, Available online , doi: 10.11932/karst2026y011
Abstract:
As a vital component of the Earth's Critical Zone, the karst critical zone exhibits highly sensitive hydrological, geochemical, and biological processes in response to environmental changes. Effective monitoring is therefore crucial for addressing key resource and environmental challenges in karst regions, including soil erosion, drought and flood disasters, and ecological conservation. However, existing monitoring stations, often constructed during different periods with varying observation indicators, suffer from incompatible equipment, transmission protocols, and data platforms. This significantly impedes data timeliness, unified management, and data sharing.To systematically resolve these issues, this paper designs and implements a standardized data convergence system based on Internet of Things (IoT) technology and cloud-native architecture. Firstly, at the device access layer, the system utilizes the MQTT protocol as its core while maintaining compatibility with standard protocols like ModBus. It also provides a flexible and extensible SDK, enabling efficient and cost-effective integration of both existing and new heterogeneous monitoring devices, thereby achieving unified access and management of multi-protocol equipment. Secondly, at the system architecture level, it employs advanced microservices design and containerization technology. Core functions are modularized through RESTful APIs, and services are containerized and automatically orchestrated using Docker and Kubernetes. This cloud-native architecture endows the system with high elasticity, scalability, and reliability, enabling self-healing, elastic scaling, and load balancing. This effectively supports the processing of massive concurrent data streams and accommodates future business growth. Regarding data storage, the system innovatively adopts a hybrid multi-type database architecture. It integrates Redis, Apache Doris, MongoDB, and MinIO to handle different scenario-scaching, real-time analysis, document storage, and object storage, respectively-ensuring highly efficient and stable data read/write operations. Furthermore, the system implements three data snapshot collection rules-immediate, change-based, and scheduled-and supports the import of historical data, guaranteeing the continuity and integrity of monitoring data. Security is embedded throughout the device end, transmission layer, and platform layer, with comprehensive data protection ensured through mechanisms such as pre-authorization, TLS encryption, and access control.The system concretely implements a three-tier application structure: a management console, a large monitoring screen, and a mobile APP. The management console handles device management, intelligent operational maintenance, and system configuration; the large screen provides a global data overview based on map visualization; the mobile terminal enables users to access device status and data anytime, anywhere. Application examples demonstrate that the system has been successfully deployed, integrating numerous monitoring stations from several national and provincial field observation and research stations, including those in Pingguo, Guangxi, and Guilin. It has accumulated over 2 million data entries, achieved a station data completeness rate of 99.94%, and operates stably. Performance tests indicate that the system maintains millisecond-level response times under simulated scenarios of millions of concurrent device connections and queries on massive historical datasets, demonstrating excellent processing capability and stability.In conclusion, this system effectively addresses the long-standing challenges of equipment heterogeneity and data silos in karst monitoring. It establishes a unified, efficient, and reliable data convergence and management platform, providing a solid data foundation for scientific research, ecological protection, and disaster early warning in karst areas. Future work will focus on the deep integration of artificial intelligence algorithms and professional models to further enhance the system's intelligence in data quality control, intelligent early warning, and analytical evaluation.