Vietnam uses supercomputers and AI to forecast storms and heavy rains

The Department of Hydrometeorology has introduced the CrayXC40 supercomputer, applying AI to improve the accuracy of forecasts and warnings of storms, heavy rains and other types of natural disasters.

Forecasts of AI models are applied by the Department of Hydrometeorology in forecasting work - Photo: NCHMF

On October 3, the Department of Hydrometeorology announced that on the occasion of the 80th anniversary of the Hydrometeorology Industry's Traditional Day (October 3, 1945 - October 3, 2025), the Department of Hydrometeorology was honored to receive the First Class Labor Medal from the President .

According to the plan, the ceremony to celebrate and receive the medal was held this afternoon. However, due to the complicated situation of rain and flood after storm No. 10 and storm Matmo (storm No. 11) which may directly affect our country, the Department of Hydrometeorology decided to postpone the ceremony to focus on storm forecasting to serve natural disaster prevention.

According to the Department of Hydrometeorology, after 80 years, the hydrometeorology industry has made great strides in forecasting storms, heavy rains, and warnings of natural disasters.

Currently, the national network of hydrometeorological stations has 2,807 stations (an increase of 44% compared to the period 2016-2023), the automation rate of meteorological monitoring is 65%, rain is 100%, thunderstorms are 100%, hydrology is 69% and oceanography is 74%, the modern weather radar network is distributed to basically cover the whole country.

Regarding the development of new technology and equipment, the Department of Hydrometeorology said it has put the CrayXC40 supercomputer into operation, a breakthrough in the industry.

With a capacity of nearly 80 TFLOPS, the CrayXC40 supercomputer allows forecasting with a resolution of 3km for the entire territory and the East Sea in just 30-40 minutes, putting Vietnam in the group of countries with strong computing capacity in the region.

For example, during the heavy rains in September 2024, this system helped update the 3km quantitative rainfall forecast map, supporting flash flood and landslide warnings to the commune level.

The department has also strongly upgraded the digital weather forecasting (NWP) model technology. High resolution models (1-3km) are synchronized with all domestic observation data, combined with forecast sources from ECMWF to reduce product delay from 5-8 hours to 2-3 hours.

The ensemble forecasting system with 32 short-term components and 51 medium-term components helps create probability maps, impact forecasts and quantitative rainfall for more than 3,000 communes and wards.

The Department of Hydrometeorology is applying many modern technologies and artificial intelligence to forecast and warn of natural disasters - Photo: NCHMF

According to the Department of Hydrometeorology, artificial intelligence (AI) has been applied by the industry in storm identification, forecasting extremely short rains, analyzing Himawari satellite images, and improving forecasting of tropical cyclone intensity.

For example, a deep learning model integrating satellite and radar data helped identify the center of typhoon Noru (2022) early when entering the East Sea, supporting 72-hour early warning.

This is an important preparatory step for Vietnam to gradually master multi-scale forecasting technology using AI, towards developing a model with a Vietnamese brand.

The Department also applies AI technology, complex models, radar and lightning positioning systems, allowing for extremely short-term warnings (0 - 6 hours) with increasing accuracy. The iWeather.gov.vn portal is operated to directly serve the people.

Along with that, the application of a real-time flash flood and landslide warning system, using soil moisture simulation technology combined with rain data analysis to provide online warnings to the commune level. This is a clear breakthrough, demonstrating a high determination in bringing science to serve life, especially in remote areas.

Thanks to technological innovation and improved forecaster skills, forecast quality has improved significantly, such as the storm forecast period being increased from 24 hours to 3 days, warnings being issued 5 days in advance, and the error in storm location at 48 hours has been halved.

Forecasts of heavy rain and flood warnings 2-3 days in advance have a reliability of about 75%, warnings of local thunderstorms 30 minutes to several hours in advance, and forecasts of severe cold and widespread heat have a reliability of 70-90%.

Back to topic

WISDOM

Source: https://tuoitre.vn/viet-nam-dua-sieu-may-tinh-ung-dung-ai-vao-du-bao-bao-mua-lon-20251003115001914.htm