The 2021 Bootleg Fire: A Hydrological Perspective through Remote Sensing and Machine Learning

Wildfires are increasing in frequency and intensity, impacting ecological systems and biodiversity. The aftermath alters hydrological cycles in affected regions, causing soil degradation through the combustion of organic matter and reducing vegetation vitality, which disrupts the natural water balance and increases the risk of erosion. Remote sensing technologies provide insights into these hydrological effects, aiding in watershed modeling and management. The 2021 Bootleg Fire, which occurred in the Fremont-Winema National Forest located in Southern Oregon, serves as a case study; it raged from July 6 until August 15, consuming over 167,000 hectares of forestland. In this research, we assess the Bootleg Fire's hydrological impact on soil and vegetation. This assessment is carried out by classifying land cover and evaluating changes in pre- and post-fire normalized difference vegetation index (NDVI) for vegetation-related hydrological impacts, along with Differenced Normalized Burn Ratio (dNBR) for burn severity to gauge hydrological impacts on soil and vegetation. Utilizing Google Earth Engine, we obtained time-series Landsat 8 surface reflectance imagery for pre- and post-fire conditions, maintaining a cloud cover threshold below 2% for data accuracy. Subsequent analysis was conducted in Matlab R2023a, employing machine learning algorithms including artificial neural networks and support vector machine to classify land cover into water, barren or impervious, grassland, shrubland, forested areas, and densely forested areas. Our analysis underscores the wildfire's role as an ecological catalyst, promoting grasslands and shrublands at the expense of tree-dominated areas, which directly correlates with shifts in soil moisture levels. Areas with higher burn severity show more pronounced changes in soil moisture, affecting the hydrological balance and influencing both immediate recovery and future resilience of soil and vegetation. This transformative effect on landscape composition points to substantial, long-term hydrological impacts on both soil moisture retention and vegetation health. The findings of this research not only bear significant implications for fire management policies and resource allocation but also serve as a foundational guide for subsequent, more focused research in understanding the intricate relationships between wildfires and hydrological disruptions.