The European Center for Medium-Term Weather Forecasting (ECMWF) has launched Pangu-Weather, an artificial intelligence (AI)-based model from Huawei Technologies (China) on its website in late July.
“The decision to incorporate Pangu-Weather into their weather forecasting was partly due to the model’s high performance. The center recognized the significant potential of AI in this area,” Tian Qi, Huawei’s chief science officer, told the newspaper. south china morning post,
Pangu-Weather test results published in scientific journal Nature Last month, it was found that the model’s predictions are more accurate than traditional methods. It can forecast weather from 1 hour to 7 days ahead at a speed of 10,000 times faster.
Pangu-Weather is designed to provide global weather forecasts on a second-by-second basis, covering aspects such as humidity, wind speed, temperature, sea water pressure and much more.
The model also shows excellent performance in predicting the path of storms, cold waves, heat waves etc. Recently, Pangu-Weather predicted a rapid turn of Typhoon Khanun on 4 August.

A flooded area after the effects of Typhoon Khanon in Daegu city, South Korea, on August 10. Photo: Reuters
Tian revealed that Huawei’s ultimate goal is to use Pangu-Weather to build a next-generation AI weather forecasting framework. The move is aimed at improving the speed and accuracy of weather forecasting to better respond to extreme weather.
Meanwhile, Florian Pappenberger, head of forecasting at the ECMWF, describes AI as a “silent revolution” in weather forecasting, showing great potential and impressive results.
Weather patterns are chaotic and cannot be accurately calculated by even the most powerful supercomputers. However, AI models have some capabilities that humans cannot. “AI can help find new patterns of atmospheric evolution from huge data,” explained Mr. Tian.
However, Xie Lingxi, a senior researcher at Huawei, believes that Pangu-Weather is not correct. For example, the model proved more accurate at predicting general storm trends but lagged behind at predicting specific landslide locations.
In addition, current AI methods may underestimate hurricane strength, partly due to the low proportion of past extreme weather data used to train AI models…
In the US, California firefighters are using AI to help them spot wildfires early through a network of more than 1,000 cameras installed across the state. The system, called ALERTCalifornia, was developed by the University of California, San Diego (UCSD) and went live last month.
theo reuters, ALERTC California recently helped locate a 3 a.m. fire in a remote, wooded area of the Cleveland National Forest. Had the AI not alerted the concerned force immediately, the fire could have spread. The result was that the fire was extinguished in 45 minutes.
In addition to the camera network, this AI platform is also gathering more information, such as aerial survey data, predicting flammable vegetation, mapping the ground beneath trees, etc., which can help measure atmospheric rivers and snow during winter. Yes, its effect can be estimated. wildfires on erosion, sediment dispersal, water and soil quality, and more.
Worst wildfire disaster in America
The death toll in the horrific forest fire that broke out on August 13 on the Maui island of Hawaii, USA, has increased to 93. The AP news agency quoted the local government as saying that the death toll in search and rescue operations and identification of people is still ongoing. In its early stages”, and warned that the number of victims could still rise. Maui County Sheriff John Pelletier said they had only reached an area that represented 3% of the expected search area.
It is considered the deadliest wildfire in the US in more than a century. Hawaii Governor Josh Green admitted: “Now all we can do is wait and support those who are still alive.” Figures from Maui County officials show that approximately 4,500 people are in need of shelter. In addition, at least 2,200 buildings were damaged or destroyed on the western side of Maui. According to Mr. Green, the physical damage is estimated to be around $6 billion.
Specifically, the AP reported that based on Hawaii’s emergency management records, the siren did not sound before the fire broke out in the town of Lahaina on the island. Local authorities sent alerts via television and radio to mobile phones, but they were limited by communication blockades.
Hawaii officials have launched an investigation into the response to the fire amid complaints from residents that they did not receive warnings. Furthermore, the effort to put out the fire faces a major challenge as there are not enough firefighters and specialized equipment. Bobby Lee, president of the Hawaii Firefighters Association, said there were only a maximum of 65 firefighters on duty at any given time, even though the extent of the fire was so extensive.
Pham Nghia