COS 123-6 - Changes in tree demography and health-status explain agreement between Landsat time series and forest structure at disturbed NEON sites

Abstract: Disturbances have variable effects on forest structure, ranging from minor damage of leaves or roots to stand-replacing mortality events. Mapping the full range of disturbance effects and how they restructure growing conditions for surviving vegetation is challenging. Structural changes that alter the quantity and arrangement of leafy tissues should be detectable with optical remote sensing, yet efforts to characterize mild to moderate forest disturbances with Landsat data have produced puzzling results. Some disturbed forest systems show strong agreement between interannual changes in canopy reflectance and ground-based measures of wood-growth, while other systems show poor agreement. These idiosyncrasies limit our ability to predict forest structural change from remote sensing.

Here, we asked whether changes in demographic and tree health-status could explain the variable agreement between Landsat reflectance and forest structural changes that result from a range of moderate forest disturbances, including fires, mechanical harvests, diseases, and insect damage from stem borers, defoliators, and aphids. We tested this by quantifying forest structure for annual inventory plots at 13 temperate flux-tower sites from the US National Ecological Observatory Network (NEON). We modeled how transitions among demographic and health-status classes explained variation in the relationship between forest structure and reflectance. We found that models that included the amount of plot basal area in live, damaged, and dead tree classes over time explained the most variation in canopy reflectance, 30% more than simpler models. The importance of damaged tree health-status in our models suggests that standard forest inventory metrics of structure that track only live and dead demographic change may miss intermediate stages of tree decline that correlate with Landsat. Future efforts to combine Landsat time series with LiDAR metrics of forest canopy change may fill these gaps in health-status measurement and allow wall to wall prediction of forest demography.