基于遥感的区域尺度森林地上生物量估算研究

森林是陆地生态系统最大的碳库,精确估算森林生物量是陆地碳循环研究的关键。首先从机载LiDAR数据中提取高度和密度统计量,采用逐步回归模型进行典型样区生物量估算;然后利用机载LiDAR数据估算的生物量作为样本数据,与多光谱遥感数据Landsat8 OLI的波段反射率及植被指数建立回归模型,实现区域尺度森林地上生物量估算。实验结果显示,机载LiDAR数据估算的鼎湖山样区生物量与地面实测生物量的相关性R2达0.81,生物量RMSE为40.85 t/ha,说明机载LiDAR点云数据的高度和密度统计量与生物量存在较高的相关性。以机载LiDAR数据估算的生物量为样本数据,结合多光谱遥感数据Landsat8 OLI估算粤西北地区的森林地上生物量,精度验证结果为:R2为0.58,RMSE为36.9 t/ha;针叶林、阔叶林和针阔叶混交林等3种不同森林类型生物量的估算结果为:R2分别为0.51(n=251)、0.58(n=235)和0.56(n=241),生物量RMSE分别为24.1 t/ha、31.3 t/ha和29.9 t/ha,估算精度相差不大。总体上看,利用遥感数据可以开展区域尺度的森林地上生物量估算,为森林固碳监测提供有力的参考数据。

Estimation of the Forest Aboveground Biomass at Regional Scale Based on Remote Sensing

Forested areas are the largest carbon pool in terrestrial ecosystems. Thus, a key link in terrestrial carbon pool research is estimating the forest biomass accurately. In this study, canopy height and density indices were calculated from LiDAR point cloud data. Statistical models between the biomass calculated from field data and LiDAR-derived variables were built. A stepwise regression was used for variable selection and the maximum coefficient of determination (R2). Techniques for improving variable selection were applied to select the LiDAR-derived variables to be included in the models. Lastly, the forest aboveground biomass as estimated by field data and LiDAR data, was regarded as sample data. The forest aboveground biomass calculated from LiDAR data, band reflectance and vegetation indices of Landsat8 OLI were used to establish the regression model for estimating the forest aboveground biomass at a regional scale. The result shows that: the correlation (R2) between the biomass estimated by LiDAR data and the biomass calculated from field inventory data was 0.81, and the RMSE of biomass is 40.85 t/ha, which means canopy height indices and density indices of airborne LiDAR point cloud data has a strong relationship with biomass. The biomass was estimated by airborne LiDAR data and Landsat8 OLI for coniferous forest, broad-leaved forest and coniferous and broadleaf mixed forest. The estimated correlation results showed that R2 was 0.51 (n=251), 0.58 (n=235) and 0.58 (n=241) respectively, and the RMSE for biomass was 24.1 t/ha, 31.3 t/ha and 29.9 t/ha respectively. The resulting estimated biomass for three different forest types is pretty much the same. On the whole, it is feasible and reliable to estimate forest aboveground biomass at regional scale based on remote sensing. The estimated biomass can provide useful data for the monitoring of forest ecosystem carbon fixation.