不同植被覆盖区ICESat-2和GF-7卫星地表高程信息对比研究  

Research on Comparsion of Surface Elevation Information of Different Vegetation Cover Types Based on ICESat-2 and GF-7 Satellite Data

作　　者：覃志刚 尤号田 黄元威 王玉 李立存 陈建军 QIN Zhigang;YOU Haotian;HUANG Yuanwei;WANG Yu;LI Licun;CHEN Jianjun(College of Geomatics and Geoinformation,Guilin University of Technology,Guilin 541006,China;Lin Zhiyuan(Beijing)Forestry Engineering Consulting Co.,Ltd.,Beijing 100020,China)

机构地区：[1]桂林理工大学测绘地理信息学院,桂林541006 [2]林之源(北京)林业工程咨询有限公司,北京100020

出　　处：《航天返回与遥感》2023年第5期91-104,共14页Spacecraft Recovery & Remote Sensing

基　　金：国家自然科学基金(41901370,42261063);广西自然科学基金(2020GXNSFBA297096);桂林理工大学科研启动基金(GLUTQD2017094);广西八桂学者专项项目(何宏昌)。

摘　　要：卫星立体影像和星载激光雷达是获取区域数字表面模型(DSM)的主要技术手段,为进一步研究利用两类数据获取地表高程结果的差异,文章选取冰云陆地高程卫星(ICESat-2)与“高分七号”卫星(GF-7)立体影像,在不同植被覆盖情况下对利用两类数据生成的地表高程结果进行对比研究。结果表明,在耕地、草地、森林三种不同植被覆盖区域下,ICESat-2强波束与GF-7生成DSM的相关性较好,相关性与均方根误差分别为0.98/6.56m、0.99/8.76m、0.95/35.17m,弱波束的相关性与均方根误差分别为1.00/4.45 m、0.99/7.09 m、0.96/29.36 m;在森林区域两类数据结果差异相对较大,两者相关性弱于草地和农田,经重新滤波处理后两类数据相关性得到提升,其强/弱波束与GF-7生成DSM的相关性均可达到0.99,均方根误差与绝对误差的平均值也大幅减小,且随着统计尺度的增加逐渐下降。同时,基于官方滤波算法所得强/弱波束数据均会高估数字表面高程,且弱波束反演精度略优于强波,但经重新滤波处理后,强、弱波束反演精度相近,且弱波束出现低估数字表面高程的现象。Satellite stereo images and spaceborne light detection and ranging(LiDAR)are the main technical means for acquiring regional digital surface model(DSM).To further investigate the difference in obtaining surface elevation results using two data,Ice,Cloud,and Land Elevation Satellite-2(ICESat-2)and Gaofen-7 satellite(GF-7)stereo images were used to generate and compare the surface elevation results under different vegetation cover conditions.The results showed that under three different vegetation coverage areas of cultivated land,grassland,and forest,the correlation between the surface elevations obtained from strong beam of ICESat-2 and GF-7 generated DSM was good,with correlation and root mean square errors of 0.98/6.56 m,0.99/8.76 m,0.95/35.17 m respectively.The correlation and root mean square errors between the surface elevations obtained from the weak beam of ICESat-2 and GF-7 generated DSM were 1.00/4.45 m,0.99/7.09 m,0.96/29.36 m,respectively.And in forest areas,there was a relatively large difference between the surface elevations obtained from the two data,and the correlation was weaker than that of cultivated land and grassland.By re-filtering the ICESat-2 data in forest area,the correlation between the surface elevations obtained from strong/weak beams and GF-7 generate DSM reached 0.99.The average value of root mean square error and absolute error also significantly decreased,and both gradually decreased with the increase of statistical scale.The strong/weak beam data obtained based on the official filtering algorithm will overestimate the digital surface elevation,and the inversion accuracy of the weak beam is slightly better than that of the strong beam.However,after re-filtering,the inversion accuracy of the strong and weak beams is similar,and the weak beam appears to underestimate the digital surface elevation.

关 键 词：地表高程信息 不同植被覆盖类型 重新滤波 

分 类 号：V445[航空宇航科学与技术—飞行器设计]