Using random forest to disentangle the effects of environmental conditions on height-to-diameter ratio of Engelmann spruce

Berlin

2024

143-156

213 - 229

e-journal

200211702

Keywords: Height-to-diameter ratio · Tree stability · Random forest · Engelmann spruce · Neighborhood competition
Engelmann spruce (Picea engelmannii) is one of the most economically important tree species in the western United States. The species is widely distributed in mid- to high-elevation mountains, so its vulnerability to wind and snow is of concern to foresters. Height-to-diameter (H–D) ratio is an important metric for assessing tree stability and resistance to windstorm and icing damage. But our understanding of the variation in H–D ratio of Engelmann spruce and its driving mechanism is limited. Based on inventory data of 9004 Engelmann spruce trees from 997 permanent plots, this study used the random forest algorithm, an important machine learning method, to develop the H–D ratio model of Engelmann spruce related to competition, site condition, climate, topography and other environmental variables. We used cross-validation to train and optimize H–D ratio model, and then used variable importance ranking and partial dependence plot to quantify and analyze the effects of environmental factors on H–D ratio. Further, structural equation was used to perform path analysis between the response variable and the explanatory variables. The results reveal that competition contributes the most to changes in H–D ratio, followed by site index, topography and climate variables.