Geospatial multi-scale GNN for urban food security in climate-stressed environments

Area of Science:

• Agricultural Science
• Environmental Science
• Computer Science

Background:

• Global food insecurity is rising due to climate change and soil degradation.
• Resilient alternative agricultural systems are crucial for risk management.
• Hydroponic and aeroponic systems offer potential but face nutrient instability challenges.

Purpose of the Study:

• To develop a geospatially integrated monitoring framework for hydroponic and aeroponic systems.
• To mitigate risks associated with nutrient instability and environmental uncertainty.
• To enhance the reliability of urban food systems.

Main Methods:

• Implementation of a Raspberry Pi-based IoT network for real-time monitoring.
• Development of an Optimized Multi-Scale Adaptive Graph Neural Network (OMSA-GNN).
• Optimization of the GNN using an Improved Sparrow Search Algorithm (ISSA).
• Integration of visual plant indices for proximal remote sensing.

Main Results:

• The OMSA-GNN framework demonstrated superior performance in forecasting growth trajectories.
• Effective management of resource-related risks in controlled environment agriculture.
• Early detection of plant physiological stress using visual indices.
• Validation with a lettuce growth dataset.

Conclusions:

• The OMSA-GNN provides a scalable approach for improving urban food system reliability.
• This framework enhances risk management in alternative agricultural systems.
• It offers a robust solution for nutrient instability and environmental uncertainty.