Mapping 2002-2012 Aboveground Biomass Carbon from LiDAR and Landsat Time Series across Northern Idaho, USA

Abstract:

The heavy investment by public and private land management entities in commercial off-the-shelf airborne lidar provides an optimum basis for a Carbon Monitoring System due to the known sensitivity of lidar to vegetation canopy structure. The ability to accurately map aboveground carbon pools from lidar and collocated field plot data has been demonstrated in many studies. Our goal is to upscale this biomass information, mapped at 30 m resolution, to the regional level using wall-to-wall, multi-temporal Landsat imagery. We use the LandTrendr approach to transform Landsat time series into annual maps of Brightness, Greenness, and Wetness along with annual change estimates of these same tasseled cap indices. These, along with ancillary layers of canopy height (e.g., GLAS-derived), topography (e.g., insolation), and climate (e.g., mean annual precipitation) are used to predict 2002-2012 aboveground carbon annually across the northern half of Idaho, USA. Ecoregion-specific models are developed to impute aboveground biomass and forest type beneath a forest/non-forest mask. Annual maps are then summarized at the county-level and compared to publically available Forest Inventory and Analysis estimates for Monitoring, Reporting and Verification.