| Even-aged forest of Populus×xiaohei T. S. Hvang et Liang with three planting densities(2m×5m(A),4m×5m(B) and 4m×10m(C))from the base of fast-growing and high-yielding plantation of poplar in Shuozhou city, Shanxi Province, were studied in this paper. The innovative results were gained. 1.High-resolution QuickBird satellite image technique was first used to evaluate wood properties. Conventional stand inventory, testing and analyzing means of wood properties, as well as image processing, segement and abstraction of high-resolution QuickBird satellite image, and also model techniques were combined organically. Models predicting wood properties and yields based on the independent variables of tree characteristics which, directly or indirectly, were obtained from high-resolution QuickBird satellite images were gained. On the basis of these models, analysis and prediction of various properties and wood yields of Populus×xiaohei was realized, providing an original research method and analysis means for evaluating and predicting wood properties and yields of Populus×xiaohei. In addition, it also provided a theoretical basis for analysis and evaluation of the wood properties of other poplar plantation. 2. During the data processing and modeling, a new analytical method was put forward for the first time to apply Conb-Douglas models of elasticity theory from economics to the wood science and technology. Elasticity of production and marginal production analysis were used between tree characteristics and wood yields.The back-propagation neural network (BP) were introduced in predicting between remote sensing techniques and wood properties ,and the result showed, it was useful method for predicting wood properties. 3.The main results were as follows. (1).Based on analysis of tree growth and wood properties of Xiaohei Poplar plantation, it was found, for three planting densities(SD) of A, B and C, the variance analysis showed that planting densities had extremely significant influences on crown widths(CW), crown lengths(CL), diameter at breast height(DBH), and distribution of diameter at breast height, ratio of diameter to height(D_H); while there was no obvious influences on tree height(H) and breast height form ratio. In addition, it's also found that the crown widths, crown lengths, and diameter at breast height increased with planting densities, and the ratio of diameter to height followed a variation pattern of "V"forms. The lower the planting density is, the higher the proportion of trees with bigger diameter class. (2).Planting densities had significant influences on the yields of some kinds of woods. With the decrease of planting density, the areas of wet heartwood (WA)and sapwood(SA), juvenile wood(JA) and stem area(SA) increased, while ratio of wet heartwood to sapwood area decreased, and the mature wood area(MA) followed a tendency: low-high-low. (3).Planting density had extremely remarkable influences on fiber width, double wall thickness, vessel ratio, fiber wall ratio, vessel number, vessel tangential diameter, basic density, modulus of elasticity (MOE) and compressive strength parallel to grain; it had remarkable influences on ratio of fiber wall to lumen and modulus of rupture (MOR); while it had no remarkable influences on fiber length, fiber lumen diameter, ratio of length to width, tissues of fiber and ray, chemical composition and coarse pulp rate. Vessel number, compressive strength parallel to grain, holocellulose, and benzene-alcohol extractives increased with planting density. More or less, there are some effects of planting density on variation of fiber anatomical parameters and basic density in the radial direction, yet the general variation pattern was not changed. (4).The relationships of wood yields and measured growth characteristics of trees (stand)(e.g, crown width,crown length,diameter at breast height, height, stem taper, planting density etc.) can be described by Conb-Douglas production equation. Therefore, this equation was applied to predict wood area (R2=0.66~0.973). In addition, the elasticity of the output and marginal production analysis method were also applied to explain the relationship between variation of wood areas of Populus×xiaohei and that of growth characteristics of trees. (5).Fairly good results (R2>0.5) were achieved when modulus of elasticity (MOE), modulus of rupture (MOR), compressive strength parallel to grain (CSG), benzene-alcohol extractives, and contents of holocellulose, were predicted based on measured growth characteristics of trees; while only 30~47% of variation could be explained by models established between wood properties such as fiber length, vessel number, vessel tangential diameter, ratio of cell wall, content of cellulose, and growth characteristics of trees. As for other wood properties, the results of prediction were really commonly. (6).Data was obtained from QuickBird satellite image featuring 0.61 m of panchromatic and 2.44 m of multispectral band wave resolution. Differential global position system (GPS) technology was employed to position each standing tree. EREAD IMAGING which was image processing system, was employed to rectification and the fusion of QuickBird satellite image, and object-oriented eCognition software, was employed to processdata from QucikBird satellite, such as, multi-scale segmentation of images, extraction of characteristics, and object-oriented classification. Finally, data of crown parameters (crown widths) were obtained. Precision of crown widths obtained from the QuickBird satellite image reached 97.36%, superior to pervious studies. (7). Diameter at breast height and tree heights were predicted based on the crown widths data obtained from QuickBird satellite image, with an accuracy of 93% and 79% respectively. Volume inside bark and volume outside bark were predicted also from QuickBird satellite image, and R2 were 0.773 and 0.796, respectively. (8). Models of wood yields and wood properties (compressive strength parallel to grain, MOE, MOR etc.) were established based on independent variables of growth characteristics of trees, which, directly or indirectly, were obtained from the QuickBird satellite image. The established models were validated, and the precision of prediction of wood yields and principal mechanical properties was found to exceed 65% and 82%, respectively. In addition, multivariate regression models and the back-propagation neural network (BP) were compared in precision of prediction, and the result showed, on the precondition of small samples, the BP neural network was superior to multivariate regression models. The main multivariate regression predicting models were as follows: JA = -5265.33632 -751.3514Ln(RSCW)+2204.19828Ln(PD)?58.89242Ln(SD) (R2=0.891) MA =433. 45338?498.96863(RSCW)+214.8088(RSCW)2 ?36.64026(RSCW)3+2.17212(RSCW)4(R2=0.644) WHA = -596.11252 +3.1966Ln(RSCW)+452.3394Ln(PD)-78.00336Ln(SD)( R2=0.899) HA = 6 16.64414+247.03516(RSCW)-12135(PD)+27038(SD) (R2=0.898) SA = exp( -44.94291+824.473(PH)-19.35549(RSCW)+208.69558(PD)+45.11216(SD)) (R2=0.648) CSG = EXP(1 0.28458+0.25954(PH)+1.02084(RSCW)?0.42044(PD)?25.93999(SD)) (R2=0.700) MOE = 54 .4627?2.97797(PH)?8.11663(RSCW)+3.12618(PD)?855.46649(1SD) (R2=0.558) According to this analysis, wood yields and main mechanical properties of Populus×xiaohei plantation are possible to be predicted accurately by high-resolution QuickBird satellite image. The capability of poplar plywood and wood quality of Populus×xiaohei werevalued based on prediction models.
|
PDF Full Text Request