Innovative big data integrationand analysis techniques for urban hazard management

Abstract

Modern early warning systems (EWS) require sophisticated knowledge of natural hazards, the urban context and underlying risk factors to enable dynamic and timely decision making (e.g., hazard detection, hazard preparedness). Landslides are a common form of natural hazard with a global impact and are closely linked to a variety of other hazards. EWS for landslide prediction and detection relies on scienti c methods and models which require input from the time-series data, such as the earth observation (EO) and ancillary data. Such data sets are produced by a variety of remote sensing satellites and Internet of Things sensors which are deployed in landslide-prone areas. Besides, social media-based time-series data has played a signi cant role in modern disaster management. The emergence of social media has led to the possibility of the general public contributing to the monitoring of natural hazard by reporting incidents related to hazard events. To this end, the data integration and analysis of potential time-series data sources in EWS applications have become a challenge due to the complexity and high variety of data sources. Moreover, sophisticated domain knowledge of natural hazards and risk management are also required to enable dynamic and timely decision making about serious hazards. In this thesis, a comprehensive set of algorithmic techniques for managing high varieties of time series data from heterogeneous data sources is investigated. A novel ontology, namely Landslip Ontology, is proposed to provide a knowledge base that establishes the relationship between landslide hazard and EO and ancillary data sources to support data integration for EWS applications. Moreover, an ontology-based data integration and analytics system that includes human in the loop of hazard information acquisition from social media is proposed to establish a deeper and more accurate situational awareness of hazard events. Finally, the system is extended to enable an interaction between natural hazard EWS and electrical grid EWS to contribute to electrical grid network monitoring and support decision-making for electrical grid infrastructure management.