Study On Spatial Correlation Of Forest Characteristics Base On Geostatistics

Intuitively, forest stands are structured spatially on all geographic scales. Based ongeo-statistical methods to analyze spatial autocorrelation and spatial heterogeneity of foreststands, we can understand fully forest spatial structure and improve estimation of forestresources accuracy, which is extremely meaningful for forest management, forest planning andresearch. In this paper, based on bureau-level permanent plot data of forest inventory inWangqing forest enterprise, we analyze spatial autocorrelation and spatial heterogeneity forforest stand invention factors (forest volume, forest above-ground biomass and tree speciesdiversity) in landscape scale. Meanwhile, based on sample plot of natural Mongolian Oakforest, we use geo-statistical methods to analyze spatial autocorrelation of tree attributesvariables (height, diameter, diameter growth and length of crown) and spatial cross-correlationamong different variables.(1) There exists no apparent spatial autocorrelation for forest volume and forestabove-ground biomass at the spatial scale of permanent plot. It means that we can regardpermanent plots as independent samples. Therefore, we can’t use geo-statistical methods toexpand the information of permanent plots to the whole study area.(2) Using trend model and universal kriging, which considers the spatial autocorrelationof residuals in trend model, estimate tree species diversity, respectively. The result shows thatthe prediction accuracy of these two methods is very high. There exists a common spatialpattern of tree species diversity in this study area. Most spatial information of tree speciesdiversity can be explained by environmental variables of trend model. So, compared to trendmodel, the prediction accuracy of universal kriging was not improved significantly. But,universal kriging shows some potential benefits: standard error map which can offer reliabilityfor any prediction point and spatial autocorrelation in the residuals of model which is often asign that an important environmental variable has been overlooked in trend model. (3) Using geo-statistical method analyzes spatial autocorrelation and spatial heterogeneityof tree attribute variables of natural Mongolian Oak forest. The result shows that there existsapparent and different extent spatial autocorrelation and anisotropy phenomena and evenspatial cross-correlation among some variables. For example, individual tree diameter growthis spatially correlated with individual tree competition index. This means that there are strongcompetitive phenomena in some compartments with lower diameter growth speed. So, we canconsider to relieve competition by some ways, such as thinning. Therefore, geo-statistical toolsare recommended for this type of spatial analysis of forest stands. Their use can provide usefulinsights into the nature of stand organization and spatial structure and can guide futuresilvicultural treatment toward higher stand productivity or better and faster regeneration.