X. Yang

Florida State University, Department of Geography, Tallahassee, Florida, USA


Because of their great environmental and ecological values, coastal marshes have been targeted in numerous interdisciplinary efforts that aim to discover patterns, relationships, and models leading to successful management and sustainable use of the resources. However, these efforts may be hindered because of the lack of adequate topographical data over coastal tidal areas.  Most of the currently available digital elevation data sets, such as GTOPO30, USGS DEMs, and NASA SRTM, are basically useless for coastal marsh areas where the terrain is featureless and of limited contrast.  Thus, there is a strong need to develop high-resolution topographic data sets for these environmentally sensitive areas. This article reports a research effort that aims to investigate the utility of integrating lidar and high-resolution imagery data for topographic and hydrological mapping with part of the North Inlet bay in Georgetown, South Carolina, USA as a case.  The study site is largely below 1 meter above the sea level, consisting of salt marshes behind barrier islands, scattered with numerous tidal creeks. The lidar data were collected in January 2003 and the entire data set was processed by using vegetation removal algorithms to create a bare-earth model.  The nominal posting density ranges from 1.5 to 5 meters. A triangulated irregular networks (TIN) model was derived from the lidar data through the Delaunay triangulation.  The TIN model was further converted into a grid-based surface model with 1-meter cell size.  Then, accuracy assessment was conducted by using a random sample of 1024 points through a cross-validation method.  The model was found to have a global RMSE of 0.05 meter, indicating very high vertical accuracy.  With this grid-based digital terrain model, hydrographical features were further extracted by using GIS-based hydrological modeling, which were validated by using a high-resolution ADAR (Airborne Data Acquisition and Registration) image covering the same area that was acquired in October 2000.  It is found that the stream networks extracted from the lidar-derived digital terrain model match the tidal creeks protracted from the ADAR image with 0.7-m spatial resolution quite well. This study demonstrates that lidar data are particularly suitable for topographic and hydrological mapping over low-lying areas but cautions should be paid upon the choice of threshold values in the GIS-based hydrological network modeling.