Percentile-Based Modeling of Height to Crown Base Distribution Using Stand-Level Variables in Even-Aged Maritime Pine Stands

Stand-level canopy base height (Cbh) is a key variable controlling crown fire initiation, yet it is commonly computed as the mean of tree height to the base of the crown (hbc), which does not reflect the lower portion of the hbc distribution governing the transition from surface to crown fire. This study investigates the relationship between physically based Cbh definitions and the hbc distribution. We develop a general multi-percentile modeling framework to estimate hbc percentiles at the stand level. Using a dataset of Pinus pinaster Aiton trials in Portugal, percentile-specific models (5th to 50th) were fitted and synthesized into a nonlinear multi-percentile formulation. Results show that the height at which canopy bulk density exceeds the critical threshold does not match mean hbc, but instead corresponds to lower percentiles, typically around the 10th percentile, varying with stand structure and age. Mean-based Cbh tends to overestimate the lower canopy boundary, reflecting its inability to capture structural variability. The final model predicts hbc at any percentile and incorporates effects of stand height, basal area, tree density, and age, ensuring positive predictions and high predictive accuracy (adjusted R2 = 0.9770; RSE = 0.4073 m; PRESS R2 = 0.9769). The framework provides a consistent representation of canopy base height for fire behavior modelling.