Research On The Spatial Distribution Of Liangshui Nature Reserve Forest Biomass Based On Geographically Weighted Regression

With the industrial development, the impact of human activities on the biosphere is growing, especially the concentration of CO2 and other greenhouse gases in the atmosphere increases year by year,causing the global temperatures rising. Through the photosynthesis of vegetation, carbon dioxide in the atmosphere can be fixed into organic substances, so the forest carbon sequestration function has become a research focus in domestic and abroad. Because the distribution of biomass of spatial nonstationarity, so get the spatial distribution of the forest biomass information timely and accurately is of great significance for accurate estimates of biomass.. In this paper, the study of the content is the use of the terrain data,square plots data and the data in protection area collected in 2010 in Heilongjiang Province LiangShui National Nature Reserve to analysis the forest carbon stocks spatial patterns in Heilongjiang Province Combined with the stand factors and topographic factors, establish the GWR model and OLS model, evaluation model of forecast results..Firstly, the study area is LiangShui National Nature Reserve, analysis by using geographic information system (GIS) spatial analysis methods. This paper adopted the exploratory spatial data analysis (ESDA) method, to explore the spatial data structure of the forest biomass spatial distribution and spatial change,the results show that forest biomass data of LiangShui National Nature Reserve does not conform to the normality, and presented on the east-west direction trends along the quadratic curve, showing trends straight up in the north-south direction; this paper further uses the Moran’s statistic spatial autocorrelation analysis, the results show that LiangShui National Nature Reserve forest biomass presents a certain degree of agglomeration, "hot spots" phenomenon occurs in the north, west and east of the research area, in the lower part of the study area is biomass "cold spots",at the same time forest biomass presents a range of random distribution. Secondly, forest biomass as the dependent variable, the average stand diameter (AVG-DBH), the number of trees per hectare (TPH), altitude (Elevation), slope (Slope) and as independent variables to build the least square model, through the variable selection, select the biomass which has significant impact on the biomass (AVG-DBH) and elevation (Elevation), to establish a linear least squares model, and the linear regression model and linear regression models were tested and residual analysis; and then use statistical software GWR4.0 establish geographically weighted regression model. And uses Moran’s statistic to describe the model residual spatial autocorrelation, results indicate that there is spatial effects of forest biomass distribution in Heilongjiang Province. Elevation and diameter affect the spatial distribution of forest biomass.Finally, by comparing between models, geographically weighted regression model is significantly better than global model (least squares model). Geographically weighted regression model can provide local model coefficient in the study area of each on the same site, the variaty of coefficient can better explain the nonstationarity of the spatial distribution of biomass. Also help forest managers understand the impact of forest management activities on forest biomass.