How A Booming Population And Climate Change Made California’s Wildfires Worse Than Ever

Data and R code to reproduce graphics in this Jul. 28, 2018 BuzzFeed News post on wildfires in California. Supporting files are in this GitHub repository.

Data

To analyze trends in California wildfires, we used the California Department of Forestry and Fire Protection (Cal Fire) Fire Perimeters Geodatabase, which records forest fires of 10 acres or greater, brush fires of 30 acres and greater, and grass fires of 300 acres or greater. The file calfire_frap.csv is derived from this database.

To put California’s wildfires in a national context, we used the US Forest Service’s Spatial Wildfire Occurrence Data For The United States, which records wildfires across the nation from 1992 to 2015. This data is in a series of CSV files in the us_fires folder.

Cal Fire’s data on buildings destroyed by wildfires per year is in the file calfire_damage.csv.

Setting up

Required packages and color palette for major fire causes (human, natural, unknown).

# load required packages
library(dplyr)
library(readr)
library(ggplot2)
library(ggthemes)
library(scales)
library(maps)
library(mapproj)

# color palette for major fire causes
cause_pal <- c("#ffff00","#d397fc","#ffffff")

Big fires have gotten more common

To plot fires by date, we used the Cal Fire alarm date.

# load and process data
calfire <- read_csv("data/calfire_frap.csv") %>%
  mutate(cause2 = case_when(cause == 1 | cause == 17 ~ "Natural",
                            cause == 14 | is.na(cause) ~ "Unknown",
                            cause != 1 | cause != 14 | cause != 17 ~ "Human"),
         plot_date = as.Date(format(alarm_date,"2017-%m-%d")))
  
# plot template
plot_template <- ggplot(calfire, aes(y=year_)) +
  geom_hline(yintercept = seq(1950, 2017, by = 1), color = "gray", size = 0.05) +
  scale_size_area(max_size = 10, guide = FALSE) +
  scale_x_date(date_breaks = "months", date_labels = "%b") +
  scale_y_reverse(limits = c(2017,1950), breaks = c(2010,1990,1970,1950)) +
  xlab("") +
  ylab("") +
  theme_hc(bgcolor = "darkunica", base_size = 20, base_family = "ProximaNova-Semibold") +
  theme(axis.text = element_text(color = "#ffffff"))

plot_template +
  geom_point(aes(size=gis_acres, x=plot_date), color="#ffa500", alpha=0.7)