A Systems Engineering Approach to Estimating Uncertainty in Above-Ground Biomass (AGB) Derived from Remote-Sensing Data

ABSTRACT We integrate systems of measurement and modeling to improve estimation of uncertainties in above‐ground biomass (AGB) derived from remote sensing. The outcome provides a unified starting point for the climate‐change carbon community to assess uncertainty and sensitivity data and methodologies, and ultimately supports decision‐making about which missions and instruments to develop for a desired cost/benefit ratio. Initial results include fusion of remote‐sensing techniques (e.g., radar and lidar), uncertainties associated with measurement and modeling, and the impact of potential uncertainty correlations across aggregated unit areas. Biomass uncertainty estimates are presented at the single‐hectare level for the forestlands of California. Using a forest biomass map of California, we calculate changes in variance (e.g., 2 orders of magnitude) as a function of uncertainty correlation assumptions, with correlations extending to spatial scales up to 100 km. Using a variogram formalism to derive the correlation shape and magnitude, we show that the estimated variance for California above‐ground biomass is between 1% and 2% (1 standard deviation) for our current best estimate of the correlation range at 5–10 km—i.e., we bound the standard deviation by a factor of 2. This contrasts with 0.025% (1 standard deviation) if one does not include the correlation term.