探索林分形高建模与评价方法,建立黑龙江重点国有林区适用于不同林分类型的形高模型。基于9 974个样地的资料,归类形成17种林分类型,选择基本覆盖所有常见林分形高模型的8种函数,分别进行拟合,利用调整后相关指数(Ra²)最大化和估计值标准差(SEE)最小化的原则,优选出不同林分类型的最适模型,并采用总相对误差(TRE)、平均系统误差(MSE)、平均预估误差(MPE)和平均百分标准误(MPSE)4个指标对最适模型进行交叉验证。建立了32个(亚)区域性最适林分形高模型,各(亚)区域内最适模型的TRE和MSE均在±0.1%内,MPE均小于3%,MPSE均小于12%,形成了黑龙江重点国有林区具有代表性的形高模型系统。建立的最适林分模型与各相应的样本点之间均有较好的契合度,模型具有较好的预估性,弥补了黑龙江重点国有林区林分形高建模方面的空缺,在森林资源规划设计调查及连续清查中具有推广使用价值。进一步验证了以全部数据拟合模型、以“刀切法”验证模型的方法是可行的。

The modeling and evaluation methods of stand form height were studied in order to establish the stand form height models suitable for different stand types in key state-owned forest areas in Heilongjiang Province.Based on the data from 9974 sample plots,17 stand types were formed by classification,8 functions which covered all common stand form height models were selected for fitting,and the optimal models were determined by maximizing the adjusted coefficient determination (Ra²)and minimizing the standard error of estimate (SEE).Four essential evaluation indices including the total relative error (TRE),the mean systematic error (MSE),the mean prediction errors (MPE)and the mean percent standard errors (MPSE)were used to cross-verify the optimal models.Optimal models for 32 sub-regions were established.The further regression analysis showed that the TRE and the MSE were within ±0.1%,the MPE was less than 3% and the MPSE was less than 12%.The representative stand form height model system of Heilongjiang key state-owned forest region was formed.These results showed that the established optimal models had good compatibility with the corresponding sample plots,and the models had good predictability.This study made up for the vacancy of stand form height modeling in Heilongjiang key state-owned forest region,which had the value of popularization and could be applied in the forest managementinventory and in the continuous forest inventory.It was further verified that the method of model fitting by using all sample data and model verification by using jack-knife method was feasible.