Contributors: Changjun LIU, Guomin LV, Kun YANG, Haitao WANG, Bin TIAN, Yong LIU, Siyuan CHANG, Ronghua LIU, Qi LIU, Tao SUN

To advance the construction of digital twin river basins, the Ministry of Water Resources has explicitly established the strategic goal of building an "integrated space-air-ground monitoring system," creating an urgent demand for efficient, precise, and real-time monitoring of hydrological elements (such as discharge and water level) in rivers, lakes, and reservoirs. Traditional contact-based streamflow measurement methods exhibit significant limitations in adaptability, safety, and coverage, failing to meet the requirements for full-domain and all-time monitoring. Although visual streamflow measurement offers advantages such as non-contact operation and wide-area coverage, it has long-standing core bottlenecks: insufficient sensing accuracy in complex environments, limited edge real-time computing capabilities, and a low level of equipment intelligence. Breaking through these technical barriers by developing integrated intelligent visual image processing chips, highly adaptable streamflow measurement equipment, and intelligent analysis platforms is key to implementing the integrated space-air-ground hydrological monitoring system. Through nearly a decade of dedicated research and development by the project team, significant breakthroughs have been achieved in algorithm-chip integration and AI-fused edge computing. These advancements have laid a solid foundation for establishing a new generation of intelligent visual streamflow measurement equipment systems, holding significant importance for elevating the modernization level of hydrological monitoring and supporting the construction of digital twin river basins.

·   Develop a series of visual streamflow measurement algorithms and models based on artificial intelligence and image recognition.

·   Construct a large-scale, high-quality measured sample database covering complex operating conditions such as rainfall and nighttime.

·   Design and develop an intelligent visual image chip that is deeply integrated with visual streamflow algorithms.

·   Manufacture a visual streamflow edge computing terminal to achieve integrated online monitoring of water level, flow velocity, and discharge.

·   Develop a mobile application for visual streamflow measurement, supporting one-click, uncalibrated measurement at any cross-section with a single measurement time period of less than 5 minutes.

·   Develop an intelligent visual streamflow management platform to enable operational applications.

·   Innovatively developed a set of high-precision visual streamflow measurement algorithms: By achieving a breakthrough fusion of AI-based space-time image velocimetry (STIV) with deep learning, the research team constructed high-precision recognition models for flow velocity, water level, and discharge. This significantly enhances sensing accuracy and robustness under complex environmental conditions (rain/fog, nighttime, low flow, etc.).

·   Established a leading domestic visual streamflow measurement sample database for complex operating conditions: The research team systematically collected millions of paired field image-hydrological data sets covering different weather conditions, time periods, and flow regimes. This provided a solid data foundation for model training and validation, leading to improved generalization capabilities.

·   Developed the first dedicated intelligent image processing chip for visual streamflow measurement: Chip-level deep integration and hardware acceleration of core STIV algorithms for velocity and water level measurement were achieved, substantially improving the real-time image processing efficiency and energy efficiency ratio at the edge.

·   Successfully developed an intelligent edge computing terminal for streamflow measurement: With self-developed image chips and AI models integrated, the terminal embedded real-time online calculation and output of water level, velocity, and discharge under complex environments, reducing reliance on cloud computing and enhancing monitoring timeliness.

·   Launched a portable mobile application for visual streamflow measurement: "Uncalibrated, one-click" rapid measurement was enabled for any river cross-section, with a single measurement time period of less than 5 minutes, greatly improving the convenience and efficiency of field surveys and emergency monitoring.

·   Constructed a cloud-edge collaborative management platform for intelligent visual streamflow measurement: Multi-terminal data aggregation, visualization, intelligent analysis/warning, and remote management were realized. This supports the operational flow and application of hydrological data, providing a core sensing tool for smart water conservancy.

The intelligent visual image streamflow measurement technology has been promoted and applied in small and medium-sized river basins across more than 13 provincial-level administrative regions, including Beijing and Shandong. It has realized automatic, rapid, and precise identification of river water levels, flow velocities, and discharges, providing critical data and decision support for regulatory authorities in flood control and early warning. To date, over 400 visual streamflow measurement stations have been deployed, accumulating more than 5 million data records. These systems have supported flood emergency responses during flood seasons, demonstrating excellent application results. The generated economic benefits exceed RMB 50 million. By providing high-precision, low-cost, and wide-coverage intelligent sensing support for the "Four Pre" (Forecast, Warning, Preview, Plan) flood control operations, the technology has effectively supported the response and post-event evaluation of multiple major emergencies, yielding significant benefits.

Figure 1. Hydrological Monitoring Technologies and Equipment for Intelligent Visual Streamflow Measurement

Figure 2. Integrated Management Module for Intelligent Visual Image Streamflow Measurement

Figure 3. Key Technologies for Intelligent Visual Streamflow Measurement

Figure 4. Application of Visual Streamflow Measurement Technology in River Channels