Study On The Carbon Density Of Natural Larch Forest Of Daxing’an Mountain Based On GWR Model

Forest carbon stock is the main factor to illustrate the characteristic of the forest.As a result, predicting the carbon stock of the forest with high accuracy is essentially important. To give exact calculation for the carbon storage of the forest ecosystem, the traditional statistical model has been widely used in the research. It laid the strengthening foundation to the carbon stocks of the individual tree level and forest stand level of different forest stand types. Currently, the whole world has been paid more attention to the evaluation for the carbon stock of the forest. So we just get the total amount of the carbon is not enough. It has significant sense to understand the spatial distribution of carbon stocks of the forest so as to provide reasonable and scientific provision for the management for the forest.Taken the PanGu forest farm in Daxing’an Mountain as the research area, the natural Larch carbon stocks for the components and total of 45 trees were used to establish the compatible carbon stock model which is suitable to this area. Based on the forest survey data of PanGu forest farm, the carbon density of the stem, branch, foliage, root and the total tree was calculated.The research made analysis for the relationship between the carbon density for the natural Larch components and altitude, canopy density, slope position, forest density, slope and the aspect. The results showed that:the altitude has positive effect on the carbon density for the stem, branch, root and the total tree. The carbon density of the components and the total amount except for the foliage got the highest value at the canopy density with the change range of 0.5-0.6, and got the minimum value at the range of 0.3-0.4 and the foliage got the highest value at the range of 0.4-0.5. As for the slope position, on the whole the components and the total carbon density all got the highest value at the medium aspect. As for the forest density, the carbon density for the components and the total decreased with the increase of the forest density. As for the slope, the carbon density for the components the total tree got the maximum value at 25-34°; For aspect, the stem, branch and total tree got the maximum at the southeast. Meanwhile the foliage and root got the maximum value at north and northwest part. The components and total got the minimum value at no-slope area.Based on the visualized tool as ArcGIS, Kriging of spatial was used to map the carbon density distribution for the stem, branch, foliage, root and total amount of the natural Larch. The results showed that:the carbon density for the stem on the whole research area was 3.4191-48.2329 t·hm-2 and the carbon density of the stem took the form of patch shape; the carbon density for the branch got the variation range of 0.2160-2.4417 t·hm-2 and the whole tendency was reduced from the northeast part to the southwest part. The carbon density of foliage was 0.2110-0.8961 t·hm-2 and had the variation tendency of reduction from the northeast part to the southwest part. The carbon density of the root was 2.0387-19.8746 t·hm-2 and the maximum value was at the southwest part and the minimum value existed in the central and northern part. The carbon density of the total tree was 3.7869-72.5695 t·hm-2. On the whole, the carbon density decreased from southwest part to the northeast part. As a result, the carbon density of stem, branch, root and total tree got the minimum value in the southwest part and got the minimum value at the central with the second of northeast. The regularity of the foliage was opposite.Stepwise method was used to illustrate the relationship between the factors and the carbon density of natural Larch and the ordinary least square (OLS) model was developed. Combined with the investigation factor of natural Larch carbon density of PanGu forest farm, GWR model was established based on GWR3.0 software. Based on Gaussian and Poisson methods, the combination of kernel function and bandwidth selection of the GWR in which the methods were not the same was used to fit the data.The results showed that GWR models were the best models. Combined with ArcGIS 10.0 software, the model coefficients were visualized and they showed the heterogenicity for the parameters of the model. The results showed that GWR model was the best and the results based on the combination of components was not the same. The carbon density GWR model for stem, root and total tree of natural Larch was the best under Poisson of adjust kernel function and AIC method. For branch, there was no significant difference between adjust and fixed kernel function under AIC. As for foliage, it was equal to each method. Taking the total tree as an example, the parameters were visualized based on ArcGIS10.0 and the results showed that it had spatial heterogeneity between carbon density of natural Larch and aspect, altitude, average DBH and forest density.