Earthquake damage assessment in urban area from Very High Resolution satellite data

The use of remote sensing within the domain of natural hazards and disaster management has become increasingly popular, due in part to increased awareness of environmental issues, including climate change, but also to the improvement of geospatial technologies and the ability to provide high quality imagery to the public through the media and internet. As technology is enhanced, demand and expectations increase for near-real-time monitoring and images to be relayed to emergency services in the event of a natural disaster. During a seismic event, in particular, it is fundamental to obtain a fast and reliable map of the damage of urban areas to manage civil protection interventions. Moreover, the identification of the destruction caused by an earthquake provides seismology and earthquake engineers with informative and valuable data, experiences and lessons in the long term. An accurate survey of damage is also important to assess the economic losses, and to manage and share the resources to be allocated during the reconstruction phase. Satellite remote sensing can provide valuable pieces of information on this regard, thanks to the capability of an instantaneous synoptic view of the scene, especially if the seismic event is located in remote regions, or if the main communication systems are damaged. Many works exist in the literature on this topic, considering both optical data and radar data, which however put in evidence some limitations of the nadir looking view, of the achievable level of details and response time, and the criticality of image radiometric and geometric corrections. The visual interpretation of optical images collected before and after a seismic event is the approach followed in many cases, especially for an operational and rapid release of the damage extension map. Many papers, have evaluated change detection approaches to estimate damage within large areas (e.g., city blocks), trying to quantify not only the extension of the affected area but also the level of damage, for instance correlating the collapse ratio (percentage of collapsed buildings in an area) measured on ground with some change parameters derived from two images, taken before and after the earthquake. Nowadays, remotely sensed images at Very High Resolution (VHR) may in principle enable production of earthquake damage maps at single-building scale. The complexity of the image forming mechanisms within urban settlements, especially of radar images, makes the interpretation and analysis of VHR images still a challenging task. Discrimination of lower grade of damage is particularly difficult using nadir looking sensors. Automatic algorithms to detect the damage are being developed, although as matter of fact, these works focus very often on specific test cases and sort of canonical situations. In order to make the delivered product suitable for the user community, such for example Civil Protection Departments, it is important to assess its reliability on a large area and in different and challenging situations. Moreover, the assessment shall be directly compared to those data the final user adopts when carrying out its operational tasks. This kind of assessment can be hardly found in the literature, especially when the main focus is on the development of sophisticated and advanced algorithms. In this work, the feasibility of earthquake damage products at the scale of individual buildings, which relies on a damage scale recognized as a standard, is investigated. To this aim, damage maps derived from VHR satellite images collected by Synthetic Aperture Radar (SAR) and optical sensors, were systematically compared to ground surveys carried out by different teams and with different purposes and protocols. Moreover, the inclusion of a priori information, such as vulnerability models for buildings and soil geophysical properties, to improve the reliability of the resulting damage products, was considered in this study. The research activity presented in this thesis was carried out in the framework of the APhoRISM (Advanced PRocedures for volcanIc Seismic Monitoring) project, funded by the European Union under the EC-FP7 call. APhoRISM was aimed at demonstrating that an appropriate management and integration of satellite and ground data can provide new improved products useful for seismic and volcanic crisis management.