• Data Visualization
• Producing Environmental Hazards
• Activity
• Working with spatial data in R
• Application: Air Quality
• Wrap-up Discussion
• Additional Application: Flooding

The language of hazards has also meaning for social issues. For example, historical redlining used “hazardous” neighborhood labels to entrench ethnoracial spatial divisions.

Los Angeles, 1939

These designations matter today!

Lane et al., 2022

Using the data handouts:

• What can you tell me about differences in exposure to air quality across neighborhoods?

• What can you tell me about differences in exposure to air quality across neighborhoods?
• What challenges came up (or might come up)?

• To assess the social implications of environmental hazards, we often need to work with spatial data
• We will use spatial analysis in R to do this

• Free, open-source, statistical analysis tool
• Normally: RStudio

• Free, open-source, statistical analysis tool
• Normally: RStudio

• Here: tinyurl.com/steele229

• The big picture: we are creating, plotting, and joining spatial data to analyze an environmental hazard

How did we join points and polygons?

How did we join points and polygons?

• We created sf objects using st_as_sf
• We plotted the sf objects together using ggplot
• We joined sf objects using st_join

• How might this be useful for studying inequalities in pollution?
• What are some challenges that might come up?
  
• Could we use a similar method to analyze another hazard (for example, flooding)?

Tyler McDaniel

tylermc@stanford.edu

• Large flood risks exist outside of FEMA floodzones

Lane et al., 2022

A hazard is a geophysical that could do harm to humans/assets.

A risk is a probability of actual harm for humans/assets.

A hazard is a geophysical that could do harm to humans/assets.

A risk is a probability of actual harm for humans/assets.

Source: USGS

If a tree falls in a forest, and no one is around, is it:

• A hazard
• A risk
• Both
• Neither

• What is a simple feature?
• Two elements:
• Geometry noting its location in space
• Attributes noting all other properties
• Advantages: flexible, reproducible, join and analyze data all in one place

• A dataframe with any kind of information
• In R, # are comments, or notes to ourselves
• c() is how we store information
• <- is how we put information into an object

# create an air quality dataframe
aqi <- data.frame("monitor" = c("1", "2", "3"),
                  "pm" = c(4, 5, 10))

# look at the dataframe
aqi

##   monitor pm
## 1       1  4
## 2       2  5
## 3       3 10

• On your own, try changing one of these numbers!

• There were 3 types of geometries in the activity. What were they?

• Points!
• Lines!
• Polygons!

• Let’s start small

# try making a point
point <- st_point(x = c(0, 1))

point

## POINT (0 1)

-Try creating your own!

# try a line
line <- st_linestring(x = matrix(data = c(0, 1, 
                                          0, 0), 
                                 ncol = 2, 
                                 byrow = TRUE))

line

## LINESTRING (0 1, 0 0)

# try a polygon
polygon <- st_polygon(x = list(matrix(data = c(0, 1, 
                                               0, 0, 
                                               1, 2, 
                                               0 ,1), 
                                      ncol = 2,
                                      byrow = TRUE)))
polygon

## POLYGON ((0 1, 0 0, 1 2, 0 1))

• Let’s create air quality monitors locations by hand:

# let's create three points
point1 <- st_point(x = c(0, 1))

point2 <- st_point(x = c(2, 3))

point3 <- st_point(x = c(0, 4))

• Can you create the points from the activity?

• %<>% is one way to permanently change an object in R

# create an sf object
aqi %<>%
  bind_cols(bind_rows(data.frame(st_sfc(point1)), 
                      data.frame(st_sfc(point2)),
                      data.frame(st_sfc(point3))))


write_csv(aqi, "aqi.csv")

-Try plotting

• Examine this object
• Does it contain attributes and geometries?
• Is it an sf object?

# look at the class
class(aqi)

## [1] "data.frame"

• Still a dataframe …

• st_as_sf() makes the data an sf object

# make it sf
aqi %<>%
  st_as_sf()

• What is the class now?

• In ggplot2, ggplot() creates a blank canvas to plot data

# create canvas
ggplot()

• geom_sf adds a “geometry” layer to the plot

# create canvas and plot
ggplot()+
  geom_sf(data = aqi)

# plot an sf object
ggplot()+
  geom_sf(data = aqi, aes(col = pm))

• We can also add polygons

# first build dataframe
zones <- data.frame("area" = c("claremont", "pomona"),
                       "students" = c(234, 120))


# then build polygons
polygon1 <- st_polygon(list(matrix(data = c(0, 1, 0, 0, 1, 2, 0 ,1), 
                                      ncol = 2,
                                      byrow = TRUE)))

polygon2 <- st_polygon(list(matrix(data = c(1, 0, 0, 0, 1, 2, 1 ,0), 
                                      ncol = 2,
                                      byrow = TRUE)))

# create an sf object
zones %<>%
  bind_cols(bind_rows(data.frame(st_sfc(polygon1)), 
                      data.frame(st_sfc(polygon2))))

• What class is zones?

# make it sf
zones %<>%
  st_as_sf()

# multiple points
ggplot()+
  geom_sf(data = zones)

ggplot()+
  geom_sf(data = zones)+
  geom_sf(data = aqi, aes(col = pm))

st_join(aqi, zones)

## Simple feature collection with 3 features and 4 fields
## Geometry type: POINT
## Dimension:     XY
## Bounding box:  xmin: 0 ymin: 1 xmax: 2 ymax: 4
## CRS:           NA
##   monitor pm      area students    geometry
## 1       1  4 claremont      234 POINT (0 1)
## 2       2  5      <NA>       NA POINT (2 3)
## 3       3 10      <NA>       NA POINT (0 4)

• Does it matter which order you put them in?

• Air quality is measured at specific monitors
• Let’s read in air quality data from 2022

# read in data
id <- "1R3N_iToKDEIRapTKgAkhbzBZW6dRtm9e"
aqi22 <- read.csv(sprintf("https://docs.google.com/uc?id=%s&export=download", id))

# make air quality data spatial
aqi22 %<>%
  st_as_sf(coords = c("X", "Y"))

# set crs
aqi22 %<>%
  st_set_crs(st_crs("NAD83"))

• Draws form Census TIGER files (Topologically Integrated Geographic Encoding and Referencing system)
• Using tigris, we can gather lots of different shapefiles from the US Census and the National Historical Geographic Information System
• For example:

library(tigris)

# gather county shapefiles
counties(state = "CA")

• Using tigris, we can gather lots of different shapefiles from the US Census and the National Historical Geographic Information System.

• For example, counties:

# gather county shapefiles
counties_ca <- counties(state = "CA")

• How would we plot the data?

# plot counties 
ggplot(counties_ca)+
  geom_sf()

• We can plot these air quality monitors
• aes(col = ) adds color to a layer

# try plotting again
ggplot()+
  geom_sf(data = counties_ca)+
  geom_sf(data = aqi22, aes(col = `Arithmetic.Mean`))

• We may want to join the data with air quality measures

# join school district boundaries with AQI
counties_ca %<>%
  st_join(aqi22)

# try plotting again
ggplot(counties_ca)+
  geom_sf(aes(fill = `Arithmetic.Mean`))+
  theme_minimal()+
  labs(col = "PM2.5 Average")

We might be interested in data on risks. For example, the National Flood Insurance Program designates the 100 year floodplain (1% annual flooding likelihood) as the primary geography of flood risk.

Floodplain Example. Source: FEMA

## Reading layer `pomona_small' from data source 
##   `/Users/tylermcdaniel/Library/CloudStorage/GoogleDrive-tylermc@stanford.edu/My Drive/Applications/Scripps/doi_10_7280_D1RH7Z__v20220829/pomona_small.shp' 
##   using driver `ESRI Shapefile'
## Simple feature collection with 1 feature and 3 fields
## Geometry type: POINT
## Dimension:     XY
## Bounding box:  xmin: -117.773 ymin: 34.0383 xmax: -117.773 ymax: 34.0383
## Geodetic CRS:  NAD83

## Reading layer `flood_hazard_small' from data source 
##   `/Users/tylermcdaniel/Library/CloudStorage/GoogleDrive-tylermc@stanford.edu/My Drive/Applications/Scripps/doi_10_7280_D1RH7Z__v20220829/flood_hazard_small.shp' 
##   using driver `ESRI Shapefile'
## Simple feature collection with 2 features and 2 fields
## Geometry type: MULTIPOLYGON
## Dimension:     XY
## Bounding box:  xmin: -117.807 ymin: 34.02632 xmax: -117.757 ymax: 34.05826
## Geodetic CRS:  NAD83

• How would you plot the flood risks?

# plot flood risks
ggplot()+
  geom_sf(data = flood_hazard_small)

ggplot()+
  geom_sf(data = flood_hazard_small, aes(fill = FLD_ZONE))

ggplot()+
  geom_sf(data = flood_hazard_small, aes(fill = FLD_ZONE))+
  geom_sf(data = pomona_small)

# join flood risks
flood_hazard_small %<>%
  st_join(pomona_small)


flood_hazard_small

## Simple feature collection with 2 features and 5 fields
## Geometry type: MULTIPOLYGON
## Dimension:     XY
## Bounding box:  xmin: -117.807 ymin: 34.02632 xmax: -117.757 ymax: 34.05826
## Geodetic CRS:  NAD83
##     FLD_AR_ID FLD_ZONE    ID   OwnerName C100yrMEAN
## 1  06037C_664        D 48083 POMONA CITY  0.3157211
## 2 06037C_5417        X    NA        <NA>         NA
##                         geometry
## 1 MULTIPOLYGON (((-117.7878 3...
## 2 MULTIPOLYGON (((-117.7572 3...

• What is a hazard?
• What is a risk?

• What are the two types of data that simple features combines?
• What are the three types of spatial “features”?

• How do we join points and polygons?
• What happens to points/polygons outside the intersection?

• How are spatial joins useful for air quality measurement?
• How are spatial joins useful for flood measurement?

The language of risks and hazards has also been applied to social issues. For example, historical redlining used “hazardous” neighborhood labels to entrench ethnoracial spatial divisions.

Los Angeles, 1939

Governments shape risks (real and perceived) that people live with.

Through risk management, states and institutions set “acceptable levels” of risks, with implications for society.

• What is an example of an “acceptable level” of risk, set by a government, institution, or group?

la_flood_risks <- read_csv("lafloodrisk_2020/lafloodrisk_2020_p_d_s_50.csv",
                           show_col_types = FALSE)


# let's look at our data
names(la_flood_risks)

##  [1] "ID"                 "APN"                "TRACT_ID"          
##  [4] "BLKGRP_ID"          "TaxRateCity"        "UseType"           
##  [7] "LotSizeSquareFeet"  "TotalAssessedValue" "C100yrMEAN"        
## [10] "P100yrMEAN"         "Q100yrMEAN"         "T100yrMEAN"        
## [13] "FEMA_ZONE"          "FEMA_SUBTITLE"      "FEMA_SFHA"         
## [16] "POP2020"            "POP20_SQMI"         "WHITE"             
## [19] "BLACK"              "AMERI_ES"           "ASIAN"             
## [22] "HAWN_PI"            "HISPANIC"           "OTHER"             
## [25] "MULT_RACE"          "MEDINC"             "SVI"               
## [28] "SOVI"               "NDI"

ggplot()+
  geom_sf(data = pomona)

pomona_small <- pomona %>%
  head(1)

st_write(pomona_small, "pomona_small.shp")

#la_verne <- st_as_sf(st_set_crs(la_verne, st_crs(flood_hazard)))

flood_hazard <- st_read("06037C_20260127/S_FLD_HAZ_AR.shp")

flood_hazard_small <- 
  st_intersection(flood_hazard, st_as_sfc((st_bbox(pomona)))) %>%
  select(FLD_AR_ID, FLD_ZONE, geometry)



st_write(flood_hazard_small, "flood_hazard_small.shp")

flood_hazard_small <- st_read("flood_hazard_small.shp")


ggplot(flood_hazard_small)+
  geom_sf(aes(fill = FLD_ZONE))+
#  lims(x = c(117.81, 117.76), y = c(34.025, 34.060))+
  geom_sf(data = pomona)

• Points!
• Lines!
• Polygons!

point1 <- st_point(x = c(0, 1))

point2 <- st_point(x = c(2, 3))

point3 <- st_point(x = c(0, 4))


# multiple points
multipoint <- st_multipoint(x = matrix(data = c(1, 1, 0, 0, 2, 0), 
                                 ncol = 2, 
                                 byrow = TRUE))


# multiple polygons
multipolygon <- st_multipolygon(x = list(list(matrix(data = c(0, 1, 0, 0, 1, 2, 0 ,1), 
                                      ncol = 2,
                                      byrow = TRUE)),
                                    list(matrix(data = c(1, 0, 0, 0, 1, 2, 1 ,0), 
                                      ncol = 2,
                                      byrow = TRUE))))

library(sf)

# load a simple features object
nc <- st_read(system.file("shape/nc.shp", package="sf"))

• You might notice “CRS” in sf objects
• What’s this? Coordinate Reference System
• Projects points on global (or North American) maps

# look at nc
head(nc)

• What is our object?

class(nc)

• Can we plot it?

# plot entire dataset
plot(nc)

# plot one column
plot(nc$AREA)

• What if we wanted to plot one county?

• Mapping with ggplot2
• We can use geom_sf() to plot simple features

library(ggplot2)

ggplot(nc)+
  geom_sf()

• In groups: Can you plot nc using ggplot() and geom_sf()?
• Can you plot just the counties with multiple polygons?
• Try changing color (fill), line width (lwd), transparency (alpha)

• How would we overlay two types of shapefiles?

• For example, counties and roads in Los Angeles County:

# gather county shapefiles
#water_la <- linear_water(state = "CA", county = "Los Angeles")

# plot counties 
#ggplot(counties)+
#  geom_sf()+
#  geom_sf(data = water_la)

# it may help to limit data to los angeles county
#county_la <- counties %>%
#  filter(COUNTYFP == "037")


#tracts_la <- tracts(county = "Los Angeles", state = "California")

# plot la county first 
#ggplot(tracts_la)+
#  geom_sf()+
#  geom_sf(data = water_la)

# get state of california shape
#ca <- states() %>%
#  filter(NAME == "California")

• Air pollution reduces lifespans by approximately 2 years, on average, globally
• 97% of world population lives in areas with “unsafe” levels of particulate matter (e.g. PM2.5)
• In the U.S., particulates are estimated to account for 100,000 deaths annually (from pulmonary and cardiac disease)

• Gains in recent decades in the U.S. are being undone by wildfires

In Los Angeles, researchers have estimate a different zone of potential flooding.

To compare official risks to experienced hazards, we’ll need to combine data from different sources. By combining data on flooding risks and spatial demographics, we can investigate what this risk management means for communities.

First, we gather data on Los Angeles:

This class will use R. We will use the following packages:

install.packages("dplyr")
install.packages("readr")
instll.packages("tigris")
install.packages("ggplot2")
install.packages("sf")
install.packages("magrittr")

library(dplyr)
library(readr)
library(tigris)
library(ggplot2)
library(sf)
library(magrittr)

Choose a neighborhood or region in Los Angeles, and combine building and flood risk data.

• Combine your map with the Los Angeles flooding data. Look at who experiences flooding in the following categories:
• inside official and unofficial floodplains
• inside official floodplains, outside unofficial floodplains
• outside official floodplains, inside unofficial floodplains
• outside official and unofficial floodplains
• Who experiences which risks? What can you conclude about what will happen in the event of a flood?
• First, draw how you might communicate these findings. Then try to visually display them using R. What changes?