Utilizing GEDI data to classify changes in forest structure due to disturbance in California

Forest disturbances, such as forest fires, drought-induced die-off, and timber harvests, are rapidly increasing and evolving in the state of California, threatening the state's natural resources. In addition, these disturbances are mass carbon source events, decreasing carbon stocks and increasing atmospheric carbon. With this in mind, California has set climate goals that include decreasing greenhouse gas emissions to 40% below what they were in 1990. To achieve this goal, it is vital to quantify the disturbances occurring in California to better constrain estimates of carbon emissions from forest disturbance. To address this need, we use spaceborne lidar data from NASA's Global Ecosystem Dynamics Investigation (GEDI) to classify disturbance based on changes to forest structure. First, we identified disturbance polygons in California affected by fire, harvest, or drought by cross-referencing datasets published by state and federal agencies. Additionally, polygons of drought-induced forest die-off were hand-drawn by looking at high-resolution satellite imagery. The total set of polygons includes 26 forest fires, 12 timber harvests, and 18 drought-induced die-offs. Overlaying these polygons with the GEDI data yielded a total of 236,185 points in affected regions and 107,241 points in the control set. From here, we applied classification algorithms including logistic regression, random forests, and support vector machines to relative height metrics from GEDI and surface reflectance from Landsat 8 to identify each type of disturbance. This work enhances our ability to use remote sensing to quantify the extent of forest disturbance and associated impacts on forest structure and carbon emissions in California.