Deciphering Active Wildfires in the Southwestern USA Using Topological Data Analysis

Abstract

The recent droughts in the American Southwest have led to increasing risks of wildfires, which pose multiple threats to the regional and national economy and security. Wildfires cause serious air quality issues during dry seasons and can increase the number of mud and landslides in any subsequent rainy seasons. However, while wildfires are often correlated with warm and dry climates, this relationship is not linear, implying that there may be other factors influencing these fires. The objective of this study was to detect and classify any nonlinear patterns in weather data by applying Topological Data Analysis (TDA) to various weather variables, such as temperature, relative humidity, and precipitation, and the five most and least intense summer fire seasons as determined by the Moderate Resolution Imaging Spectroradiometer (MODIS) active fire products. In addition to TDA, persistence diagrams and frequency plots were also used to compare fire seasons and regions in the American Southwest. Active fire seasons were more likely to have a significant correlation between the weather variables and wildfires, the Fire Weather Index (FWI) alone was not an accurate predictor for wildfires in California and Nevada, and fire weather is highly dependent upon the region and season.

A Primer on Topological Data Analysis to Support Image Analysis Tasks in Environmental Science

“…The persistence of a feature is represented by how far a persistence diagram point lies above the diagonal. We present persistence diagrams here to familiarize the reader with their use in, e.g., Kim and Vogel (2019), Tymochko et al (2020),…”

“…TDA has proven highly successful to aid in the analysis of data in a variety of applications, including neuroscience (Chung et al 2009;Gardner et al 2022), fluid dynamics (Kramár et al 2016), and cancer histology (Lawson et al 2019). In environmental science, TDA has recently shown potential to help identify atmospheric rivers (Muszynski et al 2019), detect solar flares (Deshmukh et al 2022;Sun et al 2021), identify which wildfires are active (Kim and Vogel 2019), quantify the diurnal cycle in hurricanes (Tymochko et al 2020), identify local climate zones (Sena et al 2021), detect and visualize Rossby waves (Merritt 2021), and forecast COVID-19 spread using atmospheric data (Segovia-Dominguez et al 2021). The purpose of this article is to 3 provide an intuitive introduction to TDA for the environmental science community -using a meteorological application as guiding example -and an understanding of where TDA might be applied.…”

Leveraging spatiotemporal information in meteorological image sequences: From feature engineering to neural networks

“…For example, the topological concept of persistent homology focuses on the number of connected regions, and the number of holes therein, for a varying intensity threshold in the image, which in turn allows to distinguish different types of patterns, for example, to classify the mesoscale organization of clouds (Ver Hoef et al, 2023). Persistent homology and other topological properties are emerging in several environmental science and related applications, including in the context of identifying atmospheric rivers (Muszynski et al, 2019), Rossby waves (Merritt, 2021), local climate zones (Sena et al, 2021), activity status of wildfires (Kim and Vogel, 2019), and quantifying the diurnal cycle of TCs (Tymochko et al, 2020). Generally, TDA is not used as standalone technique, but as a preprocessing step to extract important features, often to be used along with other physically interpretable features, followed by a simple machine learning algorithm, for example, support vector machines.…”

Application of Topological Data Analysis to Multi-Resolution Matching of Aerosol Optical Depth Maps

“…However, the utility of TDA in Earth science applications remains largely unexplored. In particular, to the best of our knowledge, TDA techniques have been employed in only two previous Earth science studies-identifying atmospheric river patterns (Muszynski et al, 2019) and assessing the influence of various climate variables on wildfires (Kim and Vogel, 2019). In this study, we use TDA, specifically one of its tools known as persistent homology (PH), to build a robust and reliable methodology that compares spatial patterns in aerosol optical depth (AOD) maps at different spatial resolutions based on a systematic assessment of their topology and geometry.…”