无人机影像三维重建在沙丘形态监测中的应用

沙丘形态监测对开展土地沙化防治工作具有重大意义。应用无人机，分别在2016年7月、2017年3月、7月和9月对青海湖克土沙区4个沙丘进行影像采集，借助PhotoScan软件进行三维重建生成高分辨率的沙丘正射影像和DEM数据，使用ArcGIS软件获取沙丘形态参数并分析变化。实验结果表明：基于无人机影像监测沙丘形态，具有可行性和准确性，正射影像精度较高，DEM数据通过校正后可正确反映沙丘形态，为逐天高精度监测变化提供了可能。

Application of UAV image based 3D reconstruction in morphological monitoring of dunes

Morphological monitoring of dunes has an important value to control land desertification.Compared with using differential GPS to monitor dune's morphology,the method using small civilian drone was easier and had higher reliability.This method could not only save time and labor but also produce high resolution results.And the images could turn out high resolution dune's morphological data which was smaller than 1 m.However,few studies monitored dunes'morphology by using unmanned aerial vehicle to collect images.In this paper,unmanned aerial vehicle (UAV) images of four dunes in Ketu,eastern of Qinghai Lake,Qinghai Province,China were taken by DJI Phantom 4,in July 2016,March,July and September 2017,respectively.These images were 3D reconstructed by PhotoScan to produce high resolution digital elevation model (DEM) and orthophoto images.And also dunes' morphological data was collected in the same time by using UniStrong G10A real time kinematic platform (differential GPS) to test orthophoto images' accuracy and corrected these images if there were some geometric error.For the DEM,differential GPS's data was selected to test its accuracy.The results showed that these DEM's elevation data were average 90 to 110 m higher than differential GPS's data.But it could be fixed by using raster calculator tool in ArcGIS.Then through analyzing DEM and orthophoto images which were rectified through above steps by using ArcGIS,the dynamic change of dune's elevation,length and location of crest line,and dune's volume could be understood intuitively.The results showed as follows:firstly,the orthophoto image and DEM's resolution were smaller than 1 m.And the orthophoto images' horizontal position was close to differential GPS's data.But the DEM's elevation was much higher than differential GPS's altitude.This error could be revised and after that the DEM data could accurately reflect the true morphology of dunes.Secondly,the dynamic changes of sample dune's morphology characters from July 2016 to September 2017 were consistent with previous studies.The erosion areas were mainly concentrated in the west and south of the sample dune.It also concentrated in the border area of leeward slope and the area between sand dunes where it was in the north of the sample dune.Meanwhile,the sample dune moved to the downwind direction of main wind (southwest) and its crest line rotated from east to northeast.The position of the sample dune's crest changed little.It was just at the junction of the sample dune's four crest lines which were extracted from orthophoto images in different time,respectively.Thirdly,applying high resolution DEM and orthophoto images in morphological monitoring of dunes was feasible and accurate.It made daily monitoring possible.