Water surface mapping from airborne laser scanning using signal intensity and elevation data

Abstract

In recent years airborne laser scanning (ALS) evolved into a state-of-the-art technology for topographic data acquisition. We present a novel, automatic method for water surface classification and delineation by combining the geometrical and signal intensity information provided by ALS. The reflection characteristics of water surfaces in the near-infrared wavelength (1064 nm) of the ALS system along with the surface roughness information provide the basis for the differentiation between water and land areas. Water areas are characterized by a high number of laser shot dropouts and predominant low backscatter energy. In a preprocessing step, the recorded intensities are corrected for spherical loss and atmospheric attenuation, and the locations of laser shot dropouts are modeled. A seeded region growing segmentation, applied to the point cloud and the modeled dropouts, is used to detect potential water regions. Object-based classification of the resulting segments determines the final separation of water and non-water points. The water-land-boundary is defined by the central contour line of the transition zone between water and land points. We demonstrate that the proposed workflow succeeds for a regulated river (Inn, Austria) with smooth water surface as well as for a pro-glacial braided river (Hintereisfernerbach, Austria). A multi-temporal analysis over five years of the pro-glacial river channel emphasizes the applicability of the developed method for different ALS systems and acquisition settings (e.g. point density). The validation, based on real time kinematic (RTK) global positioning system (GPS) field survey and a terrestrial orthophoto, indicate point cloud classification accuracy above 97% with 0·45 m planimetric accuracy (root mean square error) of the water-land boundary. This article shows the capability of ALS data for water surface mapping with a high degree of automation and accuracy. This provides valuable datasets for a number of applications in geomorphology, hydrology and hydraulics, such as monitoring of braided rivers, flood modeling and mapping.

LiDAR derived ecological integrity indicators for riparian zones: Application to the Houille river in Southern Belgium/Northern France

“…The mean distance between the LiDAR wetted channel extent and the field measured wetted channel extent (1418 points) was −0.71 m (based on signed values) and 0.89 m (based on absolute values), with a RMSE of 1.29 m. These results show a systematic underestimation of the wetted channel extent, which is mostly due to the rasterization process (as reported by Höfle et al (2009)). The overall classification accuracy is 76.8%, with very high user's accuracy value (97.7%) and lower producer's accuracy value (78.1%).…”

“…Nevertheless, some returns were obtained from woody debris or gravel bed, pre-classified as "water" by the service provider. In accordance with the work of Höfle et al (2009), the methodology developed here takes advantage of those very low point density areas in the LiDAR point cloud by localising them in a "Number of 'ground' returns" raster (Fig. 3).…”

“…Most of the previous studies found in the literature on the use of LiDAR data in riparian zone (see Table 2) relate to topographic and geomorphic applications (Colson et al, 2006;Hilldale and Raff, 2008;Höfle et al, 2009;Hohenthal et al, 2011;Kinzel et al, 2007;Notebaert et al, 2009). Previous studies also show the potential of LiDAR data to describe riparian vegetation; but most of these are focused on forest applications (Farid et al, 2006(Farid et al, , 2008Greenberg et al, 2012;Wasser et al, 2013).…”

Use of LIDAR in landslide investigations: a review

“…lithologies) based on the laser reflectivity because of the various causes of intensity variations. This has been applied with good results to the classification of river bed at different epochs by identifying water-land boundary (Höfle et al 2009). Some research groups are now looking at further refinement of LIDAR imaging coupled with other remote sensing techniques.…”

Remote sensing for biodiversity monitoring: a review of methods for biodiversity indicator extraction and assessment of progress towards international targets

“…Lang and McCarty (2009) distinguished wetland inundation below the forest canopy from non-inundated and transitional areas with overall accuracy up to 96.3%. Other applications include upland swamp boundary detection (Jenkins and Frazier 2010), and river (Höfle and Vetter 2009) or tidal (Brzank et al 2008a,b;Schmidt and Soergel 2013) water mapping. In addition, Lband and C-band SAR data are effectively used in wetland and mangrove characterization (Bwangoy et al 2010;Evans et al 2010;Lang et al 2008;Kumar and Patnaik 2013 Pu et al (2010), hyperspectral Hyperion data surpassed the same or higher spatial resolution data of Advanced Land Imager (ALI), Landsat TM, and IKONOS in seagrass habitat mapping.…”