Changes in GEDI-based measures of forest structure after large California wildfires relative to pre-fire conditions

Abstract

Forest productivity, biodiversity, and ecosystem services in California and the Western United States are closely tied to fire. However, fire regimes are shifting toward larger, more severe fires driven by factors such as high fuel loads and increased temperature and aridity. While multispectral satellite (e.g., Landsat) burn indices provide valuable insights into fire severity, they primarily capture top-of-canopy greenness, missing important sub-canopy changes in vegetation structure and residual fuels. The Global Ecosystem Dynamics Investigation (GEDI) spaceborne lidar mission provides current and consistent, near-global 3D forest structure measurements, enabling detailed assessment of forest changes from disturbances such as wildfire. This study utilized near-coincident, bitemporal pre- and post-fire GEDI on-orbit measurements to analyze structural changes across thirty-four large California wildfires (2019 to 2021). We examined twelve GEDI-based forest structure metrics representing a variety of 3D fuels properties, including forest height, low-stature fuels, biomass, canopy heterogeneity, volume and cover. Our broad goals were to: 1) assess GEDI's ability to detect structural changes in burned areas relative to control areas; and 2) in burned areas, explore relationships between forest structural change and factors such as pre-fire fuel loads, Landsat-based differenced Normalized Burn Ratio (dNBR), topographic slope, wind speed, vapor pressure deficit, evapotranspiration, and time since fire. Results showed significant structural loss in all GEDI structural metrics for burned areas relative to nearby controls. Pre-fire fuel loads measured by GEDI metrics were the strongest predictors of post-fire structural change, with linear models explaining an average of 46 % of variance. Model slopes showed increasing levels of pre-fire fuels were associated with large, significant post-fire decreases in canopy structure – that is, more fuels lead to higher wildfire severity, particularly for conifer forests of the Klamath, Cascades and Sierra Nevada ecoregions. One metric, measuring the proportion of waveform energy below 10 m height, increased significantly after fire in mixed and conifer forests due to canopy opening, which enhanced lidar penetration toward the ground. In contrast, the widely-used dNBR burn severity index was less correlated with GEDI-based forest structural change than pre-fire fuels, particularly in sub-canopy fuels, with models explaining no more than 32 % of the variance (average 19 %). GEDI overcomes key limitations of airborne lidar, including high cost, limited extent, and data latency, enabling scalable and timely assessments of wildfire impacts needed to manage fuels and track forest resilience and recovery. Further, GEDI metrics are physically-based and ecologically interpretable, providing complimentary information to multispectral burn severity indices.