Allison Kelly Thesis Defense
California’s North Coast hosts a Mediterranean climate with diverse topography, strong climatic gradients of temperature and precipitation, and complex species composition. Unfortunately, historical fire suppression and climate change have led to a significant increase in wildfire frequency and intensity in this unique bioregion. As such, quantifying forest structure to assess changing wildfire risk factors is critical as California moves beyond the initial stages of short-term disaster recovery and begins to develop mitigation, land management, and resource allocation strategies. Remote sensing offers a valuable method for understanding forest structure and lessening the need for costly field inventory campaigns. Here, we aim to better understand crown fire potential in Northern California oak woodlands by evaluating and estimating forest canopy structure via canopy fuel parameters at plot scale using unoccupied aerial system structure from motion (UAS-SfM) methods. In Summer 2020, images of plot-level forest structure (via Aerial Laser Scanning (ALS) and multispectral UAS) and ground-based measurements (leaf area index; LAI and canopy cover) were collected at 44 oak woodland plots (20 x 20 m) in Sonoma County, California. Using these data, we estimated total canopy cover (%), mean canopy height (m), canopy base height (m), canopy bulk density (kg/m3), tree density (number of trees/plot), and topographic elevation (m^2). This study used a tree point cloud-based segmentation method developed by Li et al. (2012) to identify individual trees. From the UAS-SfM point clouds, 68% of trees were correctly estimated. Heights for the correctly detected trees were compared to field height measurements, yielding a R2 of 0.69 and RMSE of 5.1 m. When comparing remote sensing approaches (ALS, UAS) to each other and to traditional ground-based techniques, canopy cover resulted in R2 of 0.79 and RMSE of 16.49%. When comparing UAS-derived to field-derived canopy cover, results were R2 of 0.89 and RMSE of 5.56%. The canopy base height estimate, when compared to field data, had a weak correlation (R2 = 0.33, RMSE = 5.08 m). Processing of UAS via SfM, opposed to light detection and ranging (LiDAR) technology via ALS, may offer an affordable option for rapidly and accurately estimate canopy fuel metrics such as total canopy cover, mean canopy height and tree density. However, the UAS was not able to estimate canopy bulk density and canopy base height due to its inability to penetrate dense canopies. Nevertheless, our research shows that an UAS-SfM approach may be an affordable solution for land managers to accurately quantify important canopy fuel parameters to evaluate fuels management and use as input parameters to fire behavior modeling.