DEM随机噪声误差去除方法研究——以黄土沟沿线地形剖面为例

发布时间：2018-08-17 12:27

【摘要】：在黄土高原数字地形分析中,基于DEM提取的地形参数和地貌对象是研究黄土地形地貌特征的基础。但在使用这些对象和参数进行小流域尺度的数字地形分析时,DEM数据中隐含的噪声在一定程度上会影响分析结果的可靠性。该文以分别位于陕北黄土丘陵沟壑区(绥德)和黄土塬区(淳化)的两个小流域为试验区,以国家基础地理数据库5m分辨率DEM为实验数据,以黄土地貌中最具典型特征的沟沿线为研究对象,对其地形剖面所隐含噪声的特征、提取及去除方法进行了研究。实验中采用EMD-MFDFA和EEMD-MFDFA两种组合方法对上述两个小流域黄土沟沿线地形剖面数据进行了分析。实验结果显示,沟沿线地形剖面数据噪声与地形复杂程度有很大关系;同EMD-MFDFA方法相比,EEMD-MFDFA对去除噪声成分具有更好的效果;EEMD-MFDFA方法在上述两个样区的沟沿线原始剖面与降噪后重构地形剖面的RMSE分别为2.053m和3.188m。
[Abstract]:In the digital terrain analysis of the Loess Plateau, the terrain parameters and geomorphological objects extracted based on DEM are the basis of studying the landform and geomorphological characteristics of Loess Plateau. However, the noise hidden in Dem data will affect the reliability of the analysis results to a certain extent when these objects and parameters are used for digital terrain analysis on a small watershed scale. In this paper, two small watersheds located in the Loess Hilly and Gully region (Suide) and the Loess Plateau (Chunhua) in Northern Shaanxi are taken as the experimental areas. The 5 m resolution DEM of the national basic geographic database is taken as the experimental data. In this paper, the features, extraction and removal methods of the hidden noise in the topographic profile along the gully with the most typical characteristics in the loess landform are studied. In the experiment, the topographic profile data along the loess gully of the two small watersheds are analyzed by EMD-MFDFA and EEMD-MFDFA. The experimental results show that the noise of topographic profile data along the gully is closely related to the complexity of the terrain. Compared with the EMD-MFDFA method, the EEMD-MFDFA has a better effect on noise removal. The RMSE of EEMD-MFDFA is 2.053m and 3.188m for the original section along the trench and the reconstructed terrain profile after noise reduction, respectively.
【作者单位】：南京师范大学地理科学学院虚拟地理环境教育部重点实验室;南京晓庄学院环境科学学院;
【基金】：国家自然科学基金项目(41671389、41471316、41401441)
【分类号】：P208

【相似文献】