基于辅助信息的森林蓄积量空间模拟

以北京市密云县一类清查的样地蓄积量为研究对象,结合与蓄积量相关的辅助因子,采用普通克里格法、协同克里格法对森林蓄积量进行空间插值估测,并与文献25]同一研究区的基于偏最小二乘法回归法估测结果进行比较分析。结果表明,普通克里格法、基于辅助信息的协同克里格法、偏最小二乘回归法的蓄积量估测值与实测值间的相关系数分别为0.389、0.845、0.766;基于辅助信息的协同克里格法要优于普通克里格法和偏最小二乘回归法,能够明显提高预测精度;与普通克里格法相比,所产生的均方根误差减小了71%,预测值和实测值的相关系数提高了54%。最后生成了密云县森林蓄积量空间分布图。研究表明应用地统计学方法进行蓄积量估测具有很好的应用前景,可以为森林蓄积量的估测提供一种可行的方法。

Spatial Modeling of Forest Stock Volume Based on Auxiliary Information

Taking the stock volume of continuous forest inventory in Miyun District of Beijing as research object, and combining auxiliary factors associated with stock volume, the spatial interpolation analysis of the stock volume was carried out by using the ordinary Kriging and Co-Kriging methods, and the results were compared with those of reference 25 in the same study area based on the partial least squares regression method. The results show that based on the auxiliary information, Co-Kriging method is superior to ordinary Kriging and partial least squares regression method, the correlation coefficient between the estimated value and the measured value based on Co-Kriging method was 0.845, the correlation coefficient between the estimated value and the measured value based on ordinary Kriging method was 0.389, and the correlation coefficient between the estimated value and the measured value based on partial least squares regression method was 0.766, respectively. Co-Kriging can significantly improve the prediction accuracy compared with ordinary Kriging, generating the root mean square error decreased by 71%, respectively, and the correlation coefficient between predicted values and measured values increases by 54%. Finally, the spatial distribution map of forest stock volume in Miyun was generated. The research shows that the application of geo-statistical methodology has a good application prospect, and it can provide a feasible method for the estimation of forest stock.