中文摘要：

森林郁闭度的空间分布是评价森林生产力和分解率的一个重要指标.本研究以吉林汪清林区为研究区,分别利用星载激光雷达ICESat-GLAS波形数据和多光谱遥感Landsat-TM影像对该区的森林郁闭度进行估测,然后采用多元线性回归和BP神经网络两种方法对GLAS数据和TM数据进行联合,共同估测了森林郁闭度.结果表明：单一遥感数据估测森林郁闭度时,GLAS数据的模型决定系数为0.762,TM数据的模型决定系数为0.598.将GLAS数据和TM数据联合后估测森林郁闭度时,多元线性回归模型的复决定系数为0.841,BP神经网络模型的仿真精度为0.851.表明ICESat-GLAS数据与Landsat-TM影像联合能够发挥多源遥感数据的优势,提高森林郁闭度的估测精度,并为后续的空间区域内森林郁闭度的连续制图提供可靠的方法.

英文摘要：

The spatial distribution of forest canopy cover is a critical indicator for evaluating the for- est productivity and decomposition rates. With the Wangqing Forest Region in Jilin Province of Chi- na as the study area, this study first estimated the forest canopy cover using spaceborne LiDAR IC- ESat-GLAS waveforms and Landsat-TM multispectral images, respectively, and then GLAS data and TM images were combined to further estimate forest canopy cover by using multiple linear regression and BP neural network. The results showed that when the forest canopy cover was estimated with single data source, the determination coefficient of model was 0.762 for GLAS data and 0.598 for TM data. When the forest canopy cover was estimated by combining GLAS data and TM data, the determination coefficient of model was 0.841 for multiple linear regression, and the simulation pre- cision was 0.851 for BP neural network. The study indicated that the combination of ICESat-GLAS data and Landsat-TM images could exploit the advantages of multi-source remote sensing data and improve the estimating accuracy of forest canopy cover, and it was expected to provide a promising way for spatially continuous mapping of forest canopy cover in future.