摘要
【目的】提出一种基于TanDEM-X SAR数据的RVoG模型三阶段算法反演森林冠层高度,以解决RVoG模型实际应用中模型成立条件难以严格满足、受地形影响导致森林冠层高度估测精度不高的问题。【方法】以云南省普洱市思茅区思茅松纯林和混交林为研究对象,开展经典三阶段算法、地面相位优化的三阶段算法、纯体散射复相干优化的三阶段算法和低估补偿改进的三阶段算法反演森林冠层高度试验。【结果】RVoG模型经典三阶段算法反演森林冠层高度存在低估现象(r=0.11,bias=-26.20 m,RMSE=7.16 m),地面相位优化的三阶段算法、纯体散射复相干优化的三阶段算法、低估补偿改进的三阶段算法反演森林冠层高度的估测精度较经典三阶段算法提高,其中低估补偿改进的三阶段算法反演森林冠层高度的改善效果最佳(r=0.79,bias=-1.69 m,RMSE=2.56 m);思茅松纯林的估测效果(r=0.81,bias=1.40 m,RMSE=2.27 m)优于思茅松混交林(r=0.72,bias=-3.09 m,RMSE=2.87 m)。【结论】相比经典三阶段反演算法,基于TanDEM-X SAR数据的改进三阶段反演算法估测精度更高。
【Objective】The random-volume-over-ground(RVoG)model has been widely used to estimate forest height through polarimetric and interferometric synthetic aperture radar(PolInSAR)technology,and the three-stage inversion algorithm is a geometrical method to solve the RVoG model to estimate forest canopy height.However,the conventional three-stage algorithm,solving the ground phase and the volume coherence based on the polarization scattering features of L and P bands,could not accurately estimate the ground phase and the volume coherence with TanDEM-X data of X-band.In practical application,the establishment conditions of the model were difficult to strictly meet,and due to the influences of terrain,the accuracy of tree overestimation was not high.In order to improve the estimation accuracy,an improved method of forest height retrieval based on RVoG model and TanDEM-X SAR data was proposed in this paper.【Method】The research object was Pinus kesiya var.langbianensis pure forest and P.kesiya var.langbianensis mixed forest in Simao district of Pu’er city.The experimental method included:(1)the classical three-stage inversion algorithm,(2)the improved three-stage inversion algorithm with the ground phase optimization,(3)the improved three-stage inversion algorithm with the volume coherence optimization and(4)the improved three-stage inversion algorithm with underestimation compensation.【Result】The results showed that:the forest height of method(1)based on RVoG model was underestimated(r=0.11,bias=-26.20 m and RMSE=7.16 m);the estimation accuracy of method(2),(3)and(4)was better than that of method(1),in which method(4)was the best(r=0.79,bias=-1.69 m and RMSE=2.56 m);the estimation effect of P.kesiya var.langbianensis pure forest(r=0.81,bias=1.40 m and RMSE=2.27 m)was better than that of P.kesiya var.langbianensis mixed forest(r=0.72,bias=-3.09 m and RMSE=2.87 m).【Conclusion】Compared with the classical three-stage inversion algorithm,the improved three-stage inversion algorithm based on TanDEM-X SAR data might have a better accuracy in this paper.
作者
张国飞
章皖秋
岳彩荣
Zhang Guofei;Zhang Wanqiu;Yue Cairong(Forestry College, Southwest Forestry University Kunming 650224)
出处
《林业科学》
EI
CAS
CSCD
北大核心
2022年第4期152-164,共13页
Scientia Silvae Sinicae
基金
国家自然科学基金项目(42061072)
云南省科技厅重大科技专项子课题(202002AA00007-015)
云南省教育厅项目(2018JS330)。
关键词
TanDEM-X
极化合成孔径雷达干涉测量
森林冠层高度
三阶段反演算法
RVoG模型
TanDEM-X
PolInSAR(polarimetric and interferometric synthetic aperture radar)
forest canopy height
three-stage inversion algorithm
RVoG(random-volume-over-ground)model