To achieve a micro-destructive and rapid measurement of the wood density of standing trees, this study investigated the possibility of the unified modeling of multiple tree species, the reliability of the micro drilling resistance method for measuring wood density, the relationship between drilling needle resistance and wood density, and whether moisture content has a significant impact on the model. First, 231 tree cores and drill resistance data were sampled from Pinus massoniana, Cunninghamia lanceolate, and Cryptomeria fortunei. The basic density and moisture content of each core were measured, and the average value of each resistance data record was calculated. Second, the average drill resistance, the natural logarithm of average drill resistance, and absolute moisture content were used as independent variables, while the basic wood density was used as the dependent variable. Third, the total model of the three tree species and sub-model for each tree species were established through a stepwise regression method. Finally, the accuracy of each model was compared and analyzed with that of using the average basic density of each tree species as an estimated density. The estimated accuracy of the total model, sub model, and average wood density modeling data were 90.070%, 93.865%, and 92.195%, respectively. The results revealed that the estimation accuracy of the sub-model was 1.670 percentage points higher than that of the average wood density modeling data, while the estimation accuracy of the total model was 2.125 percentage points lower than that of the average wood density modeling data. Additionally, except for Cryptomeria fortunei, the natural logarithm of drill resistance significantly influenced the wood density model at a significance level of 0.05. Moreover, moisture content significantly affected the total model and sub-models of Pinus massoniana at a significance level of 0.05. The results indicated the feasibility of using the micro-drilling resistance method to measure the wood density of standing trees. Moreover, the relationship between wood density and drill resistance did not follow a linear pattern, and moisture content slightly influenced the drill needle resistance. Furthermore, the establishment of a mathematical model for each tree species was deemed essential. This study provides valuable guidance for measuring the wood density of standing trees through the micro-drilling resistance method.

Tea disease detection is crucial for improving the agricultural circular economy. Deep learning-based methods have been widely applied to this task, and the main idea of these methods is to extract multiscale coarse features of diseases using the backbone network and fuse these features through the neck for accurate disease detection. This paper proposes a novel tea disease detection method that enhances feature expression of the backbone network and the feature fusion capability of the neck: (1) constructing an inverted residual self-attention module as a backbone plugin to capture the long-distance dependencies of disease spots on the leaves; and (2) developing a channel–spatial attention module with residual connection in the neck network to enhance the contextual semantic information of fused features in disease images and eliminate complex background noise. For the second step, the proposed channel–spatial attention module uses Residual Channel Attention (RCA) to enhance inter-channel interactions, facilitating discrimination between disease spots and normal leaf regions, and employs spatial attention (SA) to enhance essential areas of tea diseases. Experimental results demonstrate that the proposed method achieved accuracy and mAP scores of 92.9% and 94.6%, respectively. In particular, this method demonstrated improvements of 6.4% in accuracy and 6.2% in mAP compared to the SSD model.

Wood density is an important attribute that is positively correlated with many wood quality parameters. How to express wood density with drill resistance is a challenge. In this study, we determined the optimal resistance expression using the current, voltage, and power of the DC (Direct-Current) motor. Nine wood blocks crossing over the pith, with widths and heights of 2 cm and 5 cm, respectively, were obtained from three larch (Larix spp.) trees. The microdensity of each wood block at every 0.1-mm length was measured, and the average current, voltage, and power of the DC motor was recorded when the drill fed forward at every 0.1 mm by a self-manufactured resistance drill. The drill path was parallel to the lengthwise direction of the block. The linear models between the current, voltage, and power of the DC motor and wood microdensity were subsequently established, and the model with the highest R²_adj was selected to express the resistance. The adjusted R²_adj of the forward stepwise regression models between the current, voltage, and power of the DC motor and wood microdensity were determined as 0.2943. This suggests that the drilling needle resistance expressed with the current, voltage, and power or other combination may be optimal.

To improve the estimation accuracy of wood density and study the linear correlation between drill feed resistance and wood density, a new micro drill instrument prototype that can simultaneously measure the rotation resistance and feed resistance of the drill needle was designed. The test tree species included hard and soft broad-leaved trees and coniferous trees, and the absolute dry density of each wood sample was measured. The drill resistance data were tested by our newly proposed prototype and a Resistograph 650-SC, and four linear models were established to define the relation between drill resistance and the absolute dry density of the wood. The results showed that (1) the statistical indicators of each model for our proposed prototype were better than the corresponding indicators of the Resistograph 650-SC for three of the four species tested; (2) the coefficient of determination of the linear regression model between the feed resistance of our proposed prototype and the absolute dry density of wood was 0.946; and (3) the statistical indicators of the model that included rotation resistance and feed resistance were better than those of the model that only including rotation resistance. Although the proposed prototype produced a competitive level of accuracy and explicitly demonstrated that including feed resistance improved wood density measurement accuracy, the prototype should be considered a first iteration as further hardware design changes and in-forest performance assessments across wider diverse set of test species are required before a conclusive evaluation can be rendered.

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.