Remote glacier monitoring through semantic fusion of geographic and contextual data

Glacier melting, due to climate change, is a growing concern with many implications for the planet. Although remote sensing technology offers valuable solutions, the data acquired is often heterogeneous, varying in format and originating from different sensors and sources. Harmonizing this heterogeneous data is essential to ensure compatibility and integration. This paper introduces an integrated model that combines Machine Learning methods for satellite image segmentation with semantic web technologies for knowledge base construction aimed at retrieving relevant data concerning glacier monitoring implications. An ad-hoc ontology was developed to model the knowledge base about the glacier domain. At the same time, semantic segmentation allows the elicitation of relevant features combined with contextual meteorological data to populate the knowledge base. The synergy between ontology-based annotation and Deep Learning techniques for segmenting remote sensing images enables a more comprehensive assessment of glacial health, facilitating the retrieval of specific information via semantic queries. The effectiveness of the proposed system is shown by evaluating the performance of the Deep Learning models and the consistency and robustness of the ontology in processing complex queries. • Glacier melting due to climate change poses serious environmental threats. • Harmonizing remote sensing data is key to ensuring compatibility and usability. • ML segmentation combined with Semantic Technologies enables knowledge base creation. • Ontologies and deep learning enable glacial assessment through SPARQL querying.