Urban flood modelling combining top-view LiDAR data with ground-view SfM observations

Remote Sensing technologies are capable of providing high-resolution spatial data needed to set up advanced flood simulation models. Amongst them, aerial Light Detection and Ranging (LiDAR) surveys or Airborne Laser Scanner (ALS) systems have long been used to provide digital topographic maps. Nowadays, Remote Sensing data are commonly used to create Digital Terrain Models (DTMs) for detailed urban-flood modelling. However, the difficulty of relying on top-view LiDAR data only is that it cannot detect whether passages for floodwaters are hidden underneath vegetated areas or beneath overarching structures such as roads, railroads, and bridges. Such (hidden) small urban features can play an important role in urban flood propagation. In this paper, a complex urban area of Kuala Lumpur, Malaysia was chosen as a study area to simulate the extreme flooding event that occurred in 2003. Three different DTMs were generated and used as input for a two-dimensional (2D) urban flood model. A top-view LiDAR approach was used to create two DTMs: (i) a standard LiDAR-DTM and (ii) a Filtered LiDAR-DTM taking into account specific ground-view features. In addition, a Structure from Motion (SfM) approach was used to detect hidden urban features from a sequence of ground-view images; these ground-view SfM data were then combined with top-view Filtered LiDAR data to create (iii) a novel Multidimensional Fusion of Views-Digital Terrain Model (MFV-DTM). These DTMs were then used as a basis for the 2D urban flood model. The resulting dynamic flood maps are compared with observations at six measurement locations. It was found that when applying only top-view DTMs as input data, the flood simulation results appear to have mismatches in both floodwater depths and flood propagation patterns. In contrast, when employing the top-ground-view fusion approach (MFV-DTM), the results not only show a good agreement in floodwater depth, but also simulate more correctly the floodwater dynamics around small urban feature. Overall, the new multi-view approach of combining top-view LiDAR data with ground-view SfM observations shows a good potential for creating an accurate digital terrain map which can be then used as an input for a numerical urban flood model.

Actions (login required)

View Item