基于ICESat-GLAS数据和回波仿真原理识别森林类型

【目的】利用星载激光雷达波形数据对森林类型识别时,受地形、噪声和林层结构等因素影响,针叶林、阔叶林和混交林森林类型识别精度较低,为提高森林类型识别精度,需提取与森林类型相关的波形特征参数。【方法】结合回波仿真原理与林分冠层特征对GLAS回波波形进行理论分析,提出了与森林类型相关的波形特征参数Rcafit1-47、K1-47,并与其他森林类型相关的波形特征参数进行联合,建立多种波形特征参数组合,用于森林类型识别。【结果】1)针叶林和阔叶林森林类型识别时,波形特征参数组合Rcafit1-47、K1-47森林类型总体识别精度为92.86%,优于另外两种波形特征参数组合AGS、MSGS和AGS、SGS森林类型总体识别精度;2)针叶林、阔叶林和混交林森林类型识别时,波形特征参数组合Rcafit1-47、K1-47森林类型总体识别精度为77.03%,低于另外两种波形特征参数组合AGS、MSGS和AGS、SGS森林类型总体识别精度;3)波形特征参数Rcafit1-47、K1-47与其他波形特征参数组合后能够提高森林类型识别精度,其中,Rcafit1-47、K1-47、AGS、MSGS参数组合森林类型识别精度最高,针叶林和阔叶林森林类型识别精度为94.64%,针叶林、阔叶林和混交林林分类型识别精度为89.19%。【结论】提出的波形特征参数Rcafit1-47和K1-47在针叶林和阔叶林森林类型识别方面具有明显优势,而且与其他波形特征参数组合后能够明显提高针叶林、阔叶林和混交林3种森林类型的识别精度。

Forest types identification based on ICESat-GLAS data and echo simulation principles

【Objective】When using forest-borne lidar waveform data to identify forest types, it was affected by factors such as topography, noise and forest structure, and forest type identification in coniferous forest, broad-leaved forest and mixed forest was low, in order to improve forest type identification accuracy, the waveform characteristic parameters related to the forest type need to be extracted.【Method】This paper combines the echo simulation principle with the canopy feature of the stands to theoretically analyze the GLAS received waveform, and proposes the waveform characteristic parameter Rcafit1-47, K1-47 related to the forest type.And combined with the waveform characteristics of other forest types, a variety of waveform feature parameters were established for forest type identification.【Result】1) When identifying forest types in broad-leaved forests and coniferous forests, the combination of waveform feature parameters Rcafit1-47, K1-47 forest type overall identification accuracy is 92.86%, which is better than the other two waveform feature parameters combination AGS, MSGS and AGS, SGS;2) When identifying forest types in broad-leaved forests, coniferous forests and mixed forests, waveform characteristic parameters The overall recognition accuracy of the combination of Rcafit1-47, K1-47 is 77.03%, which is lower than the other two waveform characteristic parameters combination AGS, MSGS and AGS, SGS;3) the waveform characteristic parameter Rcafit1-47, K1-47 combined with other waveform characteristic parameters can improve the accuracy of forest type identification. The identification accuracy of broad-leaved forest and coniferous forest stands is 94.64%, and the identification accuracy of broad-leaved forest, mixed forest and coniferous forest stands is 89.19%.【Conclusion】The waveform characteristic parameters Rcafit1-47, K1-47 proposed in this study have obvious advantages in forest type identification of broad-leaved forest and coniferous forest, and Combined with other waveform feature parameters, the forest type recognition accuracy can be improved.