Study On Nondestructive Measurement And Modeling Methods For Standing Tree Volume

Standing-tree stock volume is a major indicator of ecological service function of forests and largely represents the mature growth stage and ecological vitality of forests, which is also important for the hotspot issues of global warming. Thus, measurement and assessment of forest standing-tree stock volume are of great significance. Because of vast area, large number of trees, and operational difficulty, measurement of standing tree volume for all trees necessitates the consumption of abundant resources. In practice, usually a tree volume model is built on basis of sampling theory and then used into assessment of forest tree stock volume. For many years, foresters cut down standard trees and acquire the samples for building a tree volume model. Despite high precision, the adoption of "cutting down trees" is faced with problems of destructiveness, high cost, and low efficiency, and violates the aim of ecological protection. This work is based on the situation of forest supervision in China and targeted at helping with measurement of forest stock volume. The methods include the generalized error propagation law, experimental comparison and analysis, modern biological mathematical models, and mathematical statistical analysis. An innovative and systemic study was performed to investigate a nondestructive precision data acquisition method, data acquisition & pretreatment, modeling method, and precision assessment.First, the standing tree accurate measurement theory based on electronic theodolite was improved and simplified. Specifically, representative large-, medium-and small-size tree samples were selected and analyzed as per the theory of generalized error propagation. Results show that the relative error of tree height measurement of electronic theodolite was 0.29%-0.89%, and the relative error of the tree volume measurement was 0.31%-0.99%, the precisions are close to those of conventional ruler analytic timber measurement theory. Then 87 cut-down trees were used in validation. The average absolute value of relative errors is 1.70% in tree height measurement and 1.75% in tree volume measurement for the electronic theodolite relative to the conventional method. Clearly, the measurements of tree height and tree volume from the electronic theodolite are highly consistent with the conventional method. Thus, the electronic theodolite can replace the conventional method based on cut-down trees, and thus meets the standards of forestry inventory and forest data table compilation. We explored to build single close-range photogrammetry for standing tree measurement based on ordinary digital photographic lens with phone/tablet computer-assisted artificial measurement. Software-based tests prove the realization of indoor-outdoor integration. With 92 larches as targets, the measurements from the electronic theodolite were used as reference to validate the feasibility of close-range photogrammetry. Results show that for close-range photogrammetry relative to the electronic theodolite, the average absolute value of relative errors is 4.37% in tree height measurement and 4.44% in tree volume measurement, respectively, with coefficients of determination of 0.907 and 0.986,indicating significant correlation between methods. The comparisons indicate that the measurements are consistent between the two methods. Thus, the close-range photogrammetry provides a new clue for standing tree measurement.Second, as per the situation of Beijing poplar planted resources, we designed a modeling sample collection scheme. Specifically, in the 12 districts and counties (Haidian, Chaoyang, Changping, Fangshan, Fengtai, Daxing, Tongzhou, Shunyi, Huairou, Miyun, Pinggu, Yanqing), the electronic theodolite was used to collect samples from three species (groups):Populustomentosa, P. × euramericana, and P. × canadensis. Totally,1678 trees were tested with at least 500 trees in each group. The abnormal data from the modeling samples were eliminated, sorted and inspected, and the adaptability was tested with the binary volume table for North China P.davidiana issued in relevant ministry standard. Results show that no abnormal data were found in the modeling samples, but the old binary volume table is not applicable for estimation of poplar tree volumes in Beijing areas.Finally, conventional weighted nonlinear least squares regression was used to build a standing tree volume model for the three tree species (groups). Tests prove that the precisions of this model are acceptable. Analysis of inter-specific differences shows that model estimations basically change as follows:P.tomentosa>P. × camadensis>P. × euramericana, indicating very large stem shape differences. Thus, separate models should be built for the three poplar species (group). Then nonlinear least squares regression was used to compare the fitted results from 14 binary tree volume models,6 unitary diameter at breast height (DBH) tree volume models, and 9 unitary ground diameter (GD) tree volume models. The models with the optimal statistical indicators and the simplest structure were selected as the optimal models. The precision indicators of the optimal models all show very balanced performances, and the modeling precision does not decrease with the increase of order. Moreover, the tree volume models proposed by forestry experts adapt in different ways to different trunk shape changes. Then error-in-variable simultaneous equations were used to build compatibility standing tree volume models for the three species (groups). Results show that the indicators of the compatibility models are better compared with the traditional method. The fitting results of 4 combinations of simultaneous equations composed of 3 tree height models and 2 GD models were analyzed comparatively. Results show that the tree volume model composed of the simplest tree height model and GD model achieved high precision.