2-spatial-data

class: title-slide
background-size: cover

<h1 class="center middle">Getting Geospatial Data<br>Into R</h1>

---

# Ways to get spatial data into R

* Load external spatial files

* Load or fetch data with specialized R packages

---

# In this section I'll be showing some code I'll cover in more detail later

---

background-image: url("images/background.png")
background-size: cover
class: center, middle, separator-slide

# Reading in existing data

---

# What function you use depends on the type of data

* Read **vector** data with the {sf} package

* Read **raster** data with the {raster} package

---

class: middle, center

# Reading vector data

---

# For vector data use `read_sf()` from {sf}

---

# You can also use `st_read()` from {sf} but `read_sf()` is more "tidy"

* `stringsAsFactors = FALSE`

* `quiet = TRUE`

* `as_tibble = TRUE`

---

# `read_sf()` to read many different file types

* Shapefiles

* Geopackages

* Geojson

* Even databases!

---

# This makes things so much easier!

![:imgcenterscale 50%](images/section-2/yay-hooray.jpg)

---

# A little more detail on the most common vector file types

---

# Shapefiles are the most common spatial files

---

# Shapefiles have been around a long time!

![:imgcenterscale 35%](images/section-2/moleman.png)

---

# Shapefiles can be unpleasant to work with

* Columns can only be 10 or fewer characters

* Inconveniently a single shapefile is represented by multiple files

---

# This is one "shapefile"

![:imgcenterscale 50%](images/section-2/oneshapefile.png)

.footnote[A shapefile can also have several other associated files]

---

# Time to move away from shapefiles

![:imgcenterscale 85%](images/section-2/Shapefile_must_die_.png)

---

# Geopackages are rapidly gaining acceptance

* Open format

* Just one file, technically they are a SQLite container

* Can store multiple layers in one file

---

# Geojson, spatial data for the web

* geojson is json (javascript) with geographic attributes

* Files can be a little larger

* Extension is usually `.geojson` (sometimes `.json`)

* Can only be latitude/longitude

---

# One point in geojson

```js
{
  "type": "FeatureCollection",
  "crs": {
    "type": "name",
    "properties": {
      "name": "urn:ogc:def:crs:OGC:1.3:CRS84"
    }
  },
  "features": [{
      "type": "Feature",
      "properties": {
        "location": "Gimme! Coffee"
      },
      "geometry": {
        "type": "Point",
        "coordinates": [-73.995001, 40.722401]
      }
    }]
}
```

---

# gimme! coffee, one of my favorites

![:imgcenterscale 75%](images/section-2/gimme_coffee.png)

---

# A note on topojson, another format for the web

* Topojson is similar to geojson but stores geometries more efficiently

* For example, a border between two countries would be stored just once.

* `read_sf()` can also read this

---

# Let's see `read_sf()` in action

---

# Simple example of `read_sf()`

```r
library(sf)
boroughs <- read_sf("boroughs.shp")
```

---

# `read_sf()` has a consistent syntax

```r
boroughs <- read_sf("boroughs.shp")
boroughs <- read_sf("boroughs.geojson")
boroughs <- read_sf("boroughs.gpkg")
```

---

# The result will be a {sf} table ... more details later

```r
glimpse(boroughs)
```

```
## Observations: 5
## Variables: 7
## $ BoroCode   <int> 1, 2, 5, 3, 4
## $ BoroName   <chr> "Manhattan", "The Bronx", "Staten…
## $ Shape_Leng <dbl> 339736.6, 397460.6, 318700.5, 576…
## $ Shape_Area <dbl> 635147797, 1182399343, 1630762350…
## $ diso       <int> 1, 1, 1, 1, 1
## $ AreaSqMile <dbl> 22.78279, 42.41274, 58.49555, 71.…
## $ geom       <MULTIPOLYGON [US_survey_foot]> MULTIPO…
```

---

# You can also read from a URL directly

```r
# Topojson
usa <- read_sf("http://bit.ly/2NhznGt")
```

```r
# Plotting discussed later
st_geometry(usa) %>% plot()
```

![:imgcenterscale 90%](images/section-2/topojson.png)

---

# A note on making non-spatial data spatial

* Addresses

* Coordinates

* Place names

---

# If you have addresses, you need to "geocode" to get coordinates

```r
# Uses Google Maps
ggmap::geocode("Hilton San Francisco Union Square")
```

```r
# Uses Open Street Map
tmaptools::geocode_OSM("Hilton San Francisco Union Square", 
        as.sf = TRUE) %>% 
  glimpse()
```

```
## Observations: 1
## Variables: 8
## $ query    <chr> "Hilton San Francisco Union Square"
## $ lat      <dbl> 37.78573
## $ lon      <dbl> -122.4104
## $ lat_min  <dbl> 37.78519
## $ lat_max  <dbl> 37.78612
## $ lon_min  <dbl> -122.4112
## $ lon_max  <dbl> -122.4096
## $ geometry <POINT [°]> POINT (-122.4104 37.78573)
```

---

# What if you have coordinates?

You need to convert them to a spatial object with {sf}

---

# Here is a table of coordinates

```r
regular_table
```

```
## # A tibble: 2 x 4
##      id latitude longitude name                  
##   <int>    <dbl>     <dbl> <chr>                 
## 1     1     40.7     -74.0 Empire State Building 
## 2     2     40.7     -74.0 One World Trade Center
```

---

# By the way X = longitude and Y = latitude

A mnemonic for latitude, longitude...

"A lat lays flat"

"Lat are steps on a ladder"

---

# We'll cover this in more detail later but...

You use `sf::st_as_sf()` to convert coordinates to an {sf} object

```r
spatial_table <- regular_table %>% 
  st_as_sf(coords = c("longitude", "latitude"), crs = 4326)
```

```r
st_geometry(spatial_table) %>% plot(cex = 2, col = "blue")
```

---

# How about place names?

You need to link them to a spatial boundary file

---

# An example with US states

```r
# My list of place names
my_states <- data.frame(NAME = c("California", "Nevada"))
my_states
```

```
##         NAME
## 1 California
## 2     Nevada
```

---

# Find an R package or spatial file with the boundaries