A new step forward in induced earthquake research in Vietnam

In Vietnam, induced earthquakes have been recorded at several reservoirs after water impoundment, such as Hoa Binh, Son La, and Song Tranh 2. At Lai Chau Hydropower Plant, the reservoir began impoundment in May 2015, and earthquakes started occurring near the reservoir shortly thereafter. Since then, earthquakes have continued. By 2021, around 1,500 earthquakes had been recorded in the area, 13 of which had a magnitude of 4.2 or more. These figures highlight the urgency of this research. The project “Study on the impact of induced earthquakes and assessment of earthquake hazard in Lai Chau, Vietnam” (code: QTIT01.02/23-24) was implemented to timely identify potential risks and propose appropriate response solutions. The project is the result of an international cooperation programme between the research team of the Institute of Earth Sciences and the Italian National Research Council (CNR).

Dr Cao Dinh Trong (right) discusses the results of analysis and calculations

According to Dr Cao Dinh Trong, principal of the project, the phenomenon of induced earthquakes in Lai Chau is no longer a hypothesis, it has been confirmed with thousands of seismic events, many of which were of considerable magnitude. Analysing the correlation between seismic activity and fluctuations in reservoir water levels will help operating units become more proactive in risk prevention.

As part of the research, scientists analysed the temporal dynamics of earthquake sequences within a 10-kilometre radius of the reservoir, at depths of up to 9 kilometres. The use of Empirical Mode Decomposition (EMD) allowed the researchers to separate signal sequences into distinct components, thereby identifying seismic oscillation cycles that match the cycles of water level changes. This is the first time that such a periodic relationship between the two variables has been confirmed under factual conditions in Vietnam.

Seismic distribution within a 10-kilometre radius from the Lai Chau dam centre

Based on the collected data, the team determined that the earthquake hazard level in Lai Chau ranges from 0.08 to 0.1g, corresponding to level VII on the MSK-64 scale—a level capable of causing significant impact on construction works. To assess the area's sensitivity to seismic activity, the team also successfully built and trained a multi-layer perceptron (MLP) neural network model with an 8-14-1 structure. Using spectral analysis, data sequence correlation testing, and modern statistical methods, the model revealed a connection between minor earthquakes in the region and the reservoir's water impoundment process. These findings provide further evidence and recommendations to improve infrastructure safety and reduce disaster risks.

Structure of the 8-14-1 MLP neural network model for data sensitivity assessment

The outcomes of the project have been transferred to relevant units, including Lai Chau Hydropower Plant, reservoir safety management units, and disaster prevention agencies, laying the foundation for early warnings, proactive operations, and more scientifically grounded policy planning.

Furthermore, the research team has published its analysis and evaluations in prestigious domestic and international scientific journals, clarifying the characteristics of induced earthquakes in Lai Chau. One notable study, published in Fractal and Fractional in 2023, used fractal and spectral analysis methods to examine the nonlinear nature and self-organising structure of earthquake sequences in the region. The study identified signs of energy accumulation prior to earthquakes. Another study, published in Entropy in 2024, applied the visibility graph method to analyse shallow earthquakes in Lai Chau. The results helped identify hidden patterns in the seismic event sequences and provided valuable scientific data for assessing earthquake hazards in the region.

The research result has clarified the correlation between reservoir water levels and seismic activity in the Lai Chau area while also introducing a new approach to geological risk assessment by integrating real-time data with artificial intelligence models. This represents an important step toward shifting from descriptive analysis to forecasting, supporting the development of seismic safety criteria and updating the national earthquake hazard map in line with international standards.

In the next phase, the research team will incorporate satellite data (InSAR) to monitor ground deformation, thereby improving the capacity to simulate and provide early warnings for earthquake sequences. Integrating multi-source data will help increase model accuracy and broaden its application scope.

Through international collaboration, the team has gained access to advanced methodologies and improved its scientific publication capacity. Building on these results, the scientists hope to expand the study to other major reservoirs nationwide, aiming to develop an early warning system and complete the national map of induced earthquake hazards. At the same time, they seek to strengthen international cooperation networks towards multi-national research programmes funded by the European Union and other international organisations.

Translated by Phuong Huyen
Link to Vietnamese version