基于GF-1PMS影像的森林郁闭度定量估测模型研究

Quantitative Estimation of Forest Canopy Density Based on GF-1 PMS Images

森林郁闭度是我国森林资源二类调查中的重要林分因子之一,为探索高分一号影像在森林郁闭度定量估测的可行性,以内蒙古自治区东北部柴河林业局为研究区,基于GF-1 PMS多光谱影像和DEM数据,以纹理特征、光谱信息和地形因子为自变量,采用k-最近邻法(k-Nearest Neighbor,k-NN)、稳健估计以及偏最小二乘法3种方法构建研究区域的森林郁闭度估测模型,辅之以二类调查小班数据和现地实测数据进行模型精度评价。结果表明:1)3种郁闭度估测方法的应用效果均能满足实际需求,这说明GF-1 PMS多光谱影像在森林郁闭度定量估测方面具有一定的潜力;2)k-NN法和稳健估计的实际应用效果明显优于偏最小二乘法,且2种模型的估测精度均>80%;3)3种郁闭度估测稳定性分析,结果显示k-NN法稳定性较好,而稳健估计法和偏最小二乘法估测模型不稳定。

Forest density is one of the important forest stand factors in China's forest resource class II surveys.In order to explore the feasibility of the GF-1 images in quantitative estimation of forest density,based on GF-1 PMS multispectral images and DEM data,this study adopted k-nearest neighbor(k-NN),robust estimation,and partial least square methods to construct a model for estimating forest density in the study area by using texture features,spectral information,and topographic factors as independent variables.The model accuracy was evaluated with small-group data and in situ measurement data from the Type II survey.The results showed that 1)the application of the three methods for estimating forest density could meet the practical needs,which indicated that GF-1 PMS multispectral images had certain potential for quantitative estimation of forest canopy density;2)the practical applications of k-NN and robust estimation methods were significantly better than partial least square method,and the rates of estimation accuracy of both models were higher than 80%;3)the stability analysis of the three depression estimations showed that the k-NN method was more stable,while the other two methods were unstable.