — Fri, 28 Nov 2025 22:00:00 GMT

Oiled Bird - Black Sea Oil Spill 11/12/07 – CC BY by Igor GOLUBENKOV (Marine Photobank)

Day 28 of 30DayMapChallenge: « Black » (previously).

There is a global oil spill dataset from 1967 to 2023 by Liu, Yin, and Cai (2025).

Config

library(glue)
library(dplyr)
library(sf)
library(janitor)
library(readxl)
library(leaflet)
library(leaflet.providers)
library(leaflet.extras)

Data

# https://figshare.com/ndownloader/files/55233773
if (!file.exists("enhanced_global_oil_spill_data_240711.xlsx")) {
  download.file("https://figshare.com/ndownloader/files/55233773",
                "enhanced_global_oil_spill_data_240711.xlsx")
}

oil <- read_xlsx("enhanced_global_oil_spill_data_240711.xlsx", 
                 sheet = "Oil spill incidents", 
                 .name_repair = make_clean_names) |> 
  mutate(release_m3 = if_else(max_ptl_release_gallons == -1, 
                              NA_real_, 
                              round(0.00454609 * max_ptl_release_gallons))) |> 
  st_as_sf(coords = c("lon", "lat"), crs = "EPSG:4326")

Map

oil |> 
  leaflet() |> 
  addTiles(attribution = r"(
           ://r.iresmi.net/">r.iresmi.net.
           data ://doi.org/10.1038/s41597-025-05601-9">Liu et al.
  )") |> 
  addProviderTiles("Stadia.AlidadeSmoothDark") |> 
  addCircleMarkers(
    radius = ~ release_m3 / 1e3,
    color = "grey",
    popup = ~ glue("{name} {open_date}

                   {commodity} - {release_m3} m³

                   {description}"),
    clusterOptions = markerClusterOptions()) |> 
  addFullscreenControl()

Figure 1: Oil spills 1967-2023

References

Liu, Yiming, Zhuoli Yin, and Hua Cai. 2025. “Enhanced Global Oil Spill Dataset from 1967 to 2023 Based on Text-Form Incident Information.” Scientific Data 12 (1): 1394. https://doi.org/10.1038/s41597-025-05601-9.