The influence of errors in Digital Terrain Models on flood flow routes

The modelling of flood inundation patterns on rural and urban river floodplains is making increasing use of high-resolution and high-accuracy Digital Elevation Models (DEMs) based on LiDAR (airborne laser scanning) data. LiDAR measurements are reported to be subject to root mean square errors of 0.15 m vertically and 0.50 m horizontally. The aim of the paper is to investigate the implications of these uncertainties in terms of the reliability of model results, concentrating on the modelling of overland flood flow using a 2D fully hydrodynamic model. A flood event caused by overflow from a river is simulated using three different DEMs: a benchmark error-free DEM, and 2 "altered" DEMs with added LiDAR errors of different magnitudes. These altered DEMs are a less accurate representation of reality, and also have an artificial roughness that is typical of LiDAR-based DEMs used in practice. The main conclusions are that: (1) this artificial roughness can significantly slow down the flow in the simulation, resulting in an incorrect representation of the flow dynamics, and (2) the LiDAR errors can have a significant impact on the routes taken by the flow, and on modelled water levels and velocities. © 2006 Taylor & Francis Group.