<the goal of this research is to support the dynamic integration of heterogeneous agricultural data sources into a comprehensive knowledge base that can support farmers in their activities, present global and real-time information to researchers and interested parties and support automatic natural hazard detection systems. For that, we divide this work into the following items: (i), modeling a natural hazards ontology and their relationships with crops; (ii), the automatic classification of the following data sources according to the natural hazards ontology: the BSV, tweets, IoT data; and (iii), enriching the natural hazards ontology with new information extracted from the different data sources.