Data Semantics for Earth Observation: A technical guide to ontology-based integration for environmental data monitoring

Abstract

The integration of semantic technologies and geospatial information, especially when introducing information from Earth observation (EO) satellites, is significantly transforming methods for monitoring and diagnosing the environment and its most critical processes. This integration provides robust, cutting-edge analytical methods, thanks to the availability of more abundant and affordable data, which improves decision-making processes. This research proposes a technical tutorial guide for the application of an ontology system architecture built on semantic ontologies for remote sensing (RS) parameters, ground-based reference measurements, and observations acquired from uncrewed aerial vehicles (UAVs), enabling their joint use by scientists in the field. Although the work was ­originally developed using the Valencia Anchor Station (VAS) database, the system architecture ensures high adaptability and scalability, making it suitable for deployment across diverse station types and for application in other study areas. Through the application of linked open data (LOD) principles, the VAS ontology model integrates 1) EO-based data from Copernicus Sentinel-1, -2, and -3 satellites and other missions such as SMOS (Soil Moisture and Ocean Salinity), SMAP (Soil Moisture Active Passive), Landsat-8, -9, Terra and Aqua MODIS (Moderate Resolution Imaging Spectroradiometer), and so on; 2) observations from UAVs; 3) in situ data from sensors networks; and 4) auxiliary information from land cover classes, administrative units, and other features of interest to enhance environmental monitoring. The ontology is designed to represent the relationships between these elements, enabling systematic organization and retrieval of environmental data. The methodology is highly adaptable and can be extended to a wide range of applications, including validation of satellite RS products, precision agriculture, urban planning or disaster management. The potential of this approach to unify diverse data sources and streamline complex analytical processes is highly significant. Given the large number of ontologies capable of handling large volumes of data, the VAS ontology is presented as an illustrative example of their potential. Among the large number of parameters of all kinds involved, as an example application, this article refers specifically to the analysis of changes in the fraction of absorbed photosynthetically active radiation (FAPAR) and related vegetation characteristics at the VAS for the period 2018–2023. The proposed ontology has been tested in the Utiel-Requena region (Valencia, Spain), serving as a pilot area to validate the effectiveness and applicability of the architecture of the proposed approach. Fusion analytical data layers use a centralized data architecture, which supports real-time visualization and the capability of analyzing temporal trends and environmental variability. Validation tests assess the correctness and strength of the model, providing simple access to structured environmental data via an REST (Representational State Transfer) application programming interface (API) built using Spring Boot. The findings demonstrate the effectiveness of field, UAV, and satellite data fusion for multiscale analysis (spatial, temporal, and spectral), offering a potential and versatile tool for the integration of diverse data sources in a wide range of applications within the EO community.