The estimation of underwater features of channel bed surfaces without the use of bathymetric sensors results in very high levels of uncertainty. A revised approach enabling an automatic extraction of the wet areas to ...The estimation of underwater features of channel bed surfaces without the use of bathymetric sensors results in very high levels of uncertainty. A revised approach enabling an automatic extraction of the wet areas to create more accurate and detailed Digital Terrain Models (DTMs) is here presented. LiDAR-derived elevations of dry surfaces, water depths of wetted areas derived from aerial photos and a predictive depth-colour relationship were adopted. This methodology was applied at two different reaches of a northeastern Italian gravel-bed river (Tagliamento) before and after two flood events occurred in November and December 2010. In-channel dGPS survey points were performed taking different depth levels and different colour scales of the river bed. More than 10,473 control points were acquired, 1107 in 2010 and 9366 in 2011 respectively. A regression model that calculates channel depths using the correct intensity of three colour bands (RGB) was implemented. LiDAR and water depth points were merged and interpolated into DTMs which features an average error, for the wet areas, of ±14 cm. The different number of calibration points obtained for 2010 and 2011 showed that the bathymetric error is also sensitive to the number of acquired calibration points. The morphological evolution calculated through a difference of DTMs shows a prevalence of deposition and erosion areas into the wet areas.展开更多
文摘The estimation of underwater features of channel bed surfaces without the use of bathymetric sensors results in very high levels of uncertainty. A revised approach enabling an automatic extraction of the wet areas to create more accurate and detailed Digital Terrain Models (DTMs) is here presented. LiDAR-derived elevations of dry surfaces, water depths of wetted areas derived from aerial photos and a predictive depth-colour relationship were adopted. This methodology was applied at two different reaches of a northeastern Italian gravel-bed river (Tagliamento) before and after two flood events occurred in November and December 2010. In-channel dGPS survey points were performed taking different depth levels and different colour scales of the river bed. More than 10,473 control points were acquired, 1107 in 2010 and 9366 in 2011 respectively. A regression model that calculates channel depths using the correct intensity of three colour bands (RGB) was implemented. LiDAR and water depth points were merged and interpolated into DTMs which features an average error, for the wet areas, of ±14 cm. The different number of calibration points obtained for 2010 and 2011 showed that the bathymetric error is also sensitive to the number of acquired calibration points. The morphological evolution calculated through a difference of DTMs shows a prevalence of deposition and erosion areas into the wet areas.
