Using vegetation structure estimates derived from multi-source remote sensing to predict dynamics of a semi-arid ecosystem in the western US

Abstract:

The distribution of species and vegetation types across the western US are expected to shift in response to climate change. Previous studies have documented the change in fire regime and the increasing fire-invasive grass cycle occurring in the western U.S. The change in vegetation structure due to climate change and invasive species alters the fuel load, making these ecosystems vulnerable to high-severity fire. Synergistic remote sensing data, such as hyperspectral data and high-resolution lidar, can be leveraged to capture the composition and structural variability of short-statured semiarid vegetation (e.g. sagebrush, annual grasses). We use a random-forests based fusion technique to integrate multi-source airborne data (hyperspectral and LiDAR) and generate spatially-explicit estimates of vegetation composition and structure (biomass, cover, density, height, LAI) and associated uncertainty across a climate and elevation gradient in southern Idaho. The results will be used to initialize an individual-based terrestrial biosphere model (Ecosystem Demography, ED2) and estimate structural dynamics under future scenarios. This study will provide a basis for understanding feedback mechanisms related to changing climate conditions, fire regimes and patterns of non-native plant invasion. The forthcoming field and remote sensing collection campaigns are also designed for parameterizing a dryland shrub plant functional type in the ED2 model.