Semantic reasoning system for monitoring natural disasters based on GeoSPARQL ontology and satellite images study cases: Flood-prone areas of Shiraz city

Reliable and automatic classification of satellite images is of great importance for disaster management and climate change monitoring. In addition to recognizing objects and entities in satellite images, the ability to reason about these entities and subsequently respond to queries raised by human operators to guide decision-making processes is important. According to recent studies by climate researchers, various parts of Shiraz are exposed to flood risk. Therefore, the Shiraz flood scenario is real. So we have defined a disaster scenario where the central part of Shiraz is mostly covered by water. One of the main goals of this research is to show the geometry of the regions on the map, enabling the map to respond to questions related to the topology and neighborhood of the regions. In this research, we presented a framework to transfer satellite image data to an interactive map that is ready for mining. In order to obtain a searchable map directly from satellite data, a CNN classifier that is sensitive to image features is used to feed this framework with labelled regions. They show their capabilities in terms of route connectivity. Representing such features in an ontology that is an extension of the existing GeoSPARQL ontology enables the system to automatically search for classified areas based on specific criteria of areas, selected based on environmental status. We have shown how by semantically enriching the representation of regions in ontocity, we can enable the system to automatically find options for regions, thus improving search time, including region revision and co-routing. This SemCityMap framework can now be used as a tool for better decision-making and situational awareness.