California Wildfires
Reid Brawer
Data Analysis
I analyzed various datasets regarding California wildfires in R to both improve my data analytic skills and coding skills. Through researching wildfires I came acorss various potential causes, and I wanted to touch on the human impacts to highlight what behaviors would have the best chance of reducing wildfire damage.
This map uses data from Cal Fire from the 2020 California wildfire season. The dark red areas had more total acres burned while the light pink areas had fewer and the gray areas had none. This acres burned data was used a basis of comparison for causes of wildfires.
The graph (left) shows the total acres burned in prescribed burns in California on the x-axis and the total number of fires in California on the y-axis. There is a possitive correlation between the number of acres burned in prescribed burns and total number of fires. However, the number of fires does not speak to total damage. The graph on the right highlights the total acres burned in wildfires and proved to be a better measure of prescribed burn effectivness.
The graph (right) shows the total acres burned in prescribed burns in California on the x-axis and the total acres burned in California wildfires on the y-axis. Although there are not enough years included to make conclusive descisions, we can see that the number of acres burned in prescribed burns impacts the following year. This pattern continues as 2020 was one of the worst fire seasons in history.
The data for these charts came from predictiveservices.gov
The graph on the left was created using data from recreation.gov that included millions of campsite reservations from all across the country. I focused this in on California and combined it with Cal Fire data that includes acres burned by county. Each point on the graph represents one county in California while larger sized and lighter colored points represent more acres burned. The x-axis and y-axis haved been scaled proportionatly on a scale of 100x100 for easier comparison. The main take away here is that some of the counties with the most fires also had a lot of campers, but many of the counties with lots of fires had few campers and likewise many counties with many campers had few fires.
This data set may be incomplete as it only includes government campsite reservations and not reservations from sites like Airbnb, Hipcamp, or ones made in person or not made at all. However, it is a large enough data set to give an accurate representation of camper traffic in California.
Annotated Bibliography
Related Projects
Aldhous, Peter. “How A Booming Population And Climate Change Made California’s Wildfires Worse Than Ever.” BuzzFeedNews, July 2018, https://www.buzzfeednews.com/article/peteraldhous/california-wildfires-people-climate
Buzz feed did a complete project on wildfires across the country in R with various visuals and analysis similar to my project scope. This is a good source for datasets and visual ideas. It establishes a straightforward approach that compares California to the rest of the country with especially interesting graphs comparing human and non-human caused fires.
Williams, Sky B.T. “Wildfire Destruction — A Random Forest Classification of Forest Fires.” Towards Data Science, Feb. 2018, https://towardsdatascience.com/wildfire-destruction-a-random-forest-classification-of-forest-fires-e08070230276
Sky Williams is a data scientist that used both air temperature data and over a century of fire data to make a prediction model of a fire's potential size. Her model gives more attention to small fires that could potentially be false alarms to mitigate the potential for fires with larger potential to go unnoticed.
Stewart, Jessica, “Artist Uses Fire and Smoke to Create Incredible Paintings of Birds.” My Modern Met, May 2017, https://mymodernmet.com/steven-spazuk-fire-painting/
Steven Spazuk is a modern artist that utilizes the impressionist technique of fumage in which candles are used to apply soot to a canvas or paper. His draw to creating are with fire is that “fire consumes, warms, and illuminates but can also bring pain and death; thus its symbolic meaning varies wildly”. He goes on to explain that he mostly uses it around living things, mainly birds, to discuss “life’s fragility”.
Contextualization: air quality, fire causes, medical concerns
V. Leone & R. Lovreglio, “Human fire causes: a challenge for modeling” University of Basilicata, Potenza, Italy, https://www.researchgate.net/profile/Emilio_Chuvieco/publication/44157960_The_GOFCGOLD-Fire_Program_a_mechanism_for_international_coordination/links/54aea1c10cf2b48e8ed45417/The-GOFC-GOLD-Fire-Program-a-mechanism-for-international-coordination.pdf#page=90
Pierre-Louis, Kendra, and John Schwartz, “Why Does California Have So Many Wildfires?” The New York Times, 3 Dec. 2020, https://www.nytimes.com/article/why-does-california-have-wildfires.html
This NY Times article gives context to what is causing the fires, and why they are getting worse every year. Their sections of causes may also be good starting places for data analysis.
Shapiro, Ari, “Fire Expert On How Indigenous Land Management Could Help With Fires In California” NPR All Things Considered, Oct. 2020, https://www.npr.org/2020/10/13/923377261/fire-expert-on-how-indigenous-land-management-could-help-with-fires-in-californi
This gives context to the history of indigenous burns and how society has strayed from a “for the common good” mindset for a more individual focused mindset.
Jordan, Rob “5 Questions: Researchers discuss wildfires’ health impacts.” Stanford Medicine, Aug 2020, https://med.stanford.edu/news/all-news/2020/08/researchers-discuss-health-impacts-of-wildfires.html
Five Stanford medical professionals gave their opinions about the harms of polluted air quality as well as ways of reducing the impact of particles such as PM2.5 with the use of cloth or N-95 masks.
Miller, Rebecca K., Christopher B. Field, and Katherine J. Mach, “Barriers and enablers for prescribed burns for wildfire management in California” nature sustainability, Vol 3, Feb. 2020, https://www.nature.com/articles/s41893-019-0451-7.pdf
This study mainly addresses barriers to executing prescribed burns in California. The information comes from a series of interviews conducted with all stakeholders involved from landowners, to random citizens, to officials on the relevant boards. In the planning phase burns can be halted from “negative public opinion” and “liability concerns” Prescribed burns can only happen if the landowner (often private) allows it to, funding is raised, the air quality board gives approval, and a certified burn expert is present while it happens. The primary three barriers have been identified as “risks, limited resources, and regulations”
United States Environmental Protection Agency, https://www.epa.gov/ghgemissions/inventory-us-greenhouse-gas-emissions-and-sinks
The emissions statistics that I use in the rational section came from the EPA. Instead of breaking down the data into industry and household use they also include “transportation”, “electricity”, and “agriculture”. Agriculture seems like it should belong to “industry” while transportation and electricity would be split between industrial and household use.
Datasets
Recreation.gov, https://www.recreation.gov/use-our-data . Accessed 20 Sept. 2020
This is a dataset that contains every campsite reservation in the country over the past five or so years. There are over three million reservations recorded here. The campsite reservations do not account for all campsites used in a given year. Data from private sites is most likely not shared with recreation.gov and some sites do not require online reservations.
CalFire, https://www.fire.ca.gov/stats-events/ . Accessed 20 Sept. 2020
Cal Fire has many datasets that can be useful. Specifically, in the Redbook section of the incident reports there are several tables that I can convert to a csv file. They also have a mapdataall csv file in incident reports that has basic information on every fire in California. The benefit of this is that it is fully updated whereas the Redbooks only start in 2018.
https://www.predictiveservices.nifc.gov/intelligence/2019_statssumm/wildfire_charts_tables19.pdf
This dataset contains both number of prescribes burns conducted across the U.S. from 2009-2019 as well as the number of acres burned in total. The data is separated into ten regions and was gathered through collaboration from various fire agencies. This data will be contextualized with the acres burned data.
Approaches to Data Analysis and Visualization
DuBois, Luke R. “Insightful Human Portraits Made from Data | R. Luke DuBois.” YouTube, YouTube, 19 May 2016, www.youtube.com/watch?v=9kBKQS7J7xI.
Luke R. Dubois finds interesting ways to visualize or experience events through data. He maintains the most important human connections behind any piece and believes “turning us into statistics is something that is done at our peril.”
Dixon, Michael, “Types of predictive analytics models and how they work” Selerity, Dec. 2019, https://seleritysas.com/blog/2019/12/12/types-of-predictive-analytics-models-and-how-they-work/
This article covers a brief overview of various types of predictive analytics as well as providing tips on terms and concepts. Based on these definitions I would most likely pursue a time series model which forecasts future numerical values for some data points based on historical data.
Adams, Amy. “Reflections on the California Wildfires.” Stanford News, 28 Nov. 2018, https://news.stanford.edu/2018/11/28/reflections-california-wildfires/.
This is a slightly dated, but still relevant article about the causes of wildfires and their predicted continual progression. Experts speak to data loss that occurs from not recording fires smaller than 10 acres in many counties across the state. They suggest we must find exact locations of fires with longitude and latitude (not just estimates), record how they happen, and how often they happen.
Zonin, Alessandro. “Emoji Speak Louder than Words.” Social Network Analysis, 2 Sept. 2019, https://alessandrozonin.wordpress.com/2019/09/02/emoji-speak-louder-than-words/.
This is a short article with some visual examples with data taken from Socioviz. The article covers how emojis can be used to convey sentiments or to make comparisons between two or more similar things. In my case this could be California and Oregon or Argentina.
Benney, Marnie, and Pete Kistler, “Creative Tools to Generate AI Art”, 2019, https://aiartists.org/ai-generated-art-tools
This site offers a compiled list with summaries of the uses of various machine learning tools. TensorFlow and Deep Dream may be the most useful for training before and after photos of fires.
Cortez, Paulo, and Anìbal Morai, “A Data Mining Approach to Predict Forest Fires using Meteorological Data”, https://repositorium.sdum.uminho.pt/bitstream/1822/8039/1/fires.pdf
This study focused on real time meteorological data in Portugal to predict the size of a forest fire. The four variables they took into consideration were rain, wind, temperature, and humidity. Systems such as this have up to a 90% accuracy rate with small fires.
Lulu.com is the online book publisher I used to create and print my artist book. They source projects out to local print shops around the country to fufill custom book orders for nearly any user needs.