地球信息科学学报 ›› 2011, Vol. 13 ›› Issue (1): 118-125.doi: 10.3724/SP.J.1047.2011.00118

• 遥感技术与应用 • 上一篇    下一篇

基于街道空间数据及GPS测量的SRTM-DEM 校正和插值细化

肖飞1,2, Parrot J. F.2, 杜耘1, 凌峰1   

  1. 1. 中国科学院测量与地球物理研究所,武汉 430077;
    2. LAGE, Instituto de Geografia,UNAM,Mexico City 04510, Mexico
  • 收稿日期:2010-08-25 修回日期:2011-01-19 出版日期:2011-02-25 发布日期:2011-02-25
  • 作者简介:肖飞(1978-),男,陕西人,博士,研究方向:地形地貌数字分析及GIS资源环境应用。E-mail:xiaof@whigg.ac.cn
  • 基金资助:

    国家自然科学基金项目(40801045);湖北省自然科学基金项目(2009CDB138);中国科学院知识创新项目(0609211120);国家自然科学基金项目(40801186)共同资助。

Correcting and Downscaling SRTM DEM Using Auxiliary Street Map and GPS Data

XIAO Fei1,2, PARROT J.F.2, DU Yun1, LING Feng1   

  1. 1. Institute of Geodesy &|Geophysics, Chinese Academy of Sciences, Wuhan 430077, China;
    2. LAGE, Instituto de Geografia, UNAM, Mexico City 04510, Mexico
  • Received:2010-08-25 Revised:2011-01-19 Online:2011-02-25 Published:2011-02-25

摘要: 针对城市地区SRTM 数字高程模型,本文引入街道空间分布信息作为辅助数据,并结合GPS实地测量,对DEM中具有空间相关性的误差部分进行修正,从而尝试在一定程度上提高SRTM数据的垂直精度和空间分辨率。本文以墨西哥城Xico地区为例,解析SRTM DEM像元分辨率单元内地物表面平均高程值的组成结构;分析形成DEM高程值的地物组分之间的空间关系并根据GPS实地测量数据进行检验;探讨SRTM DEM中具有空间相关性系统偏差的产生原因;计算街道空间分布和SRTM 数据中与其具有空间相关性的系统误差之间的定量关系;并评估SRTM数据在该区的总体高程偏差。根据辅助数据与DEM系统误差之间的空间关系,通过GIS空间运算实现SRTM DEM的校正和细化。结果表明,与原始SRTM数据相比,校正细化后的DEM其整体高程偏差,以及与街道建筑结构相关的系统误差,均能够得到一定程度的校正和改进,并能较好地反映城区实际地面高程和城区建筑结构信息。在其他地形数据缺乏情况下,尽管计算结果受SRTM自身数据精度所限,但利用含有空间定位信息的辅助数据,对现有较低空间分辨率的DEM进行误差校正及细化,仍可在一定程度上消减与辅助数据相关的空间误差,提高DEM数据精度,从而为相关研究提供更精确的基础地形资料。

关键词: SRTM, DEM, 空间分辨率, 误差, 细化

Abstract: We developed and tested a geo-statistical methodology to correct the spatially correlated errors of SRTM DEM and downscale the spatial resolution of the DEM using auxiliary street map and GPS data in city areas. The methodology was based on geo-statistics and spatial analysis techniques in GIS. The calculation procedures were illustrated using a case study in the Xico area of Mexico City. Spatial structures of all the surface features contributing the elevation value within a single SRTM resolution cell were separated into basic components. Then the syntagmatic relationships among the basic components were evaluated and verified using GPS survey data in the study area. Afterwards, spatial relations between the structure of the basic components and the errors surface of the DEM were analyzed using GIS spatial analytical methods. From spatial analysis, the error surface of SRTM DEM proved to be spatially correlated with spatial patterns of streets in the study area. A global error of the DEM was also identified through the above process. Accordingly, the spatially correlated random errors and the global error can be located and corrected by using the auxiliary street map and GPS data. Since the auxiliary street map has finer resolution than the SRTM DEM and is spatially correlated with the error surface of the DEM, the coarse-resolution DEM in the city area can be downscaled to a finer resolution DEM according to the spatial relations between DEM and the auxiliary street map using geo-statistical methodology. The result shows that the downscaled DEM has better performance on the representation of real topography and urban structure. Even though the result was determined by the accuracy of the original DEM, the methodology presented here is effective at downscaling of spatial resolution and the correction of the spatially correlated errors for SRTM in city areas, and could be helpful in topographical information collection provided that there are deficient of other higher resolution DEMs.

Key words: SRTM, DEM, spatial resolution, error, downscale