Data Paper Zone II Versions EN4 Vol 4 (2) 2019
Download
DOM and DEM dataset of the Drôme River remotely sensed with a UAV (2005 – 2009)
 >>
: 2018 - 05 - 21
: 2018 - 08 - 17
: 2019 - 07 - 12
118 2 0
Abstract & Keywords
Abstract: In the past few years, a new low-cost, efficient photogrammetry mapping method called structure from motion (SfM) was applied to the high-spatial resolution three-dimensional topographic data collection. As the latest and a very significant advance in digital surveying, SfM has only been applied to the geosciences to produce DOM, three-dimensional (3D) point clouds, and DEM since 2011. SfM-derived products have been used to detect topographical changes and hence to infer dynamic processes in glacial, fluvial, coastal, hillslope, dryland, volcanic, and shallow underwater environments. In this study, historical imagery archives of 2005 – 2009 acquired from a UAV are processed by using SfM to create hyper-spatial resolution DOM and DEM. A statistical analysis indicates that the results have an overall accuracy of about 0.13 – 0.23m. Precise and detailed data obtained from UAV imagery and SfM photogrammetry are essential for describing fluvial geomorphology and also helpful to detect, monitor and predict fluvial geomorphic processes.
Keywords: fluvial remote sensing; UAV; SfM; DOM; DEM
Dataset Profile
English TitleA dataset of DOM and DEM of the Drôme River (2005 – 2009)
Corresponding authorZhang Yaonan (yaonan@lzb.ac.cn)
Data authorsCui Dandan, Kristell Michel, Hervé Piégay, Zhang Yaonan
Time range2005 – 2009
Geographical scope44°37′10″–44°39′40″N, 5°25′00″–5°26′50″E (a 5 km reach in Drôme River)
Spatial resolutionDOM: 0.1 m, DEM: 0.2 m/0.3 mData volume5.56 GB
Data format*.tif
Data service system<http://www.sciencedb.cn/dataSet/handle/639>
Sources of fundingInformatization Program of the Chinese Academy of Sciences: Construction and application of technology cloud on studies of environmental evolution in the cold region (XXH13506); National Natural Science Foundation of China (91125005); Incubation Foundation for Special Disciplines of the National Science Foundation of China (J1210003).
Dataset compositionThe dataset consists of two parts of data, one for DOM and the other for DEM. It comprises ten data files in total. Take data of the year 2006 as an example – the data files are recorded as 2006_DOM.tif and 2006 _DEM.tif.
1.   Introduction
In recent years, the integration of computer vision and traditional photogrammetry technology has brought a new topographic measurement technology called Structure from Motion (SfM). This method only needs multi-view photos of the target object and uses a high-efficiency feature matching algorithm to obtain high-quality three-dimensional topographic data of overlapping regions accurately, quickly and efficiently from photos. Since 2011, SfM has been used in the field of earth sciences. Accurate topographic data plays an important role in the study of river geomorphology.1.2. The resolution of most remote sensing images is difficult to meet the geomorphological characteristics of small and medium-scale rivers (river width <200 m).3.Compared with other traditional methods (such as total station, differential global positioning system, laser scanning, etc.), SfM has an irreplaceable advantage due to its low cost, efficient and true three-dimensional information in fluvial morphological monitoring and sediment budget studies.4.5.6.7.8.9. At present, in the study of fluvial geomorphology, SfM is a powerful supplement to three-dimensional remote sensing. SfM with obtaining high-resolution images by Unmanned Aerial Vehicle (UAV) can obtain high-precision topographic data and construct high-quality DEM to study the change of river topography quantitatively.10.11.12.
This paper used the high-resolution UAV images from 2005 to 2009 to understand the situation of the Drôme River. Based on SfM photogrammetry, the DOM and DEM were constructed with 3D terrain data processing technology which is suitable for the non-expert users. DEM was compared with differential GPS data to analyze the accuracy of archived UAV image-derived products and to maximize the potential of data applications.
2.   Data collection and processing
2.1   study area
The Drôme River (Fig. 1) is located in the south of the Alps, in the southeast of France, and belongs to the left tributary of the Rhône River. The Drôme River is 110 km long and has a drainage area of 1,640 km2. The 5 km reach of the Drôme River is selected as the study area. The width of the study reach channel is between 10 and 200 m. The riverbed is mainly composed of loose sandstone, gravel, and pebbles. The river channel is wide and shallow with little curvature. There are no dams along the river. The river is not fixed, and the migration is rapid13.. Adjacent to the channel, riparian vegetation is dense.


Figure 1   Study area
2.2   data collection and processing
The main steps of DOM and DEM generation from images is shown in Figure 2.


Figure 2   Workflow of the processing of DOM and DEM
The study area was monitored annually by the French National Center for Scientific Research (CNRS, Centre National de la Recherche Scientifique) from 2005 to 2009. These images are high -overlap, high-resolution, true-color digital images that cover the entire study area using a digital camera with a Pixy drone system. The image format is JPEG. See Table 1 for detail information on flight time, sensor, number of images, flight height, ground control point, and checkpoint.
Table 1   Overview of data collection and data processing
Time2005/05/23-272006/05/15-192007/05/21-252008/09/29-302009/06/16-19
SensorCanon G5Canon G5Canon G5Canon G9Sony DSLR-A350
Image resolution(cm)7.24.84.56.45.6
focal length(mm)7.27.287.430,35
Image number563647982387365
Flight height(m)200134197271373
Number of GCP23038015070120
Number of CP89104962935
DOM(m)0.10.10.10.10.1
DEM(m)0.20.20.20.30.3
X error(m)0.060.070.080.080.08
Y error(m)0.070.080.080.090.1
Z error(m)0.090.170.140.160.19
Overall error(m)0.130.20.180.20.23
Before the flight experiments, mark targets are evenly distributed along the river corridor, marked as a bright red area of 50 × 50 cm. In the field, we use the Tempo RTK5800 to obtain their spatial coordinates by receiving high-precision real-time differential signals. Among these points, a part of them is clearly identified on the image taken as ground control points to georeference, and the remaining points are used as checkpoints to evaluate the quality of the data product.
Agisoft PhotoScan software is a professional software for UAV images using SfM method to construct 3D models with geographic coordinates. This paper uses this software. Verhoever introduced the related algorithms in detail.14.The main processing steps are shown in Figure 2: (1) In order to ensure the accuracy of data product, preliminary quality detection can eliminate distortion, blur, and out of the study area images. The UAV pre-processed images were imported into PhotoScan. (2) Image alignment. This step calculated the overlapping image matching points and estimated the position of each image. Then, a sparse point cloud was generated. (3) Importing ground control points with precise geographic coordinates, transforming the point cloud from the image space coordinate system to the real world space coordinate system. Furthermore, optimizing the model and obtaining the real spatial position of the camera and the sparse point cloud. (4) Calculation of the depth information. A dense point cloud was generated. (5) DOM and DEM with spatial geographic coordinate information were exported in this final step. The resolution and projection type can be adjusted in this step.15.The resolution of the output data is shown in Table 1. The projection type is RGF93_Lambert_93.
3.   Sample description
The study area is the 5 km long river channel and the surrounding landscape of the Drôme River between Luc-en-Diois and Recoubeau-Jansac. The dataset consists of two parts of data, one is DOM, the other is DEM. There is a total of 10 Geotiff format data files, and the geographic coordinate system is RGF93_Lambert_93. The spatial resolution and error of the data can be seen in Table 1. With a spatial resolution of 0.1 m, the DOM with 8-bit integer pixel depth clearly shows the river landscapes such as river channel and vegetation. The DEM data pixel depth is 32-bit floating point type, and the spatial resolution of the different years is slightly different, which is 0.2 m or 0.3 m. Referring to Table 1 for details. The elevation of the entire area is between 480 and 580 m. Taken the 2006 data as an example, the DOM data is named 2006_DOM.tif, the DEM data is named 2006_DEM.tif, and the data result is shown in Figure 3.


Figure 3   DOM and DEM of 2006
4.   Quality control and assessment
The data set of this paper is mainly controlled by the following means:
1. Data source quality control. In order to generate high precision data, UAV images are inspected and selected to ensure that high quality images cover the entire study area with high overlap.
2. Quality control during processing. The UAV image is processed by the currently available popular commercial SfM software AgiSoft PhotoScan Professional Edition. At the same time, the ground control points with uniform distribution, sufficient quantity, and high precision are added as the georeferencing to register and optimize the point cloud model.
3. Data quality assessment. Select some points as checkpoints, and check the root mean square error (RMSE, Root Mean Square Error) in the X (longitude), Y (latitude), and Z (height) directions and overall error (Table 1). Statistical analysis of error can provide a clear understanding of DOM and DEM data quality.
5.   Value and significance
DOM and DEM datasets derived from UAV images can be used for mapping, river feature visualization, quantitative analysis and background images for other data. The data has a high spatial resolution can be used to analyze the forest vegetation coverage of the riverbank, and as an ecological model, hydrological model parameters to determine vegetation productivity, biomass and watershed analysis. Combined with the terrain data of different times in the same region, it is possible to accurately understand the dynamic change information of the river channel, analyze the variation of sediment transport in the river channel, scouring and silting of the beach, and monitoring the change of river topography.
6.   Usage notes
The 2005-2009 DOM and DEM datasets of Drôme River can be read and operated in popular GIS and remote sensing software such as ArcGIS, PhotoScan, SuperMap, ENVI, and ERDAS.
1.
PIEGAY H, KONDOLF G M, MINEAR J T, et al. Trends in publications in fluvial geomorphology over two decades: A truly new era in the discipline owing to recent technological revolution? [J]. Geomorphology, 2015, 248: 489-500.
2.
WOODGET A S, CARBONNEAU P E, VISSER F, et al. Quantifying submerged fluvial topography using hyperspatial resolution UAS imagery and structure from motion photogrammetry [J]. Earth Surface Processes and Landforms, 2015, 40(1): 47-64.
3.
KONDOLF G M, PIEGAY H. Tools in Fluvial Geomorphology [M]. John Wiley & Sons, 2016.
4.
MCKEAN J, ROERING J. Objective landslide detection and surface morphology mapping using high-resolution airborne laser altimetry [J]. Geomorphology, 2004, 57(3-4): 331-351.
5.
CAVALLI M, TAROLLI P, MARCHI L, et al. 2008. The effectiveness of airborne LiDAR data in the recognition of channel-bed morphology[J]. Catena, 73(3): 249-260.
6.
MORETTO J, DELAI F, RIGON E, et al. 2012. Assessing short term erosion-deposition processes of the Brenta River using LiDAR surveys[J]. WIT Transactions on Engineering Sciences 73, 149-160.
7.
FONSTAD M A, MARCUS W A. High resolution, basin extent observations and implications for understanding river form and process [J]. Earth Surface Processes and Landforms, 2010, 35(6): 680-698.
8.
THOMA D P, GUPTA S C, BAUER M E, et al. Airborne laser scanning for riverbank erosion assessment [J]. Remote sensing of Environment, 2005, 95(4): 493-501.
9.
NOTEBAERT B, VERSTRAETEN G, GOVERS G, et al. Qualitative and quantitative applications of LiDAR imagery in fluvial geomorphology[J]. Earth Surface Processes and Landforms, 2009, 34(2): 217-231.
10.
DIETRICH J T. Riverscape mapping with helicopter-based Structure-from-Motion photogrammetry [J]. Geomorphology, 2016, 252: 144-157.
11.
TAMMINGA A D, EATON B C, HUGENHOLTZ C H. UAS‐based remote sensing of fluvial change following an extreme flood event [J]. Earth Surface Processes and Landforms, 2015, 40(11): 1464-1476.
12.
MIŘIJOVSKÝ J, LANGHAMMER J. Multitemporal monitoring of the morphodynamics of a mid-mountain stream using UAS photogrammetry [J]. Remote Sensing, 2015, 7(7): 8586-8609.
13.
TOONE J, RICE S P, PIÉGAY H. Spatial discontinuity and temporal evolution of channel morphology along a mixed bedrock-alluvial river, upper Drôme River, southeast France: Contingent responses to external and internal controls [J]. Geomorphology, 2014, 205: 5-16.
14.
VERHOEVEN G. Taking computer vision aloft-archaeological three-dimensional reconstructions from aerial photographs with photoscan[J]. Archaeological prospection, 2011, 18(1): 67-73.
15.
Wei Zhanyu, Arrowsmith Ramon, He Honglin, et al.Accuracy analysis of terrain point cloud acquired by “structure from motion” using aerial photos[J].Seismology and Geology, 2015, 37( 2) : 636-648.
Data citation
Cui D, Zhang Y, Hervé P and Kristell M. DOM and DEM dataset of the Drôme River remotely sensed with a UAV (2005 – 2009). Science Data Bank, DOI: 10.11922/sciencedb.639(2019).
Article and author information
How to cite this article
Cui D, Zhang Y, Hervé P and Kristell M. DOM and DEM dataset of the Drôme River remotely sensed with a UAV (2005 – 2009). China Scientific Data 4(2019). DOI: 10.11922/csdata.2018.0032.zh.
Cui Dandan
Contribution: the processing of UAV images, the production of data products.
Ph.D candidate. research area: UAV remote sensing, fluvial geomorphology monitoring.
Zhang Yaonan
Contribution: academic guidance for remote sensing processing and application of UAVs.
yaonan@lzb.ac.cn
Ph.D, researcher, doctoral supervisor. research area: geoscience e-science research based on data, models, calculations and visualization, models and applications of high-performance computing in geosciences, scientific engineering science calculations and visual analysis.
Kristell Michel
Contribution: UAV remote sensing application guidance and flight plan design.
Ph.D, researcher, doctoral tutor; research area: river landscape.
Hervé Piégay
Contribution: planning and implementation of flight experiments, data collection, and processing.
engineer, associate researcher; research area: GIS.
Informatization Program of the Chinese Academy of Sciences: Construction and application of technology cloud on studies of environmental evolution in the cold region (XXH13506); National Natural Science Foundation of China (91125005); Incubation Foundation for Special Disciplines of the National Science Foundation of China (J1210003).
Publication records
Published: July 12, 2019 ( VersionsEN4
Released: Aug. 17, 2018 ( VersionsZH2
Published: July 12, 2019 ( VersionsZH3
References
中国科学数据
csdata