Edge Correction of Voronoi Diagram in Forest Spatial Structure Analysis
Liu Shuai,Zhang Jiang,Li Jianjun,Zhou Guoxiong,Wu Shuci
DOI: https://doi.org/10.11707/j.1001-7488.20170104
2017-01-01
Abstract:[Objective]In order to provide technical support for the analysis and control of forest spatial structure,this paper put forward a novel edge correction method,that is,distance determinant of Voronoi nodal.[Method]To assess the performance of distance determinant of Voronoi nodal,this study compares and analyzes this new method with other edge correction methods by firstly introducing the quantitative evaluation index RMSE,then applying uniform angle index, aggregation index and other spatial structural index to form RMSE,and finally setting up different simulation plots and survey plots. [Result]Although buffer zone,nearest-neighbor correction,distance determinant of Voronoi nodal are all called minus-sampling correction methods,their performances and effects are affected by several factors,such as sample size,tree spatial pattern,forest spatial structure indices,plot conditions,etc. buffer zone is simple and easy to operate, but its samples are seriously wasted and there also exist leakage and errors,which leads to a usual high level of RMSE curve. Distance determinant of Voronoi nodal can be seen as a correction method which adaptively changes with the actual border forest distribution. It has greater ability to correct and can maximize the elimination of edge effects in plots,thus which is better than the other edge correction methods when calculating uniform angle index,size ratio and aggregation index with sufficient samples. However,there exist some shortcomings such as error determinant,low sample utilization ratio,etc. The performance of nearest-neighbor correction stands fell the middle of the above-mentioned two methods,and its overall level of RMSE was higher than that of distance determinant of Voronoi nodal. [Conclusion]Sample size is closely related to edge correction methods,any changes in the plot size or forest density are ultimately reflected as theincrease or decrease of sample size. Therefore,maintaining adequate samples for edge correction is particularly important. Spatial distribution is another important factor which would affect edge correction. When the spatial pattern is random or uniform,the important role of edge correction can be better reflected. In addition,after edge correction,the calculation accuracy of all kinds of spatial structure indices can be improved significantly and adequate forest sample sizes are needed to calculate the aggregation index.