Prediction of diameter using height and crown attributes: A case study

Abstract

Recent advances in remote sensing provide increasingly detailed forest information in a timely and cost-effective manner. Individual tree stem diameter, an important variable for operational forest inventory, cannot be determined directly from remotely sensed data; stem diameter must be estimated from ancillary measures of tree crown, tree height, and/or measures related to stand structure. In this study, we developed predictive models of diameter as a function of height and crown attributes using a nonlinear mixed-effects approach. Long-term silvicultural experiments provided data for several species: black spruce, Douglas-fir, and lodgepole pine. Addition of crown area and crown area of larger trees to diameter-height models significantly improved predictive performance, resulting in low root mean squared error values between 0.9 and 1.8 cm (10 -11% of mean diameter). Our models indicated the need to include additional explanatory variables at wider levels of spacing and thinning. This case study highlights several practical implications for developing and refining individual tree diameter models based on variables that can be remotely sensed. Copyright © 2012 by the Society of American Foresters.