Hydraulic experts discuss ‘digital brain’ to protect 6,800 reservoirs

According to Dr. Nguyen Van Manh, Head of the Department of Science and Technology at the Institute for Water Resources Planning, an integrated data platform will be the backbone that enables more proactive, safer management as weather patterns grow increasingly extreme.

Fragmented data and the pressure of extreme rainfall

Speaking at the Forum on Digital Transformation for Reservoir Safety on November 21, Dr. Manh outlined how Viet Nam’s reservoir-operation system is being pushed to its limit. The country has around 6,800 reservoirs, but only 300 provide 80% of total regulated storage, and just 200 reservoirs are equipped with floodgates that allow active flood control. “Most reservoirs still follow rigid rule curves and lack the flexibility needed to respond to extreme conditions,” he said.

The North Central region, which holds the largest reservoir capacity nationwide, has faced nonstop storm-flood events in 2024 - 2025. Unusual storm tracks, off-season heavy rainfall and rapidly rising flood peaks have moved outside traditional patterns, causing rising economic losses. “Modernizing reservoir operations is no longer just a technical matter, it is a political and social necessity,” Dr. Manh emphasized.

Large reservoirs must simultaneously ensure flood control, downstream safety, hydropower, irrigation and ecological flows. Achieving these layered objectives requires three elements: complete data, strong modelling tools and trained personnel. “If even one pillar is weak, risk becomes inevitable,” he said.

The Institute currently maintains basic information for all 6,800 reservoirs, including real-time water-level data for about 500 of them and inputs from roughly 300 monitoring stations. Yet the landscape remains fragmented. Many datasets are inconsistent, managed under different formats or updated at different intervals.

The irrigation and drainage network comprising 45,000 hydraulic structures, relies heavily on weekly or monthly reports. Such delays make it difficult to support hour-by-hour decisions during rapidly developing flood situations.

One of the clearest improvements is the Vrain network of 2,600 automatic rainfall stations, more than 1,100 located in northern provinces. These deliver real-time rainfall data and are integrated into national and provincial disaster-monitoring systems. However, Dr. Manh cautioned that raw data often contains noise, missing values or inconsistencies. “Abundant data does not automatically guarantee accuracy,” he said, noting the need for filtering, cross-validation, and quality-control before data is input into models.

The Institute is also building a more complete water-resource database, incorporating inflow records, upstream-downstream characteristics, and storage curves. But Dr. Manh argued that Viet Nam needs a true “digital brain,” in which datasets from ministries, provinces, operators, Vrain and the national meteorological service follow a single set of standards and reside in one integrated data architecture.

Technology adoption improves but not fast enough

For over a decade, the Institute has supported flood-season operations for major reservoirs such as Cua Dat, Ngan Truoi, Ta Trach and Ia Mor, and provided joint-operation guidance for large river basins including the Red–Thai Binh, Ma, Ba and Srepok. It also develops seasonal water-availability forecasts and water-use plans across regions.

The workflow involves: analyzing operational rules; collecting rainfall and water-level data; forecasting inflows using MIKE NAM and HEC-HMS; modelling downstream hydraulics using MIKE 11, MIKE FLOOD and satellite-based GIS; and issuing daily or emergency operational bulletins. During the flood season, experts evaluate hundreds of inflow/outflow scenarios for each basin.

The forecasting system integrates three data groups: observations from more than 3,000 rainfall stations, short- and medium-range forecasts from the Viet Nam Meteorological and Hydrological Administration, global models including GFS, CFS (US) and ICON (Germany).

“Multiple forecast sources help, but uncertainty in rainfall remains the biggest challenge because it drives inflow to reservoirs,” Dr. Manh said.

Artificial intelligence is being used in early-stage tasks, such as data compilation and automated reporting. Yet the modelling system still relies heavily on older tools. “Many hydraulic models are outdated and not suited for ultra-short-term or multi-scenario forecasting,” he said. As a result, operators often remain reactive rather than proactive.

Transboundary basins like the Red River - Thai Binh River add further complexity because flows are affected by upstream reservoirs in China. Downstream flood-risk maps must be continuously updated and tightly linked to rainfall, terrain and reservoir-operation data to support real-time decision-making.

To enhance prediction capabilities, Dr. Manh recommended that the Ministry revise guidelines for downstream flood-risk mapping. New technologies, such as digital twins, location-based services (LBS) and high-resolution elevation data (DEM/LiDAR), should be incorporated. “Each reservoir should have a live model that is constantly updated and ready to provide timely operating recommendations,” he said.

A central theme of Dr. Manh’s presentation was the need for a sector-wide data platform in accordance with TCVN 13238:2020 / ISO/IEC 20546:2019. The system should adopt a river-basin data architecture, with reservoirs serving as core data objects. Observations, structural information, storage volumes, inflows, releases and downstream conditions must be seamlessly shared across ministries, provinces and operators.

Data-quality systems, including noise filtering, gap-filling, and version tracking, are crucial. And the platform must remain operational year-round. “Even without floods, the database must continue running. The data must stay clean, models updated and personnel ready,” Dr. Manh said.

Dr. Manh emphasized that digital transformation cannot rely solely on equipment or software. Sustaining a team of modelling experts is essential, supported by a long-term training and coordination mechanism with local authorities. He also called for the Ministry to issue uniform technical standards and pricing norms for data development and forecasting services, ensuring consistency across projects.

He concluded that digital transformation in reservoir operations “is not about adding a few new models or sensors. It is about building an interconnected ecosystem where data, tools and people interact seamlessly.” If successful, Viet Nam can shift from reacting to floods to anticipating and preparing for them, strengthening safety for millions of people living downstream.