1

In the addon of the session, we counted the charging stations located within North-Rhine Westphalia (NRW). Still, we did not show how to get a point layer of NRW charging stations (“charger_nrw”).

Subset the data file yourself by relying on the spatial information of the file charging_points_ger.csv and a polygon of NRW. There are two ways to achieve this. How many chargers are located within NRW?

You need two datasets for that: the point layer charging_points_ger.csv (remember to adjust the crs) in the ./data folder and polygons of NRW. For the latter, you can again use the osmdata syntax.

There are two functions you can explore: sf::st_join and sf::st_intersection(). The default of sf::st_join() will leave you with a ‘left-join’ and return a data object with all chargers and matching district information for those that are located within NRW. You can reset the option to perform an ‘inner-join’ and keep only the observation that lay within the predefined area (sf::st_join(x , y, join = "", left = FALSE)).

solution 

# load charger
charger_ger <- 
  # Read charging station points datae
  readr::read_delim("./data/charging_points_ger.csv", 
                                 delim = ";") %>%
  # Filter out rows with missing longitude or latitude
  dplyr::filter(!is.na(longitude) & !is.na(latitude)) %>%
  # Convert data frame to sf object
  sf::st_as_sf(coords = c("longitude", "latitude"), crs = 4326) %>%
  # Reproject the spatial data to the desired CRS (Coordinate Reference System)
  sf::st_transform(crs = 3035) 
## Rows: 60560 Columns: 7
## ── Column specification ───────────────────
## Delimiter: ";"
## chr (3): operator, federal_state, type
## dbl (4): latitude, longitude, power_kw,...
## 
## ℹ Use `spec()` to retrieve the full column specification for this data.
## ℹ Specify the column types or set `show_col_types = FALSE` to quiet this message.
#  use the OSM function
nrw <-
  osmdata::getbb(
    "Nordrhein-Westfalen", 
    format_out = "sf_polygon"
  ) %>% 
  .$multipolygon %>% 
  sf::st_transform(3035)

# option 1
charger_nrw <- 
  charger_ger %>% 
  # Subset point data to sampling area
  sf::st_intersection(nrw)

# option 2
# spatial join
charger_nrw <-
  charger_ger %>%  
  sf::st_join(
    # point layer nrw
    nrw, 
    # chose intersect or within
    join = sf::st_intersects,
    # option FALSE will 
    # keep only the chargers 
    # which could be joined
    left = FALSE
  )
 
nrow(charger_nrw)
## [1] 11081
# 11081 chargers in NRW

2

Did the operationalization of train station accessibility convince you? The INKAR database offers another approach: Proportion of residents with max. 1000m linear distance to the nearest public transport stop in the district. We have everything needed to create this indicator on a smaller scale as well. What is the mean share of residents with max 1000m linear distance to the nearest train station in a 5km neighborhood of our fake respondents?

You can run the code below to load all the data you need.

nrw <-
  osmdata::getbb(
    "Nordrhein-Westfalen", 
    format_out = "sf_polygon"
  ) %>% 
  .$multipolygon %>% 
  sf::st_transform(3035) 

set.seed(1234)

fake_coordinates <-
  sf::st_sample(nrw, 1000) %>% 
  sf::st_sf() %>% 
  dplyr::mutate(
    id_2 = 
      stringi::stri_rand_strings(10000, 10) %>% 
      sample(1000, replace = FALSE)
  )

nrw_pt_trainstops <- sf::st_read("./data/nrw_pt_osmtrainstops.shp", crs = 3035)

inhabitants_ger <-
  z11::z11_get_100m_attribute(Einwohner)

As always, there are several ways to do this. Anne tried to keep the workflow as close to the functions taught in this course as possible and suggests the following steps:

  1. Create a point layer with the centroids of all grids in NRW based on the z11 population layer.

  2. Calculate the distance to the next train station for each grid.

  3. Create a column that equals 0 if the distance to the next train station is >1000m and contains the number of inhabitants if <1000.

  4. Rasterize the sf data object to receive two raster objects: number of inhabitants and number of inhabitants with a max 1000m distance to the next train station.

  5. Calculate the mean for the 5km buffer for each “respondent” for each raster.

  6. Calculate the share.

In the add-on slides of the raster session, Stefan introduced and gave some information on how to transform the raster to points (and back). To get a sf point layer for the raster object you can use terra::as.points() %>% sf::st_as_sf(). To rasterize the object, you can use terra::rast(vals = .$colname, resolution = 100).

solution 

# Extract inhabitants points for North Rhine-Westphalia (NRW)
nrw_pt_inhabitants <- 
  inhabitants_ger %>% 
  terra::crop(., nrw) %>% 
  # convert to points
  terra::as.points() %>%   
  # convert to sf object
  sf::st_as_sf() 

# Find the nearest train station for each grid centroid
nearest_station <- 
  sf::st_nearest_feature(nrw_pt_inhabitants, nrw_pt_trainstops)

# Calculate distances from inhabitants to the nearest train station
distances <- sf::st_distance(nrw_pt_inhabitants, 
                             nrw_pt_trainstops[nearest_station,], 
                             by_element = TRUE) 

# Create a raster representing population with train access within 1000m for NRW 
nrw_rast_inhabitants_access <- 
  nrw_pt_inhabitants %>% 
  dplyr::mutate(train_access = ifelse(as.numeric(distances) <= 1000, lyr.1, 0)) %>% 
  terra::rast(vals = .$train_access, resolution = 100) 

# Create a raster representing population  for NRW
# NOTE: You could also use the 'old' raster. However, by converting the
# data the extent of the raster layer might not fit anymore and needs adjustments.
nrw_rast_inhabitants <- 
  nrw_pt_inhabitants %>% 
  terra::rast(vals = .$lyr.1, resolution = 100) 

# Extract population within 5km buffers around each respondent
population_buffers <- 
  terra::extract(
    nrw_rast_inhabitants, 
    fake_coordinates %>% 
      sf::st_buffer(5000) %>% 
      terra::vect(), 
    fun = mean,
    na.rm = TRUE
  )

# Extract population within 5km buffers considering train access within 1000m
population_access_buffers <- terra::extract(
  nrw_rast_inhabitants_access, 
  fake_coordinates %>% 
    sf::st_buffer(5000) %>% 
    terra::vect(), 
  fun = mean,
  na.rm = TRUE
)

# Combine population data with train access information
linked_df <- 
  fake_coordinates %>% 
  dplyr::mutate(
    population = population_buffers[[2]],
    population_trainaccess = population_access_buffers[[2]],
    share_access = population_access_buffers[[2]] / population_buffers[[2]]
)

# Summary of the new data
summary(linked_df)
##           geometry        id_2          
##  POINT        :1000   Length:1000       
##  epsg:3035    :   0   Class :character  
##  +proj=laea...:   0   Mode  :character  
##                                         
##                                         
##                                         
##    population    population_trainaccess
##  Min.   :18.71   Min.   : 3.409        
##  1st Qu.:22.20   1st Qu.: 5.301        
##  Median :28.78   Median : 8.328        
##  Mean   :28.42   Mean   : 8.204        
##  3rd Qu.:33.67   3rd Qu.:10.704        
##  Max.   :41.75   Max.   :16.084        
##   share_access   
##  Min.   :0.1698  
##  1st Qu.:0.2404  
##  Median :0.2854  
##  Mean   :0.2795  
##  3rd Qu.:0.3174  
##  Max.   :0.3879