Project Overview 

 

Building Resilience to Wildfires in the Western United States: Predictive Modeling in a Coupled Climate and Human System

​

Fire is a natural and ecologically crucial process that is a major player in Earth’s terrestrial carbon budget. The rate and nature of future climate change is highly uncertain due in part to uncertainty in how future fire activity will affect the carbon storage capacity of Earth’s continents. Fire has also emerged in recent decades as an increasingly important influence on human safety and health. Large wildfires in California, the Amazon, and Australia in the past two years killed over 100 people, destroyed tens of thousands of structures, and exposed millions to toxic particulate matter. Global wildfire models are at odds with each other about how global fire activity will change in the future, and they even disagree on how fire has changed in the past. We must improve our understanding of how fire responds to changes in climate, land cover, and human activities (e.g., ignitions, suppression efforts) to improve our understanding of future fire impacts on climate, ecosystems, and human safety and health. For my PhD research under the guidance of Dr. Park Williams, I propose to develop a predictive wildfire model in which wildfire activity is driven by the complex interaction between climate, vegetation, and humans. The study region will be the western United States, which has particularly good observational data and has experienced a large increase in wildfire activity over the past several decades.

 

While fire is widely studied in the western United States, much of the work has focused exclusively on the impacts of climate and has been limited by a focus on only the largest fires. In cases when smaller fires were considered, a constraint was the limited accuracy of the fire sizes and locations. This research will combine government records of wildfire occurrence with NASA/USGS Landsat satellite data available through Google Earth Engine to create a new database documenting with unprecedented accuracy all western US fires larger than 200 hectares (ha) since 1984, expanding the Landsat-based record of very larger (>404 ha) wildfires compiled US Forest Services’ Monitoring Trends in Burn Severity (MTBS) program.

 

I will use the new database to explore the drivers of variability and change in regional wildfire area. Previous work found that as aridity increases, burned area increases exponentially. However, this relationship cannot hold forever, as the amount of vegetation available will eventually limit how much fire can continue burning. I will investigate the nature of this exponential relationship and determine when and why it should be expected to break down. I will further assess how climate effects on wildfire are modulated by land-cover and human variables (e.g., fuel abundance, human population density) and will explore a range of quantitative techniques to model these complex relationships. This modeling work will include machine learning approaches (e.g., recurrent neural networks). To assess the complex effects of human behavior and culture on wildfire, I will conduct a study specifically on areas on either side of the US-Mexico border, where climate is essentially identical but human relationships with wildfire and vegetation are quite different.

 

My proposed research aims to expand NASA’s sustained efforts to bring satellite data and products to the public through the NASA Applied Sciences Division and the NASA Disasters Team. With an improved understanding of climate-fire interactions, the created database and developed fire statistical model will bring new insights into disaster response and research, and help NASA’s domestic and global partners make informed decisions on fire management. With new understanding of climate-fire interactions, I look forward to collaborating with NASA and its partners to guide new environmentally sustainable policies and fulfill the agency’s goal to safeguard and improve life on Earth through space technologies.