A Method for Estimating Tree Growth Potential with Back Propagation Neural Network

Abstract

Tree growth potential is crucial for maintaining forest health and sustainable development. Traditional expert-based assessments of growth potential are inherently subjective. To address this subjectivity and improve accuracy, this study proposed a method of using Backpropagation Neural network (BPNN) to classify tree growth potential. 60 Pinus tabulaeformis (Carr.) and 60 Platycladus orientalis (Linn.) were selected as experimental trees in the Miyun Reservoir Water Conservation Forest Demonstration Zone in Beijing, and 95 Pinus massoniana (Lamb.) and 60 Cunninghamia lanceolate (Linn.) were selected as experimental trees in the Jigongshan Nature Reserve. The average annual ring width of the outermost 2 cm xylem of the experimental trees were measured by discs or increment cores, and the wood volume increment of each experimental trees in recent years were calculated. According to wood volume increment, the growth potential of experimental trees was divided into three levels: strong, medium, and weak. Using tree height, breast height diameter, average crown width as input variables, using growth potential level as output variables, four sub models for each tree species were established; Using tree species, tree height, breast height diameter, average crown width as input variables, using growth potential level as output variables, a generalized model was established for these four tree species. The test results showed that the accuracy of the sub models for Pinus tabulaeformis, Platycladus orientalis, Pinus massoniana, and Cunninghamia lanceolate were 68.42%, 77.78%, 86.21%, and 78.95%, respectively, and the accuracy of the generalized model was 71.19%. These findings suggested that employing BPNN is a viable approach for accurately estimating tree growth potential.