The purpose of this tutorial is:
The end goal of this tutorial is for R users with minimal spatial experience to be able to search for, download and visualise Vicmap data. While this tutorial is aimed at spatial beginners, it can still be used as a reference for more experienced users.
Vicmap is the Victorian Government’s catalogue of spatial datasets. The catalogue features 463 datasets across land, property, infrastructure and environment and is the most authoritative suite of spatial data in Victoria.
This data catalogue is freely available to the public and can be accessed via several methods. For users of R, the fastest way to access up to date Vicmap datasets is to utilise the Web Feature Service (WFS). WFS is a standardised interface to request geographic information, regardless of the platform on which it is stored.
WFS requires a URL that contains the instructions for the query and is written in WFS specific terminology. This of course requires understanding of the WFS terminology and the time-consuming process of manually building the URL string.
VicmapR simplifies WFS queries by taking user supplied keywords and
automatically building the correctly formatted URL string.
The functions in VicmapR therefore provide a much faster and more robust
method for acquiring Vicmap data.
VicmapR uses lazy evaluation, in which we query but do not collect the data from the Vicmap database. This is called making a ‘Vicmap Promise’. A Vicmap Promise contains the dataset information but not the data itself. This promise can be filtered for a subset of the data before collecting. Using lazy evaluation we can collect the desired subset, instead of the entire dataset.
Note: A previous version of this tutorial used swooping birds as the example, but due to the migration of platforms that data was not available at time of writing
To demonstrate VicmapR’s features, we will download and explore a dataset from the Vicmap catalogue. The data is a spatial dataset of Melbourne’s tram network.
First, we need to search for the dataset. We can view all layers, or
do a keyword search. Use ignore.case = TRUE
for a
case-insensitive search. listLayers()
by default now
provides a column of ‘Abstract’ and ‘metadataID’. You can remove these
columns by setting the abstract argument to FALSE
(e.g. listLayers(abstract = FALSE)
)
#All layers
<- listLayers()
all_layers
# Case insensitive keyword search
<- listLayers(pattern = "tram", ignore.case = T) search
Viewing our search results shows us the name of the dataset and its description.
Once we have the name of the Vicmap dataset we would like to download, in this case ‘open-data-platform:ptv_metro_tram_route’, we can query the Vicmap database.
If we are still not sure from the name and description that this is the right dataset, we can look at the Vicmap Promise to get a snippet of the dataset contents.
vicmap_query(layer = "open-data-platform:ptv_metro_tram_route")
We can see that the dataset contains the route numbers and names. At
a glance this data should be adequate. If you want more extensive
information of the data you can use get_metadata()
to
download a list of (i) metadata about the data, (ii) a data dictionary
(work-in-progress) and (iii) a link to the metadata url.
<- vicmap_query(layer = "open-data-platform:ptv_metro_tram_route") %>%
metadata get_metadata() %>%
1]]
.[[
kbl(metadata) %>%
kable_styling()
The summary also shows that there are rows in this dataset. This isn’t a very large dataset, but if we are using this data in an interactive report or application, waiting to download the full dataset may be impractical.
Using lazy evaluation, we can use pipes to filter and subset the data, so that we only collect the desired subset. For example, let’s just look at the first 10 rows.
<- vicmap_query(layer = "open-data-platform:ptv_metro_tram_route") %>%
query head(10) %>% collect()
query
Now let’s only look at the tram route data. Note that VicmapR
datasets will retain id
and geometry
columns
even if not selected in the VicmapR promise. As this a simple features
(sf) dataset, the geometry corresponding to the features does not need
to be selected as it is always attached. The geometry column can only be
removed intentionally, for example, by using the
st_drop_geometry()
function in the sf
package.
<- vicmap_query(layer = "open-data-platform:ptv_metro_tram_route") %>%
tram_select filter(operator_name == "Yarra Trams") %>%
filter(num_of_stops > 20) %>%
select(route_id, route_short_name, trip_headsign, operator_name, route_km) %>%
collect()
%>% head(5) %>%
tram_select select(-id) %>% #condense table for viewing
kable() %>%
kable_styling()
By filtering the Vicmap_promise
before
collecting, we have downloaded only a subset of rows of
data.
mapview
packageOften the first thing we want to do is see the data on a map. The
simplest and quickest way to visualise spatial data is to use the
mapview
package:
library(mapview)
mapviewOptions(fgb = FALSE)
# Because the dataset is not big we can download all the data as 'tram_route'
<- vicmap_query(layer = "open-data-platform:ptv_metro_tram_route") %>%
tram_route collect()
# plot data
mapview(tram_route)
While the output is basic, it is very simple to achieve and lines can be clicked on to obtain the details from the other fields.
leaflet
For more customization, you might want to consider plotting your map
in leaflet
. Leaflet is JavaScript library that allows users
to produce interactive maps that work efficiently across most desktop
and mobile platforms. Using the leaflet package in R, we have the
flexibility of plotting more complex map products.
We will start by plotting the most basic map in leaflet. Instead of
using a single function like mapview()
, in leaflet we need
to add each component to the map. At a minimum we specify the leaflet
object, add a basemap and add the markers. Note: differing to mapview,
leaflet does not include marker labels by default, these must be added
with the popup
parameter.
library(leaflet)
# Create a palette
<- colorFactor("Accent", levels = tram_route$trip_headsign) #define a colour palette
pal
%>%
tram_route leaflet() %>%
addProviderTiles(providers$CartoDB.Positron) %>% #add third party base map
addPolylines(
color = ~pal(trip_headsign),
weight = 2,
stroke = 0.5, #removes outline
fillOpacity = 0.8,
popup = paste0("<b>Route No.: </b>", tram_route$route_short_name, "<br>", #format the popup with html tags
"<b>Route Name: </b>", tram_route$trip_headsign, "<br>",
"<b>Distance (km): </b>", tram_route$route_km)
)
Let’s say we want to know about the tram routes only in our LGA (e.g. Darebin). We can apply a spatial filter to our dataset so that only the tram routes in our suburb are shown. Where do we get the spatial data for our LGA? From the Vicmap catalogue of course!
The sf
package for R provides tools for dealing with
‘simple feature’ datasets, i.e. a data frame or tibble with a geometry
list-column. The sf package allows us to manupulate spatial data and
perform spatial operations. In this case, we want to find which points
in our tram route data intersect with the polygon for our suburb
polygon.
To do this we can use sf::st_intersection()
. Before
performing any spatial operations on a dataset, you want to make sure
the coordinate reference system (crs) is correct and if there are
multiple datasets, that the crs is the same for each dataset. You can
check the crs system with sf::st_crs()
.
# Suburb polygon for Darebin
<- vicmap_query(layer = "open-data-platform:lga_polygon") %>%
darebin select(lga_name) %>%
filter(lga_name == "DAREBIN") %>%
collect() %>%
st_make_valid() # magic fix for some spatial data
# Check crs for each spatial dataset
::st_crs(darebin) == sf::st_crs(tram_route)
sf
# Intersection
<- sf::st_intersection(tram_route, darebin)
darebin_trams mapview(darebin_trams) + mapview(darebin, alpha.regions = 0.3, col.regions= "green")
We probably regularly travel a bit further afield though, so lets
look at a radius from our home address. We can create the radius
geometry with st_buffer()
but first need to define the
coordinates for the centre of the radius (home). Don’t forget to specify
the crs as 4326 as we are dealing with a latitude and longitude
coordinate system.
<- -37.75215888969604
lat <- 145.02927170548745
lon
<- st_sfc(st_point(c(lon, lat)), crs = 4326) home
We have a problem though, latitude and longitude units are degrees
and st_buffer()
assumes units of meters. To convert from
degrees to metres we need to project our latitude and longitude, which
are represented as a point on a curved surface, onto a flat plane. To do
this, we use a projection standard chosen specifically for the zone
containing the coordinates. The standard is called the EPSG code and for
our coordinates is zone 55.
# Project coordinates
<- st_transform(home, "+proj=utm +zone=55")
home_utm
# Get buffer
<- sf::st_buffer(home_utm, dist = 10000)
home_radius
mapview(home_utm) + mapview(home_radius)
Now we can use this radius to filter our tram route data. Don’t forget, we need to convert the crs to 4283 to work with our tram route data
st_crs(home_radius) #crs is UTM Zone 55
<- st_transform(home_radius, crs = 4283)
home_radius st_crs(home_radius)
<- st_intersection(tram_route, home_radius)
tram_route_10k mapview(tram_route_10k) + mapview(home_radius, alpha.regions = 0.3, col.regions = "green")
Let’s say we feel like a bike ride. We’ve picked a route but want to know which areas to avoid so as not to get caught in tram tracks.
We can solve this problem in a similar way, using
st_intersection()
.
# Load bike route
<- sf::st_read(system.file("shapes/cycle_route.geojson", package="VicmapR"), quiet = F) %>%
route_line ::st_transform(4283)
sf
#Add buffer to route - we need to convert to m to do this (same as before)
<- st_transform(route_line, "+proj=utm +zone=55")
route_m <- st_transform(tram_route, "+proj=utm +zone=55") # convert to same crs for st_intersection
tram_m
<- st_intersection(route_m,tram_m)
trams_en_route
# Recorded tram tracks within 300m of the route
mapview(trams_en_route, col.regions = "Black") + mapview(route_line)
The VicmapR package offers tools for geometric filtering that avoid the need for sf package. The WFS Geoserver on which Vicmap is based supports several geometric filters - see the full list here.
These geometric filters allow us to perform basic spatial manipulation while retaining the benefits of lazy evaluation. So as before, if we want to filter our data before downloading, we can do so with more complex spatial operations. This is particularly useful when presenting VicmapR data in an interactive map, as it reduces the time for downloading data between re-rendering the map.
Let’s revisit our examples above. To get the tram routes only in
Darebin we can filter with INTERSECTS()
.
# trams in Darebin
<- vicmap_query(layer = "open-data-platform:ptv_metro_tram_route") %>%
darebin_trams filter(INTERSECTS(darebin)) %>%
collect()
mapview(darebin_trams) + mapview(darebin, alpha.regions = 0.3, col.regions = "orange")
And finally, the tram routes on our cycling route. To find tram
tracks within 30m we use the INTERSECTS()
function and the
cycling route with an added 300m buffer. Note: VicmapR geometric filters
will be simplified, thus if precise intersections are desired, it is
advised to do a cleanup of the filtered data once collected.
<- sf::st_read(system.file("shapes/cycle_route.geojson", package="VicmapR"), quiet = F) %>%
route_line ::st_transform(4283)
sf
# Condense line object
<- route_line %>%
route_poly ::st_transform(3111) %>%
sf::st_buffer(30) %>%
sf::st_cast("POLYGON") %>%
sf::st_transform(4283)
sf
<- vicmap_query(layer = "open-data-platform:ptv_metro_tram_route") %>%
route_intersection filter(INTERSECTS(route_poly)) %>%
collect() %>%
st_intersection(route_poly)
::mapview(route_poly) + mapview::mapview(route_intersection, color = "red", lwd = 5) mapview
After completing this tutorial you should have a basic understanding of plotting spatial data and performing some basic spatial operations. This tutorial is a quick guide and designed only to get you started. For further learning, see the following resources: