Study On Stand Structure Of Larix Kaempferi Artificial Forest And Its Increment Prediction

Larix kaempferi, as an exotic tree, was well planted more than 50 years in Changlingang Forest Farm in Jianshi county, Hubei province and played an important role in developing local economy, adjusting industrial structure, enriching biodiversity and improving ecological environment.For the purposes of better management and utilization of Larix kaempferi artificial forest, based on the concept of establishing digital forestry management, in this study digital elevation model (DEM) was established, degree of slope and aspect of slope of topographic eigenvalues were extracted. This provided original data source for the further evaluation of forest habitat quality. Stepwise regression was used to select main factors in the evaluation of forest habitat quality. Degree of slope, aspect of slope, elevation and soil depth had significant effects on the growth of Larix kaempferi and were used as factors of forest habitat quality evaluation by quantifying qualitative factors in the light of regulated standards. Grey fixed weight clustering was used in the quantitative evaluation of forest habitat quality and the method of analytic hierarchy process(AHP) was used in determining weights of each factor. Weights of elevation, soil depth, degree of slope and aspect of slope were eventually calculated to be 0.0616, 0.1737, 0.2910 and 0.4737 respectively, in which shady slope ranked the first.Values of soil depth were obtained by using the method of Ordinary Kriging (OK) in geostatistics. The exponential model with considerable small error was selected as the variogram, in which partial sill was 17.35, nugget was 90.814, sill was 108.164 and spatial variable index was 83.9%, and this meant comparable weak spatial correlations and that the variability of soil depth derived from random factors.Tree height, diameter at breast height and timber volume are key stand measuring factors in studying stand structure and its increment. The analysis of 20 random sample plots sampled from 43 temporary sample plots indicated that these plots in accord with normal distribution shared 65%, and all of these plots were in accord with Weibull distribution. The further evaluation of forest habitat quality in 20 sample plots by the method of grey fixed weight clustering and the statistic of individual tree number according to the similarity of habitat quality and diameter class showed that all plots didn't coincide with normal distribution, but plots with good or bad habitat quality coincided with Weibull distribution, while plots with the medium habitat quality didn't coincide with Weibull distribution. Data of average diameter at breast height, tree height and timber volume at different age class were obtained by using the method of mean tree in temporary sample plots. The correlation between average tree height(H) and diameter at breast height(D) met the formulaH = 0.4365D^1.3292 , and the correlation coefficient was 0.9957. H-Weibull distribution wasinduced from D-Weibull distribution and the table of theoretical probability distribution for tree height was made. Seven methods were used to analyze the total increment after the analysis of the table of theoretical probability distribution for tree height. The analysis of current annual growth was obtained by the first order derivatives of artificial neural network, empirical Levacovic equation and theoretical Richard growth equation. With the standard of the least square sum of residues fitted by total increments of three stand measuring factors, the best methods to predict diameter at breast height, tree height and timber volume were respectively artificial neural network, theoretical Richard growth equation and grey-Markov model, in which the square sum of residues were 0.29, 0.4900 and 0.0002 respectively. Compared with the summation of the square sum of residues for three stand measuring factors by the above seven methods, the order from the smallness to bigness were theoretical Richard growth equation, empirical Levacovic equation, grey-Markov model, grey artificial neural network, artificial neural network, grey genetic algorithm and Grey model(GM) (1,1) in turn. Moreover, the square sum of residues for three stand measuring factors by theoretical Richard growth equation were all smaller than 1. As a result, the best method to study on the increment of three stand measuring factors was theoretical Richard growth equation. The main innovative ideas in this study are as the following:1 Based on exacting characteristic factors from DEM model and the geostatistic study on soil depth, the law of stand structure of Larix kaempferi artificial forest was studied by evaluating its habitat quality quantitatively.2 The artificial neural network model was used in studying increments of three stand measuring factors and the first order derivatives of artificial neural network function was used to obtain the current annual growth.3 In determining with function of four main factors, degree of slope, elevation and soil depth could be expressed as whiten functions, but aspect of slope could not. One reason is that the value was -1 when there was no aspect of slope, but other values were 0°-360° based on exacting values of aspect of slope in Arc View software. The other reason is that Larix kaempferi was sunny tree showed in many research documents, but it can't bear above shading and grows better on shady or semi-shady slope than on sunny or semi-sunny slope. Considering the continuity of aspect of slope valuation and usual custom, the current classic whiten function and the research similar to the past was not found.