Hierarchical Bayesian spatial models for predicting multiple forest variables using waveform LiDAR, hyperspectral imagery, and large inventory datasets

► We detail a modeling framework for predicting forest variables with uncertainty. ► We propose an approach for dimension reduction of LiDAR and hyperspectral data. ► Analysis demonstrates the need to meet model assumptions to draw correct inference. ► Addition of multivariate spatial random effects provides improved predictive inference. ► Dimension reduction of the spatial random effects is needed in application.