Original scientific paper
https://doi.org/10.15177/seefor.21-04
Height to Crown Base Modelling for the Main Tree Species in an Even-Aged Pedunculate Oak Forest: A Case Study from Central Croatia
Saray Martín-García
; föra forest technologies SLL, Campus Duques de Soria s/n, ES-42004 Soria, Spain
Ivan Balenović
; Croatian Forest Research Institute, Division for Forest Management and Forestry Economics, Trnjanska cesta 35, HR-10000 Zagreb, Croatia
Luka Jurjević
; Croatian Forest Research Institute, Division for Forest Management and Forestry Economics, Trnjanska cesta 35, HR-10000 Zagreb, Croatia
Iñigo Lizarralde
; föra forest technologies SLL, Campus Duques de Soria s/n, ES-42004 Soria, Spain
Krunoslav Indir
; Croatian Forest Research Institute, Division for Forest Management and Forestry Economics, Trnjanska cesta 35, HR-10000 Zagreb, Croatia
Rafael Alonso Ponce
; föra forest technologies SLL, Campus Duques de Soria s/n, ES-42004 Soria, Spain
Abstract
The height to crown base (hcb) is a critical measure used in many investigations as an input variable to investigate the vigour of the stands, the social position of the trees, and to evaluate the behaviour of forest fires, among other uses. Though measuring height-related variables in the field is always time-consuming, the foremost benefits offered by modelling hcb are that it permits to generalize and average a very uneven attribute and, furthermore, provides insights about which tree and stand variables have a significant impact on hcb. However, there are many species in which models of the crown base height have not been developed in Croatia. The objective of this research was to develop a height to base crown model for each of the main species present in the two-layered mixed stands of this study. According to previous investigations, logistic models provide the highest precision and require the lowest inventory cost owing to less frequent measurements. Tree- and plot-level variables with distance-independent competition indexes were studied in the fitting model. In this research, we obtained models for the main stand species: Acer campestre (root mean squared error (RMSE) = 2.28 m, R2 = 82.80%); Alnus glutinosa (RMSE = 1.78 m, R2 = 85.36%); Carpinus betulus (RMSE = 2.47 m, R2 = 67.55%); Fraxinus angustifolia (RMSE = 2.46 m, R2 = 82.45%); Quercus robur (RMSE = 2.60 m, R2 = 80.57%); Tilia sp. (RMSE = 2.01 m, R2 = 89.07%); and Ulmus laevis (RMSE = 1.71 m, R2 = 92.42%). The combination of the total height, tree, and plot-level variables with distance-independent competition indexes contributed to the prediction accuracy of proposed model significantly.
Keywords
two-layered mixed stands; distance-independent competition indexes; logistic models; leave-one-out cross-validation
Hrčak ID:
260496
URI
Publication date:
30.6.2021.
Visits: 915 *