基于窗函数的林区ICESat-GLAS波形数据消噪研究  

Denoising of Forest ICESat-GLAS Waveform Data Based on Window Function

作　　者：王爱娟[1] 邢艳秋[1] 邱赛[1] 王蕊[1] 

机构地区：[1]东北林业大学森林作业与环境研究中心,黑龙江哈尔滨150040

出　　处：《西北林学院学报》2016年第1期214-220,共7页Journal of Northwest Forestry University

基　　金：国家自然科学基金项目(41171274);中央高校基本科研业务费专项(DL12EB07)

摘　　要：在利用对地学激光测高系统(the Ice,Cloud and land Elevation-Geoscience Laser Altimeter System)数据估测森林结构参数时,需对原始波形进行去噪处理,以提高估测精度。以吉林汪清林区为例,提出了基于窗函数的林区GLAS数据消噪方法,选取了5种窗函数对GLAS数据进行消噪并比较其消噪精度。结果表明:窗函数对林区波形数据消噪具有较好的效果,窗函数消噪法的信噪比SNR最高为40.488 679,均方根误差RMSE最低为0.000 335;GLAS数据经窗函数消噪后能够合理地预测林区冠层高度,预测冠层高度与实测冠层高度的回归精度r从0.725增至0.820;本研究所选的几种窗函数中布拉克曼窗函数的消噪效果较好。结果说明了窗函数在对ICESat-GLAS波形数据消噪中具有很大的应用潜力。To improve the accuracy of estimating forest parameters by ICESat-GLAS （the Ice, Cloud, and land Elevation-Geoscience Laser Altimeter System） waveform data,the original waveform data needs to be denoised. In Wangqing forest area of Jinlin Province, based on the window function, methods of ICESat- GLAS waveform data denoising were proposed in the paper,and 5 window functions were carried out to de- noise the GLAS waveforms and analyze the denoising accuracy accordingly. The results showed that, the window function had a better effect on forest waveform data denoising,and the highest average signal-to- noise was 40. 488 679,and the lowest average root-mean-square error was 0. 000 335. The GLAS data de- noised by window function could reasonably predict the forest canopy height, and the regression precision R of the measured canopy height with estimated canopy height ranged from 0. 725 to 0. 820. The Braque Mann window function was better than others for the GLAS waveform denoising. These results suggested that the window function had a great application potential on the GLAS waveform denoising in the forests.

关 键 词：ICESAT/GLAS 林区波形 消噪 窗函数 

分 类 号：S771.8[农业科学—森林工程]