UCLA

As global warming and urban development continue, areas with warm and dry climates have become more prone to wildfires in recent years. Not only do wildfires cause instant public health issues and financial losses, but they also result in long-term damage in the local biosphere. To predict and mitigate wildfires, it is important to apply statistical methods to perform a comprehensive analysis on the occurrences of wildfires to understand the trends and correlations. Los Angeles County is an important and typical area to analyze, as it has a dry, warm Mediterranean climate, and is surrounded by hills and mountains. Those mountains are usually covered by deciduous trees and shrubs, which contributed to dry, fallen leaves that are prone to get ignited by winds in winter. Moreover, Los Angeles County is densely populated and has limited land for urban developments. As a result, many regions near foothills have been urbanized, including cities or communities of Pacific Palisades, Brentwood, West Hollywood, Los Feliz, La Canada Flintridge, Altadena, Sierra Madre, and Arcadia. As these parts of the county sit just south of either the Santa Monica Mountains or the San Gabriel Mountains, wildfires that occur in these mountain forests are a nonnegligible threat in winter. The Eaton Fire, for instance, caused severe loss in Altadena, Sierra Madre, and Arcadia, while the Palisades Fire had very negative socioeconomical influences on Pacific Palisades and Brentwood. Both wildfires occurred in January 2025, when precipitation was less than normal and a major Santa Ana wind event occurred, which had synergistic effects that dramatically elevated the risk of wildfires. To further understand the occurrences of individual wildfires in Los Angeles County as well as their relationships, I use statistical approaches including Stoyan Grabarnik (SG) estimation, Moran’s I method, Local Indicators of Spatial Association (LISA), and Social Network Analysis to form a clear and multidimensional reflection. The Stoyan Grabarnik method is an efficient and predictive way that can help analyze the likelihood of wildfires, which is indispensable for fire prediction and prevention. Moran’s I and LISA are more focused on determination of occurrence distribution, which facilitates the identification of high-risk areas in a period. Social Network representations provide us with information about relationships among wildfires based on different parameters and are thus helpful for understanding the mechanism and interaction of wildfires.