- Project report
The aim of the study project HydroChange³ was to carry out an investigation of the fluvial dynamics of the Neckar river and its gravel bar near Ilvesheim. For this purpose, terrestrial laser scanning (TLS) was used to acquire the necessary data; appropriate methods for data processing were applied. The whole project was divided into five different groups and tasks building upon each other.
In order to receive the necessary topographical information in form of a DTM for further study aims, an automated workflow for DTM generation was applied. The basis for this is a point cloud, captured via TLS, which was registered and geo-referenced using the RiSCANPRO software from the laser scanner manufacturer Riegl; additionally the registered point cloud was adjusted to previous point clouds from 2012 and 2011. During DTM generation, objects such as vegetation were removed using a morphological filter, which iteratively erodes such objects if a predetermined height difference is exceeded. Besides the requirement of a height difference, considering the echo ratio within the erosion process leads to the best results. Compared to reference data, the best result shows a RMSE of 9.8 cm; similar results (10.9 cm) were received in the study of HÄMMERLE et al. (2013). Overall, the errors increase with dense vegetation by trend and are lowest on the gravel bar without vegetation. However, for final evidence, more reference data are necessary especially at areas with dense vegetation (e.g. inside the bush).
Since the multitemporal analysis shows a height difference of the gravel bar between 2012 and 2013 of 30 cm in average, the DTM's RMSE of 9.8 cm was also considered as a limiting factor as well as a possible displacement between the different years' point clouds. In addition to the deviation indicated by the DTM, the calculated height difference of 2012/2013 also shows differences compared to single measured reference data. Considering these values, the calculation of the gravel bar's volume balance indicates a high error potential, which is similar to the results of HÄMMERLE et al. (2013). However, the multitemporal analysis shows in general that the gravel bar is affected by ongoing fluvial processes and the calculation of the volume balance shows a trend for the processes occurring on the gravel bar. Furthermore the quite high erosion rate of 30 cm in average corresponds to an excessively high flow rate and a water stream with a high energy level during the major flood in May and June 2013. With regard to flood events, and especially the one in June 2013 in Germany, the necessity for flood predictions via hydraulic modeling is shown. For this purpose, high-resolution topographic data can improve such predictions. A comparison of five DTMs with different resolutions (SRTM, ASTER, ALS, TLS) relative to inundation area and depths shows significant differences. The two lowest resolution DTMs, SRTM (90x90m) and ASTER (30x30m), overestimated the total flooded area and did not trace the river course. Even though this data is publicly available, it is not suitable for hydraulic modeling in the detailed scale of our study. In contrast, ALS (10x10m and 1x1m) data brought up realistic results compared to the flood of 3rd June 2013. Analysis based on the DTMs derived from TLS (0.25x0.25m) provides most detailed performance. Due to the high cost and labor-intensive gathering of the data, it is sufficient and advising to perform hydraulic modeling over broad areas with ALS, and investigate only in smaller areas with TLS data.
Since hydraulic modeling can be improved by knowing the roughness values of the terrain, several methods were compared to derive such values. The TLS data could be used as expected. The Kinect data could not meet the preceding results, whereas SfM data proved to be very good. Further research could be inquired concerning the registration of the individual data within the global coordinate system, which was not possible during this campaign. The SfM data acquisition forms an effective upgrading of the 3D analysis toolbox. Working accurately and more detailed than TLS data, it should be tested on further surfaces, too.
In order to present the study project a video and web site were created. The video gives an insight about the whole study campaign by presenting the field work and the different research aims of the working groups. The web site is used to publish all research results and background information about the laser scanning technology; the video can be viewed on the web site as well.