资助项目:中国科学院“一带一路”科技合作专项“中国科学院国际合作局对外合作重点项目”(131C11KYSB20160061); 中国科学院A类战略性先导科技专项“地球大数据科学工程”(XDA19070302)
第一作者:蒋广鑫(1995—),男,江苏省泰州市人,硕士研究生,主要研究地理信息科学。E-mail:1364282438@qq.com 通信作者:谢元礼(1968—),男,湖南衡阳人,副教授,主要研究自然地理学和测绘学。E-mail:37911220@qq.com
探究常用的3种DEM在中国不同地貌区的误差分布特征,为3种DEM在不同研究区的应用提供支撑。选取中国5类典型区(华北平原、黄土高原、青藏高原、塔里木盆地和云贵高原)为研究区域,以ICESat/GLAS GLAH14点数据为参考高程,选取目前最常用的3种DEM数据,SRTM 1 Arc-Second Global(SRTMGL1),SRTM Version 4(SRTM V4),ASTER GDEM V2(ASTER V2)作为精度评价数据,结合地形要素和地表覆盖,对3种DEM在5类研究区域的高程精度进行了对比分析。结果 表明:(1)在5类研究区中,总体上SRTMGL1数据精度优于SRTM V4和ASTER V2,其中以塔里木盆地3种DEM精度最高,云贵高原精度最低。(2)不同坡度区间内,SRTMGL1的数据精度皆高于ASTER V2和SRTM V4,平原地区(<10°),SRTM两类DEM精度相近,可互为补充,高植被覆盖、大坡度区域,SRTMGL1数据精度一般为最优,ASTER V2精度优于SRTM V4。(3)坡向对于3种DEM的影响较小,误差分布较均匀。(4)3种DEM在耕地和人造表面覆盖区域的精度最高,在林地覆盖区域,DEM精度最低。(5)不同地貌对3种DEM的精度具有较大的影响。SRTMGL1精度最优且受地形和地表覆盖影响最小,其次,SRTM V4在平原地区精度较好,受地形影响最大,ASTER V2在植被覆盖的地貌复杂区域精度较好,两者互有优劣。
This study explored the error distribution characteristics of 3 commonly used DEMs in different landform areas in China to support for application of 3 DEMs in different research area. This study selected 5 typical regions of China(North China Plain, Loess Plateau, Qinghai-Tibet Plateau, Tarim Basin and Yunnan-Guizhou Plateau)as research areas, with ICESat/GLAS GLAH14 points data as the reference elevation, and 3 most commonly used DEM data, SRTM 1 Arc-Second Global(SRTMGL1), SRTM Version 4(SRTM V4), ASTER GDEM V2(ASTER V2)were selected as the accuracy evaluation data. Combined with terrain factors and land coverage, a comparative analysis about the elevation accuracy of the 3 DEMs in the 5 types of research areas was made. The results show that:(1)in the 5 types of research areas, the SRTMGL1 data accuracy is better than SRTM V4 and ASTER V2 in general, among which the 3 kinds of DEM in the Tarim Basin have the highest precision and that in the Yunnan-Guizhou plateau has the lowest precision;(2)the data accuracy of SRTMGL1 is higher than that of ASTER V2 and SRTM V4 in different slopes; the plain area(<10°), two types of SRTM DEM have similar precision, which can complement each other; in high vegetation coverage and large slope area, SRTMGL1's data accuracy is generally optimal, ASTER V2's accuracy is close to SRTMGL1's, and SRTM V4 is the most affected by slope and has the lowest accuracy;(3)the aspect has little effect on the 3 types of DEM and can be ignored;(4)the 3 types of DEM have the highest precision in the area covered by cultivated land and artificial surface, and the DEMs have the lowest precision in the forest coverage area;(5)different geomorphological divisions have the greater impact on the accuracy of the 3 DEMs. SRTMGL1 has the best accuracy and is least affected by terrain and surface coverage. Secondly, SRTM V4 has better accuracy in the plain area and is most affected by the terrain, ASTER V2 has better accuracy in the area with complex landforms and covered by vegetation, and both have their own advantages and disadvantages.