An Investigation of the California Wildfire Crisis

By Adhvaith Vijay(Project Lead), Danielle Goldwirth, Tara Jaigopal, Jaclyn Chiu, Angelina Kim

Embers engulf a hill behind the Bidwell Bar Bridge | September 9, 2020

A Brief Overview

Over the last few decades, more and more evidence has led us to conclude climate change is wreaking havoc on our environment. However, like many phenomena, this change is gradual and often it is difficult to see the impact of climate change on our day-to-day lives. Wildfires serve as warnings that the impact of climate change is both severe and, for many Californians, very close to home. In a wildfire risk analysis performed by the Insurance Information Institute, California leads all other states with nearly 2,000,000 at-risk properties — all the while enforcing some of the strictest national standards for the construction of fire-resistant properties.

No amount of policy will truly put an end to the billions of dollars lost every year due to wildfires. Mother Nature stops for no one, and many of the same homes built under strict protocols have been burned to the ground.

Both anthropogenic and environmental factors are shaping the fires we see today. Our ability to study fires using satellite data points to a growing trend in wildfire frequency and severity especially along the West Coast of the United States. As such, we sought to analyze and explain many of the wildfire trends across California at both a county and regional level.

Historical Growth

In order to best represent geospatial data, we decided to overlay a scatterplot of latitude and longitude coordinates of major wildfires onto a map of California. Each circle represents a wildfire incident that has been color-coded and sized up proportionately to accurately represent the number of acres burned.

Run the animation and hover over each circle for information on specific incidents

After running the animation a couple of trends appear. Since the onset of 2017, there have been far more damaging fires across the state with coastal and forested areas hit hardest. Before 2018, fires were far more equally distributed in both size and frequency all over California. However, from 2018 onwards, there has been a rampant increase in deadly wildfires across Northern California compared to other regions in the state. 2020 alone has seen the most destructive fires in modern history with the August Complex Fire and LNU Lightning Complex of Northern California burning an estimated 1.4 million acres to date. The August Complex Fire (depicted as the large, bright yellow circle in the 2020 pane), was extinguished only last month in November after an unprecedented cost of $264.1 million dollars.

One striking oddity in the trend of wildfire growth is the plummet of wildfires in 2019. With the lowest levels since 2004, the destruction done in 2019 amounted to only 260,000 acres — a mere 6% of the damage done in 2020. However, experts vehemently believe this trend will not change any long-term patterns and was simply an irregularity due to unusually heavy precipitation.

From this animation, it is clear that the most destructive fires occur in more heavily-forested areas — centered around Northern California. However, can the same be said about the frequency of wildfire incidents?

Hover over each bar for year-specific information

In this bar chart, broken up by region (NorCal and SoCal), it is clear that while the number of incidents in Southern California is increasing at a slight linear rate, the number of incidents in Northern California appears to be growing exponentially year-on-year. Reasons for this stark division can be attributed to the presence of more vegetation in NorCal which acts as a veritable timebomb waiting to ignite during the dry season.

With the exception of 2019 — which was an anomalous year for wildfires — there continues to be an overarching trend of both increased frequency and severity of wildfires, especially in Northern California.

County-Level Analysis

The figure below is a choropleth map — a thematic map colored in proportion to a specific statistical variable. In this case, this choropleth has been partitioned to visualize the cumulative acres burned since 2003 on a county-by-county basis. As expected, Northern California trends towards more land destruction.

Hover over choropleth for information on specific counties

Heavily damaged areas, depicted in purple and black, are also among the most forested regions in California. The San Joaquin Valley, an inland area constituting much of the Central Valley sports a more Mediterranean climate and is far less forested. This, in turn, accounts for fewer acres burned compared to the more northerly Sacramento Valley.

Source: Drought Monitor NOAA Climate.gov

Prolonged droughts and the rapid drying of vegetation accounts for the increase in wildfires. However, variability in temperature and precipitation are often the real instigators. Years of drought in California have traditionally been followed by very wet weather leaving behind vegetation that turns into fuel for wildfires. Cyclical weather patterns along with strong, warm winds prime wildfires for destruction. As seen above, Northern California is the most victim to moderate-severe drought compared to the rest of the state. This dichotomy might be explained by Southern California’s access to the Colorado River Aqueduct, which supplies more than a billion gallons of water a day.

Hover over each bar for county-specific information

The bar graph shown above reveals that Lake County, Shasta County, and Trinity county suffer the brunt of wildfire damage across California. The reason these three counties lead the pack is that they all encompass national parks and are more heavily forested than any other region in the state.

Coincidentally the Shasta-Trinity National Forests — the largest National Forest in California was also the site of the August Complex Fire — the most destructive wildfire in modern California history.

Across California, many minor incidents occur every day and are often contained without loss of property or life. Even with fewer wildfire incidents relative to other counties, Lake County is still responsible for more than 2 million acres worth of destruction over the last two decades.

To better understand the impact wildfires have on private property we drew from data provided by the insurance industry. The below choropleth map serves to highlight figures provided by Verisk Analytics concerning the percentage of California households at high to extreme risk for wildfires.

Hover over choropleth for risk-assessment of various counties

Counties in blue indicate less than 20% of households are threatened, whereas red counties indicate a larger percentage of households at severe risk for wildfires.

Alpine, Trinity, Tuolumne, and Mariposa counties had the highest concentration of severely at-risk households in Northern California. The reason for this disparity is because households in Northern counties are dispersed at the edge of forested areas and are often directly in the way of wildfires. In many instances, stray embers carried by the wind make their way to rain gutters, bursting into flames and engulfing hundreds of properties.

Investigating Major Wildfire Incidents

After investigating wildfire impact on a county-by-county level we shifted to analyzing specific wildfire incidents. The bar chart below shows the top wildfires incident in terms of acres burned.

Hover over each bar for incident-specific information

Among the wildfires listed, the top six all took place in Northern California between 2018 and 2020. Furthermore, the August Complex — the single-largest wildfire and the largest fire complex in recorded California history — is more than twice as large as any other recorded incident. Blazing across Mendocino, Humboldt, Trinity, Tehama, Glenn, Lake, and Colusa counties, the August Complex Fire has been coined as a “gigafire” — a term never before used in California history to signify a blaze that burns at least a million acres.

From this subset of the Top 10 Wildfires in California history, the average duration that wildfires burned for was a whopping 59.5 days. Naturally, the question arises: What Makes Wildfires So Hard to Put Out? Wildfires behave in many ways like a combustion-powered hurricane. By channeling air and fuel upward, forests often spontaneously combust without actually coming into contact with flames. This combination of explosive growth and hellish conditions often renders fire-support teams from both the ground and air useless. In addition, since fires are most prevalent during the dry season, a lack of humidity leaves humans without the aid of Mother Nature.

Seasonal Disparity

For many Californian natives, fire season has always been marked by the end of Summer well into late Fall — making up the months of August, September, and October. To better understand the annual pattern of wildfires, we created a boxplot of Acres Burned across wildfire incidents from 2003–2020. By grouping wildfires by the month they started, we aimed to recognize and explain the seasonal disparity in damage.

Hover over each point for incident-specific information

Right from the start, it’s clear that wildfires dominate the autumn months. The peak of fire season spans from late July through September and is marked by radical wildfires destroying twice the number of acres compared to incidents in the spring and early summer.

Interestingly, there likely exists two distinct fire seasons in California. A study by University of California authors found that California cycles between what is called the Summer Fire Season, and the Santa Ana Fire Season.

Characterized by dry winds blowing towards the coast from the interior, the Santa Ana Fire Season occurs from October through April — striking more developed areas and inflicting more economic damage.

The Summer Fire Season, however, can take place anywhere in the state and often impacts remote/wild areas — as was the case with the August Complex Fire which engulfed the Mendocino, Six Rivers, and Shasta-Trinity national forests. Making up the rest of the calendar year, the Summer Fire Season is between June and September.

With the Santa Ana Fire Season inflicting more economic damage, the Summer Fire Season accounts for the most land destruction with millions of acres destroyed every year.

In order to prove that there exist two distinct fire seasons in California, we tried to replicate the results made by University of California researchers.

We decided to apply a form of unsupervised machine learning known as K-means Clustering to see if the “clusters” of months we identify align with the months outlined in the University of California paper. The idea behind this was to apply K-means clustering on key metrics that might reveal if a set of months truly deserves to belong in their own distinct fire season. As per the findings made by University of California researchers, we wanted to investigate metrics that are most representative of fires during the Summer Fire Season and Santa Ana Fire Season. For example, University of California researchers observed that ‘Summer Fires’ are more inclined to burn more slowly, while ‘Santa Ana Fires’ tend to burn along the coast. As such we take into account the average fire duration by month as well as the percent of fires located in coastal counties. Other metrics include the average acres burned and total wildfires incidents on a month-by-month basis.

To ensure that the algorithm weights each metric with the same relative importance during the clustering process, we normalized each column between 0 and 1 using the MinMaxScaler function in Python’s Scikit Learn library. After normalizing my metrics, we were able to use the K-means algorithm to partition my data into clusters as seen below.

There appear to be two distinct clusters based on this dataframe. Cluster A, identified by ‘1’ exists from June through September. Cluster B, identified by ‘0’ exists from October through April.

Cluster A is striking in that there are more wildfire incidents, a greater number of acres burned on average, and longer wildfires. These results share similarities to the ‘Summer Fire Season’ description outlined by University of California researchers. In the same vein, fires in Cluster B tend to take place along coastal areas and burn fast — characteristic of the Santa Ana Fire Season.

Hover over each point for incident-specific information

To better visualize the disparity between fire seasons we applied a color scale to the boxplot from earlier. Months in orange constitute the Santa Ana Fire Season, whereas months in red represent the more destructive Summer Fire Season.

Human Activity

There are a variety of ways to gauge which incidents can be described as the worst wildfires in California history. Different metrics include size (acres burned), deadliness (lives lost), and destruction (infrastructure destroyed). In order to better understand the part humans play in instigating the worst wildfires in California history we look at the following subsets: the top 20 largest wildfires, the top 20 deadliest wildfires, and the top 20 most destructive wildfires.

While wildfires have historically been sparked by natural conditions, the increasing amount of fires caused by human activity has become of great concern. In the figure above, human-related activities and human-infrastructure (i.e. powerlines) emerge as the largest known motivators of the worst wildfire incidents in California history.

Whoever said the small things don’t matter has never seen a match start a wildfire.

-Beau Taplin

Human causes of destructive blazes are well documented. From sparks generated on the rim of a flat tire (2018 Carr fire) to a hunter’s signal flare (2003 Cedar fire), humans have played a role in some of the most destructive fires across California. As recently as September 2020, the El Dorado Fire — which went on to claim one life and burn 22,744 acres — was ignited by a pyrotechnic device at a gender reveal party.

Nonetheless, the question still remains: which wildfires can be called the most severe incidents in modern California history? To answer this question we created a bipartite graph of all wildfire incidents that map to 2+ top 20 subsets.

Mapped to each of the top 20 subsets, there are certain incidents that are among the largest, deadliest, and most destructive fires of all time. Highlighted in crimson above, these incidents are the North Complex, Carr, and Cedar Fires. In essence, we can objectively call these three incidents the worst wildfires in modern California history.

Looking Outwards

Climate change is a driving factor in the increased severity of wildfires in the modern era. Rising temperatures have created drier conditions, milder winters, and longer fire seasons. Owing in part to variable weather patterns, wildfires are burning larger swathes of California than ever before — turning entire forests into matchboxes. This in turn creates a cyclical pattern, with emissions from fires adding more carbon to the atmosphere. As seen above, since 1985 climate change has played an increasing role in aggravating and compounding the wildfire crisis. Especially in the last decade, climate change accounts for nearly 50% of the acres burned as a direct consequence of forest fires.

As we assess the devastating impacts of fires on the environment and human lives, it is crucial to acknowledge our outsized role in the escalation of this crisis and shift our focus towards the prevention of these tragedies.

Visit our GitHub for more details about our analysis process.

Get the Medium app

A button that says 'Download on the App Store', and if clicked it will lead you to the iOS App store
A button that says 'Get it on, Google Play', and if clicked it will lead you to the Google Play store