Visualizing geospatial data

class: center, middle, title-slide

.title[
# Visualizing geospatial data
]
.author[
### Claus O. Wilke
]
.date[
### last updated: 2024-04-08
]

---

background-image: url("https://clauswilke.com/dataviz/geospatial_data_files/figure-html/world-orthographic-1.png")
background-position: left 50% top 75%
background-size: 45%

## Parallels (latitude) and meridians (longitude)

???

Figure from [Claus O. Wilke. Fundamentals of Data Visualization. O'Reilly, 2019.](https://clauswilke.com/dataviz)

---
background-image: url("https://clauswilke.com/dataviz/coordinate_systems_axes_files/figure-html/worldmap-four-projections-1.png")
background-position: left 50% top 75%
background-size: 65%

## There are many ways to project onto a 2D plane

???

Figure from [Claus O. Wilke. Fundamentals of Data Visualization. O'Reilly, 2019.](https://clauswilke.com/dataviz)

---
background-image: url("https://clauswilke.com/dataviz/geospatial_data_files/figure-html/world-mercator-1.png")
background-position: left 50% top 70%
background-size: 45%

## There are many ways to project onto a 2D plane

.absolute-bottom-left[
Mercator projection: Shapes are preserved, areas are severely distorted
]

???

Figure from [Claus O. Wilke. Fundamentals of Data Visualization. O'Reilly, 2019.](https://clauswilke.com/dataviz)

---
background-image: url("https://clauswilke.com/dataviz/geospatial_data_files/figure-html/world-goode-1.png")
background-position: left 50% top 60%
background-size: 80%

## There are many ways to project onto a 2D plane

.absolute-bottom-left[
Goode homolosine: Areas are preserved, shapes are somewhat distorted
]

???

Figure from [Claus O. Wilke. Fundamentals of Data Visualization. O'Reilly, 2019.](https://clauswilke.com/dataviz)

---
background-image: url("https://clauswilke.com/dataviz/geospatial_data_files/figure-html/usa-orthographic-1.png")
background-position: left 50% top 60%
background-size: 45%

## Projecting the US

.absolute-bottom-left[
Alaska, Hawaii, and the lower 48 are far apart; difficult to show on one map
]

???

Figure from [Claus O. Wilke. Fundamentals of Data Visualization. O'Reilly, 2019.](https://clauswilke.com/dataviz)

---
background-image: url("https://clauswilke.com/dataviz/geospatial_data_files/figure-html/usa-true-albers-1.png")
background-position: left 50% top 60%
background-size: 50%

## Projecting the US

.absolute-bottom-left[
A fair, area-preserving projection
]

???

Figure from [Claus O. Wilke. Fundamentals of Data Visualization. O'Reilly, 2019.](https://clauswilke.com/dataviz)

---
background-image: url("https://clauswilke.com/dataviz/geospatial_data_files/figure-html/usa-albers-1.png")
background-position: left 50% top 60%
background-size: 50%

## A common visualization. What happened to Alaska?

.absolute-bottom-left[
Alaska and Hawaii were moved closer; Alaska was also reduced in size
]

???

Figure from [Claus O. Wilke. Fundamentals of Data Visualization. O'Reilly, 2019.](https://clauswilke.com/dataviz)

---
background-image: url("https://clauswilke.com/dataviz/geospatial_data_files/figure-html/usa-albers-revised-1.png")
background-position: left 50% top 60%
background-size: 50%

## A fair visualization of the 50 states

.absolute-bottom-left[
Alaska is the largest state; 2.2 the size of Texas
]

???

Figure from [Claus O. Wilke. Fundamentals of Data Visualization. O'Reilly, 2019.](https://clauswilke.com/dataviz)

[//]: # "segment ends here"

---
class: center middle

## Choropleth mapping: Coloring areas by a data value

---
background-image: url("https://clauswilke.com/dataviz/geospatial_data_files/figure-html/population-density-counties-1.png")
background-position: left 50% top 60%
background-size: 50%

## US population density as a choropleth map

.absolute-bottom-left[
Alaska has very low population density
]

???

Figure from [Claus O. Wilke. Fundamentals of Data Visualization. O'Reilly, 2019.](https://clauswilke.com/dataviz)

---
background-image: url("https://clauswilke.com/dataviz/geospatial_data_files/figure-html/population-density-counties2-1.png")
background-position: left 50% top 60%
background-size: 50%

## US population density as a choropleth map

.absolute-bottom-left[
Alaska has very low population density
]

???

Figure from [Claus O. Wilke. Fundamentals of Data Visualization. O'Reilly, 2019.](https://clauswilke.com/dataviz)

---
background-image: url("https://clauswilke.com/dataviz/geospatial_data_files/figure-html/median-income-counties-binned-1.png")
background-position: left 50% top 60%
background-size: 50%

## US median income as a choropleth map

.absolute-bottom-left[
A binned color scale can make the map more readable
]

???

Figure from [Claus O. Wilke. Fundamentals of Data Visualization. O'Reilly, 2019.](https://clauswilke.com/dataviz)

---
background-image: url("https://clauswilke.com/dataviz/geospatial_data_files/figure-html/median-income-states-1.png")
background-position: left 50% top 60%
background-size: 50%

## Choropleth maps can be misleading

.absolute-bottom-left[
Large area of Alaska makes it appear very rich; remember, it's mostly empty
]

???

Figure from [Claus O. Wilke. Fundamentals of Data Visualization. O'Reilly, 2019.](https://clauswilke.com/dataviz)

---
background-image: url("https://clauswilke.com/dataviz/geospatial_data_files/figure-html/median-income-statebins-1.png")
background-position: left 50% top 60%
background-size: 50%

## A cartogram heatmap may be preferable

.absolute-bottom-left[
Each state is shown as an equally sized square
]

???

Figure from [Claus O. Wilke. Fundamentals of Data Visualization. O'Reilly, 2019.](https://clauswilke.com/dataviz)

[//]: # "segment ends here"

---
class: center middle

## Maps and layers

---
background-image: url("https://clauswilke.com/dataviz/geospatial_data_files/figure-html/sfbay-overview-1.png")
background-position: left 50% top 60%
background-size: 50%

## Maps show data in a geospatial context

.absolute-bottom-left[
Wind turbines in the San Francisco Bay Area
]

???

Figure from [Claus O. Wilke. Fundamentals of Data Visualization. O'Reilly, 2019.](https://clauswilke.com/dataviz)

---
background-image: url("https://clauswilke.com/dataviz/geospatial_data_files/figure-html/sfbay-layers-1.png")
background-position: left 50% top 60%
background-size: 50%

## Maps are composed of several distinct layers

.absolute-bottom-left[
Wind turbines in the San Francisco Bay Area
]

???

Figure from [Claus O. Wilke. Fundamentals of Data Visualization. O'Reilly, 2019.](https://clauswilke.com/dataviz)

---
background-image: url("https://clauswilke.com/dataviz/geospatial_data_files/figure-html/shiloh-map-1.png")
background-position: left 50% top 60%
background-size: 50%

## The concept of aesthetic mappings still applies

.absolute-bottom-left[
Location of individual wind turbines in the Shiloh Wind Farm
]

???

Figure from [Claus O. Wilke. Fundamentals of Data Visualization. O'Reilly, 2019.](https://clauswilke.com/dataviz)

[//]: # "segment ends here"

---
class: middle center

## Making geospatial visualizations in R

---
background-image: url("https://user-images.githubusercontent.com/520851/50280460-e35c1880-044c-11e9-9ed7-cc46754e49db.jpg")
background-position: left 20% bottom 10%
background-size: 65%

## The **sf** package: Simple Features in R

.absolute-bottom-right[
Artwork by <a href="https://twitter.com/allison_horst/status/1071456081308614656">Allison Horst</a>
]

???

Artwork by <a href="https://twitter.com/allison_horst/status/1071456081308614656">Allison Horst</a>

---

## Getting the data

We'll be working with the `texas_income` dataset:

.tiny-font[

```r
texas_income <- readRDS(url("https://wilkelab.org/SDS375/datasets/Texas_income.rds"))

texas_income
```

```
Simple feature collection with 254 features and 4 fields
Geometry type: MULTIPOLYGON
Dimension:     XY
Bounding box:  xmin: -106.6456 ymin: 25.83738 xmax: -93.50829 ymax: 36.5007
Geodetic CRS:  NAD83
First 10 features:
    FIPS    county median_income  moe                       geometry
1  48001  Anderson         41327 1842 MULTIPOLYGON (((-96.0648 31...
2  48003   Andrews         70423 6038 MULTIPOLYGON (((-103.0647 3...
3  48005  Angelina         44223 1611 MULTIPOLYGON (((-95.00488 3...
4  48007   Aransas         41690 3678 MULTIPOLYGON (((-96.8229 28...
5  48009    Archer         60275 5182 MULTIPOLYGON (((-98.95382 3...
6  48011 Armstrong         59737 4968 MULTIPOLYGON (((-101.6294 3...
7  48013  Atascosa         52192 3005 MULTIPOLYGON (((-98.80479 2...
8  48015    Austin         53687 3810 MULTIPOLYGON (((-96.62085 3...
9  48017    Bailey         37397 8652 MULTIPOLYGON (((-103.0469 3...
10 48019   Bandera         49863 7193 MULTIPOLYGON (((-99.60332 2...
```
]

---

## The **sf** package: Simple Features in R

.tiny-font[

```r
texas_income
```

---

## The **sf** package: Simple Features in R

.tiny-font[

```r
texas_income$geometry
```

```
Geometry set for 254 features 
Geometry type: MULTIPOLYGON
Dimension:     XY
Bounding box:  xmin: -106.6456 ymin: 25.83738 xmax: -93.50829 ymax: 36.5007
Geodetic CRS:  NAD83
First 5 geometries:
```

```
MULTIPOLYGON (((-96.0648 31.98066, -96.06305 31...
```

```
MULTIPOLYGON (((-103.0647 32.52219, -103.0005 3...
```

```
MULTIPOLYGON (((-95.00488 31.42396, -95.00334 3...
```

```
MULTIPOLYGON (((-96.8229 28.16743, -96.82127 28...
```

```
MULTIPOLYGON (((-98.95382 33.49637, -98.95377 3...
```
]

---

## The **sf** package: Simple Features in R

.tiny-font[

```r
texas_income %>% 
  filter(county == "Travis")
```

```
Simple feature collection with 1 feature and 4 fields
Geometry type: MULTIPOLYGON
Dimension:     XY
Bounding box:  xmin: -98.17298 ymin: 30.0245 xmax: -97.36954 ymax: 30.62825
Geodetic CRS:  NAD83
   FIPS county median_income moe                       geometry
1 48453 Travis         61451 591 MULTIPOLYGON (((-98.15927 3...
```
]

---

## ggplot supports simple features with `geom_sf()`

.tiny-font.pull-left[

```r
# plot all of Texas
ggplot(texas_income) + 
  geom_sf()
```
]

.pull-right.width-50[
<img src="geospatial-data_files/figure-html/texas-no-aes-out-1.svg" width="100%" />

]

---

## ggplot supports simple features with `geom_sf()`

.tiny-font.pull-left[

```r
# plot only Travis County
texas_income %>% 
  filter(county == "Travis") %>%
  ggplot() + 
  geom_sf()
```
]

.pull-right.width-50[
<img src="geospatial-data_files/figure-html/texas-sf-filter-out-1.svg" width="100%" />

]

---

## ggplot supports simple features with `geom_sf()`

.tiny-font.pull-left[

```r
# plot the ten richest counties
texas_income %>% 
  slice_max(median_income, n = 10) %>%
  ggplot() + 
  geom_sf()
```
]

.pull-right.width-50[
<img src="geospatial-data_files/figure-html/texas-sf-filter2-out-1.svg" width="100%" />

]

---

## ggplot supports simple features with `geom_sf()`

.tiny-font.pull-left[

```r
# color counties by median income
texas_income %>%
  ggplot(aes(fill = median_income)) + 
  geom_sf()
```
]

.pull-right.width-50[
<img src="geospatial-data_files/figure-html/texas-no-coordsf-out-1.svg" width="100%" />
]

---

## ggplot supports simple features with `geom_sf()`

.tiny-font.width-50.pull-left[

```r
# highlight the ten richest counties
texas_income %>% 
  mutate(
    top_ten = rank(desc(median_income)) <= 10
  ) %>%
  ggplot(aes(fill = top_ten)) + 
  geom_sf() +
  scale_fill_manual(
    values = c(
      `TRUE` = "#D55E00",
      `FALSE` = "#E8EEF9"
    )
  )
```
]

.pull-right.width-50[
<img src="geospatial-data_files/figure-html/texas-sf-mutate-out-1.svg" width="100%" />

]

---

## ggplot supports simple features with `geom_sf()`

.pull-left.width-50[.tiny-font[

```r
# highlight the ten richest counties
texas_income %>% 
  mutate(
    top_ten = rank(desc(median_income)) <= 10
  ) %>%
  ggplot(aes(fill = top_ten)) + 
  geom_sf(color = "black", linewidth = 0.1) +
  scale_fill_manual(
    name = NULL,
    values = c(
      `TRUE` = "#D55E00",
      `FALSE` = "#E8EEF9"
    ),
    breaks = c(TRUE),
    labels = "top-10 median income"
  ) +
  theme_minimal_grid(11)
```
]

We apply styling as usual
]

.pull-right.width-50[
<img src="geospatial-data_files/figure-html/texas-sf-mutate2-out-1.svg" width="100%" />

]

---

## We can customize the projection with `coord_sf()`

.tiny-font.pull-left[

```r
ggplot(texas_income) + 
  geom_sf(
    aes(fill = median_income),
    color = "black", linewidth = 0.1
  ) +
  scale_fill_continuous_sequential(
    palette = "Blues", rev = TRUE
  ) +
  theme_minimal_grid(11)
```
]

.pull-right.width-50[
<img src="geospatial-data_files/figure-html/texas-styled-out-1.svg" width="100%" />
]

---

## We can customize the projection with `coord_sf()`

.tiny-font.pull-left[

```r
ggplot(texas_income) + 
  geom_sf(
    aes(fill = median_income),
    color = "black", linewidth = 0.1
  ) +
  scale_fill_continuous_sequential(
    palette = "Blues", rev = TRUE
  ) +
* coord_sf() +
  theme_minimal_grid(11)
```
]

.pull-right.width-50[
<img src="geospatial-data_files/figure-html/texas-coord-sf-out-1.svg" width="100%" />
]

---

## We can customize the projection with `coord_sf()`

.tiny-font.pull-left[

```r
ggplot(texas_income) + 
  geom_sf(
    aes(fill = median_income),
    color = "black", linewidth = 0.1
  ) +
  scale_fill_continuous_sequential(
    palette = "Blues", rev = TRUE
  ) +
  coord_sf(
*   # Texas Centric Albers Equal Area
*   crs = 3083
  ) + 
  theme_minimal_grid(11)
```
]

.pull-right.width-50[
<img src="geospatial-data_files/figure-html/texas-coord-sf-3083-out-1.svg" width="100%" />
]

.absolute-bottom-left[
Reference: https://spatialreference.org/ref/epsg/3083/
]

---

## We can customize the projection with `coord_sf()`

.tiny-font.pull-left[

```r
ggplot(texas_income) + 
  geom_sf(
    aes(fill = median_income),
    color = "black", linewidth = 0.1
  ) +
  scale_fill_continuous_sequential(
    palette = "Blues", rev = TRUE
  ) +
  coord_sf(
*   # Texas Centric Lambert Conformal Conic
*   crs = 32139
  ) + 
  theme_minimal_grid(11)
```
]

.pull-right.width-50[
<img src="geospatial-data_files/figure-html/texas-coord-sf-32139-out-1.svg" width="100%" />
]

.absolute-bottom-left[
Reference: https://spatialreference.org/ref/epsg/32139/
]

---

## We can customize the projection with `coord_sf()`

.tiny-font.pull-left[

```r
ggplot(texas_income) + 
  geom_sf(
    aes(fill = median_income),
    color = "black", linewidth = 0.1
  ) +
  scale_fill_continuous_sequential(
    palette = "Blues", rev = TRUE
  ) +
  coord_sf(
*   # Web Mercator (Google Maps)
*   crs = 3857
  ) + 
  theme_minimal_grid(11)
```
]

.pull-right.width-50[
<img src="geospatial-data_files/figure-html/texas-coord-sf-3857-out-1.svg" width="100%" />
]

.absolute-bottom-left[
Reference: https://spatialreference.org/ref/sr-org/7483/
]

---

## We can customize the projection with `coord_sf()`

.tiny-font.pull-left[

```r
ggplot(texas_income) + 
  geom_sf(
    aes(fill = median_income),
    color = "black", linewidth = 0.1
  ) +
  scale_fill_continuous_sequential(
    palette = "Blues", rev = TRUE
  ) +
  coord_sf(
*   # Longitude-Latitude WGS84 (GPS)
*   crs = 4326
  ) + 
  theme_minimal_grid(11)
```
]

.pull-right.width-50[
<img src="geospatial-data_files/figure-html/texas-coord-sf-4326-out-1.svg" width="100%" />
]

.absolute-bottom-left[
Reference: https://spatialreference.org/ref/epsg/4326/
]

---

## We can customize the projection with `coord_sf()`

.tiny-font.pull-left[

```r
ggplot(texas_income) + 
  geom_sf(
    aes(fill = median_income),
    color = "black", linewidth = 0.1
  ) +
  scale_fill_continuous_sequential(
    palette = "Blues", rev = TRUE
  ) +
  coord_sf(
*   # Alaska Albers equal area
*   crs = 3338
  ) + 
  theme_minimal_grid(11)
```
]

.pull-right.width-50[
<img src="geospatial-data_files/figure-html/texas-coord-sf-3338-out-1.svg" width="100%" />
]

.absolute-bottom-left[
Reference: https://spatialreference.org/ref/epsg/3338/
]

[//]: # "segment ends here"

---

## We can get map data from the rnaturalearth package

.pull-left[

A world map:

.tiny-font[

```r
library(rnaturalearth)

sf_world <- ne_countries(returnclass='sf')
ggplot(sf_world) + geom_sf()
```
]]

.pull-right.width-50[
<img src="geospatial-data_files/figure-html/world-map-out-1.svg" width="100%" />
]

---

## We can get map data from the rnaturalearth package

.pull-left[

A map of the lower 48:

.tiny-font[

```r
sf_us <- ne_states(
  country = "United States of America",
  returnclass='sf'
)

sf_us %>%
  # exclude Alaska (US02), Hawaii (US15)
  filter(!code_local %in% c("US02", "US15")) %>%
  ggplot() + geom_sf()
```
]]

.pull-right.width-50[
![](geospatial-data_files/figure-html/lower-48-map-out-1.svg)
]

---

## Further reading

- Fundamentals of Data Visualization: [Chapter 15: Visualizing geospatial data](https://clauswilke.com/dataviz/geospatial-data.html)
- **sf** package documentation: [Simple Features for R](https://r-spatial.github.io/sf/index.html)
- **ggplot2** reference documentation: [`geom_sf()`, `coord_sf()`](https://ggplot2.tidyverse.org/reference/ggsf.html)