TOWARDS MORE RESILIENT SMART CITIES: MT-InSAR MONITORING OF URBAN INFRASTRUCTURE USING MACHINE LEARNING TECHNIQUES

Abstract

Abstract. Global climate change makes the maintenance and resilience of cities one of the greatest challenges facing civilization. Constant monitoring is essential work to determine if the city and its infrastructure are well preserved. This monitoring has become increasingly simple and affordable, thanks to the advancement of technology, so the current trend is to systematically monitor the entire city (this concept is the paradigm of the smart city). There are several methods for infrastructure monitoring including GPS, mobile-mapping, video-surveillance, etc. However, this type of method has a series of disadvantages, such as the impossibility of obtaining large-scale data or the unavailability of information of the previous, current or after state that an event has occurred in the study-area. This can be solved with monitoring based on satellite images, since these have historical and constant coverage over time, with good resolution to identify urban structures and cover large study areas. The use of satellite radar images through MT-InSAR is booming because it is a powerful remote sensing technique capable of detecting displacements on the earth's surface. This technique can be combined with Machine-Learning to perform predictive analysis in urban environments and detect infrastructure failures. This predictive monitoring capable of anticipating risks is one of the objectives of the new urbanism. For this reason, this work analyzes the collapse of a pier, which occurred in Vigo, a city in NW Spain, through radar satellite images (Sentinel-1), MT-InSAR and Machine-Learning. The result is the possibility of anticipating structural failures thanks to the predictive monitoring.