GEOGLAM, an initiative supported by the G20, aims to enhance market transparency and provide early warnings in regions facing food insecurity through the power of EO.
During the GEO Symposium and the ODOK Workshop 2024, GEOGLAM facilitated the Open Science Day to foster collaboration and data sharing among organizations focused on agricultural monitoring.
At this event, VITO, a partner organization of GEOGLAM and project coordinator of ScaleAgData, highlighted the innovative work being done in ScaleAgData. The presentation focused on the importance of making in-situ sensor data accessible and how this information can bolster EO-based monitoring tools.
A key aspect of the discussion revolved around the methodological frameworks currently being developed to integrate sensor data into various agricultural workflows. These methodologies aim to tackle persistent challenges in agricultural monitoring.
Among the innovative strategies presented were self-supervised and few-shot learning frameworks designed to maintain model accuracy even when reference datasets are limited.
Additionally, continuous learning and data assimilation methodologies were introduced, allowing for the effective use of real-time reference data without the need for frequent model retraining.
The effectiveness of these frameworks was exemplified through a yield estimation model, which demonstrated significant accuracy improvements when utilizing the new methodologies.
This showcase underscored the potential of in-situ data to enhance EO monitoring, paving the way for more precise agricultural insights.
The connection with GEOGLAM was sought, discussing how ScaleAgData can contribute to their working groups, such as their in-situ working group as well as the Essential Agricultural Variables working group.
For more information on GEOGLAM and ScaleAgData's in initiatives, please visit GEOGLAM and scroll through this ScaleAgData website.
