Precise models of the impact of explosions in urban environments provide novel and valuable information in disaster management for developing precautionary, preventive and mitigating measures. Yet to date, no methods enabling accurate predictions of the process and effect of detonations at particular locations exist. We propose a novel approach mitigating this gap by combining state-of-the-art methods from photogrammetric 3D reconstruction, semantic segmentation and computational based numerical simulations. In a first step, we create an accurate urban 3D reconstruction from georeferenced aerial images. The resulting city model is then enriched with semantic information obtained from the original source images as well as from registered terrestrial images using deep neural networks. This allows for an efficient automatic preparation of a 3D model suitable for the use as a geometry for the numerical investigations. Using this approach, we are able to provide recent and precise models of an area of interest in an automated fashion. Within the model, we are now able to define the explosive charge size and location and simulate the resulting blast wave propagation using CFD simulation. This provides a full estimation for the expected pressure propagation of a defined charge size. From these results, arising damages and their extent, as well as possible access routes or countermeasures, can be estimated. Using georeferenced sources allows for the integration and utilization of simulation results into existing geoinformation systems of disaster management units, providing novel inputs for training, preparation and prevention. We demonstrate our proposed approach by evaluating expected glass breakage and expected damages impairing the structural integrity of buildings depending on the charge size using a 3D reconstruction from aerial images of an area in the inner city of Graz, Austria.
展开▼
