LiDAR Approach
Aboveground biomass will be modeled based on the relationship between variables derived from the LiDAR data at plot locations and aboveground biomass (AGB) measured in-situ at the plot locations. Metrics likely to used can be found in the attached report \cite{McGaughey2014}. See CloudMetrics section for detail on metrics and their derivation. The use of sub-meter precision GPS ensures that the LiDAR point-cloud and the plot data are coincident spatially. The relationship between LiDAR metrics and plot-measured AGB will be modeled and AGB will be predicted on a\(\frac{1}{20}^{\mathrm{th}}\) acre continuous grid across the entire project area. See Section 5.1.3 in \citet{Tittmann2015a} for further detail on model development and calculation of sampling error.
Sequential sampling
New Forests proposes two possible options to use LiDAR derived aboveground carbon inventory in sequential sampling for verification. The two approaches are based on paired and unpaired approaches to sequential sampling.
Paired sampling approach
This is our preferred option. Sequential sampling will be conducted using a paired test on field-measured, permanent monumented plots by strata as is done for plot-based inventories. LiDAR-based model inputs, methods, and results coincident with monumented plots will be provided along with field-based measures to allow verifiers to test modeled LiDAR plot data directly. This approach is similar to how standard height-diameter regressions are
tested currently.
A hybrid sampling design will be used to 1. enable paired sequential sampling of measured plots and 2. to ensure that the model used to predict biomass density from LiDAR is calibrated using the full range of potential AGB densities. To ensure that only randomly selected plots are considered for sequential sampling, a random-origin systematic design will be employed across the project area. In addition, mapped biomass density estimated from remote sensing data will be used to evaluate weather the randomized samples reflect full range of variability in aboveground biomass density. If no plots fall in areas with the highest and lowest biomass densities, additional plots will be installed within those areas. This step will insure that the model developed form LiDAR will be robust in AGB prediction across the range of biomass densities, not just for randomly selected plot locations. All of the samples will be used to develop and test the LiDAR \(\Longrightarrow\)AGB model.
Unpaired sampling approach
A secondary option would be to use unpaired plots to conduct sequential sampling. In this approach, inventory plot locations would not be randomized but would be located based on a-priori knowledge of biomass densities derived from remote sensing sources . A minimum of 40 inventory plots per strata will be used to develop and test the LiDAR model. In this way, the LiDAR\(\Longrightarrow\)AGB model will be robust across the range of biomass densities. The unpaired test will be conducted comparing randomly selected plots measureed by the verifier against the mean AGB for a given stand.