Fayrene Nguyen
Is KBAABB a good proxy of TRUE FIA data?
The Forest Inventory and Analysis (FIA) program aims to provide comprehensive data on national forest resources. However, challenges in accessing all forest areas necessitate estimating forest attributes from sampled data. Determining which estimator to use requires assessing estimator properties on a synthetic population of forest attributes. These synthetic forest populations are only useful if they accurately reflect the variability of the actual forest characteristics. This project evaluates the fidelity of KBAABB (k-Nearest-Neighbors approximated to Bayesian bootstrap) synthetic forest populations, which were generated to emulate the true distribution of forest attributes in the United States.
Our team approached this multifaceted project from various perspectives, each focusing on different states, yet all contributing to the overarching objective of validating these synthetic datasets. The analysis suggests that KBAABB synthetic populations do indeed serve as effective proxies for the true FIA data. This efficacy is supported by several key findings: the alignment of the synthetic data points population and spatial structures with well-known ecological zones, the close replication of FIA distributions, and the approximate preservation of important spatial patterns and underlying relationships within the data. These characteristics collectively indicate that the KBAABB artificial populations are credible and suitable for use in simulations where a representative forest population is required.