Geohazards such as earthquakes, wildfires, and floods threaten infrastructure, economies, and lives, necessitating real-time monitoring for timely intervention. Sage Continuum, a cyberinfrastructure project, deploys ARM64-based NVIDIA Xavier NX edge devices to collect, process, and transmit sensor data efficiently. However, resource constraints on these devices pose challenges for performance monitoring.

This study introduces a lightweight monitoring plugin leveraging Linux virtual filesystems (procfs and sysfs) to capture CPU usage, memory utilization, and energy consumption. Through systematic experiments, we analyze the tradeoff between performance and accuracy, identifying optimal sampling intervals that minimize monitoring overhead while preserving data fidelity. Results highlight practical strategies for resource-efficient monitoring, enabling improved application scheduling and workload management on edge devices. These findings offer actionable insights for optimizing geohazard monitoring systems and advancing edge computing efficiency.