Leveraging Cloud, Earth Observation and In-Situ Sensors for Agri-Environmental Monitoring and Policy Decision-Making

Cloud infrastructures play a significant role in delivering secure, scalable and efficient data processing for Earth Observation (EO) and agricultural management applications. As part of the ScaleAgData project, we present a hierarchical Agri-Environmental Monitoring Tool running on a private cloud infrastructure. The system combines data from EO, in-situ sensors and farm management information systems (FMIS), including parcel calendars, to provide farmers and policymakers multi-scale insights.  

The solution is cloud-based and designed with an underlying architecture that ensures both scalability and interoperability, leveraging OGC-compliant data formats where applicable. EO and in-situ data streams can be processed and analyzed efficiently with the help of containerized apps and microservices to facilitate modular development and simplify deployment. By using a web-based dashboard with hierarchical design, stakeholders can navigate from overviews at the municipal level to individual parcels. Aggregated summaries that comply with Common Agricultural Policy (CAP) criteria are useful to policymakers and farmers can get comprehensive parcel-level metrics to optimize irrigation, pesticide use and other agro-related activities.  

Specifically, the tool combines EO data to derive vegetation indices (e.g., NDVI, EVI) and other parameters requiring advanced processing for crop type classification. Furthermore, these datasets are enriched with in-situ sensor measurements (e.g. soil moisture, weather data) and farm logs managed within FMIS (irrigation schedule, pesticide usage). Parcel-level data (L1) is processed to generate statistics, which are then calibrated with nearby parcels data with similar properties and crop type(L2), serving as control level, and finally extrapolated to the municipal level (L3) using spatial averaging techniques  to provide indicators related to irrigation water, pesticide, fertilizer usage, etc.  Farm calendars stored within FMIS provide a reliable source of ground-truth data, enhancing the tool’s ability to validate aggregated metrics. The aggregation at L2 and L3 allows for the identification of regional trends and patterns in agricultural practices, empowering policymakers and stakeholders to implement targeted interventions at both levels, thereby promoting sustainable agriculture.   

This work showcases the potential of private cloud infrastructures to enhance agri-environmental monitoring by processing and integrating heterogeneous data streams (EO, in-situ sensors and farm log data) into a unified system. The system is being applied in diverse agricultural regions of Greece (Crete, Thessaly, Macedonia) with ongoing validation efforts aimed at refining its accuracy and adaptability. Future work includes the integration of cloud-based machine learning models and EO-derived evapotranspiration data to enhance the efficiency of extrapolating parcel-level (L1) and regional (L2) metrics into policy-level indicators (L3). Additionally, alternative aggregation methods, such as model-based approaches, spatial regression, and interpolation techniques like Kriging, will be tested to improve the accuracy and reliability of aggregated insights.