摘要
树高是森林资源管理领域的重要参数,获取准确、大范围、空间连续的树高信息对于森林资源监测、管理和碳循环研究等都具有重要的意义。利用星载激光雷达ICESatGLAS波形数据,以大兴安岭塔河林业局营林区为研究区,针对不同地形条件选择合适的算法提取出GLAS光斑内的森林平均树高。联合GLAS波形数据与多角度光学遥感MISRBRF数据,应用随机森林机器学习算法获取塔河林场树高图,实现树高信息提取从点到面尺度上的扩展。用实测样地数据进行检验,其结果为:R2=0.72,RMSE=1.83 m,精度为85.22%。激光雷达数据和多角度光学遥感影像协同,弥补了2种数据各自的不足,为森林生物量以及碳储量的估算提供准确的基础数据。
Tree height is an important parameter in the field of forest resource management, so it is significant in forest resource monitoring and carbon cycling research to obtain accurate, large-scale and spatially continuous tree height. The paper uses the waveform data from space-borne Lidar ICESat/GLAS, and chooses proper algorithms with topographic correction to extract forest tree height average values from GLAS foot-print data. Then, we apply the random forest machine learning algorithm to process tree height values from GLAS foot-print data and MISR/BRF data from multi-angle optical remote sense, and get the tree height map of Tahe Forest Farm, namely, which makes the scale of extracted tree height values extend from point to plane. Finally, the field measured data was utilized to test the results, which showed that R2, RMSE and precision were 0. 72, 1.83 m and 85.22% , respectively. This approach integrates Lidar data with multiangle imagines of optical remote sense, which makes up of the shortcomings of each. In addition, the results also provide accurate reference guidelines for the prediction of forest biomass and carbon storage.
出处
《北京林业大学学报》
CAS
CSCD
北大核心
2014年第4期8-15,共8页
Journal of Beijing Forestry University
基金
"十二五"国家科技支撑计划项目(2011BAD37B01)
