Detailed Digital Surface Model (dsm) Generation and Automatic Object Detection to Facilitate Modelling of Urban Flooding

This paper investigates the necessity of detailed surface representation in order to generate a feasible platform for the detailed simulation of urban runoff and surface flooding caused by heavy rainfalls. In recent years, urban flooding has become one of the major environmental hazards and it has caused damage to the buildings and properties over billion of Euros. Expansion of urban catchment areas mainly to cater for increased demand of accommodation, transportation and financial activities has resulted in complex and sprawling urban surface pattern. Behaviour of surface runoff and flooding caused by storms would also tend to be complex over this urban surface. Tackling excess storm water from huge urban catchment by sewerage system is not alone a feasible solution as there is always limitation of sewer pipe capacity. Provision of overland flow management within distinct surface features as well as good knowledge of surface characteristics is logically a potential option left to support detailed modelling and analysis of urban flooding. With this aim in mind, high resolution (<0.5 m) Digital surface model (DSM) has been generated from airborne LIDAR data. Special emphasis has been put on identifying each tiny element (Triangle/pixel) of generated DSM. A ‘dual reciprocal approach’ of model based and pixel based automatic object identification has been applied in order to successfully describe relevant surface characteristics in terms of rainfall-runoff and surface flow. In contrast to traditional LIDAR Digital Terrain Model (DTM) generation, attention will be paid to generate LIDAR DSM which is believed to simulate surface runoff more accurately as it considers rainfall interception by trees and direct input of water to sewer system from building roofs and other impervious surfaces.