New AI model brings breakthroughs in five-day regional weather forecasting
Institute of Atmospheric Physics, Chinese Academy of Sciences
image:
Diagram illustrating the process of cascaded prediction
view moreCredit: Congqi Cao
For decades, medium-range weather forecasting—predicting conditions 1 to 5 days ahead—has relied heavily on traditional numerical models. However, this approach often struggles when applied to specific regions with limited historical data.
Researchers at Northwestern Polytechnical University in China have now proposed a novel deep learning–based framework that dramatically improves the accuracy of forecasts, even when data are limited.
To address key challenges in regional forecasting, the team introduced a new method that integrates three major innovations: the use of semantic segmentation models originally designed for medical image analysis; a learnable Gaussian noise mechanism that improves the model’s robustness; and a cascade prediction strategy that breaks the forecasting task into manageable stages. The study is published in Atmospheric and Oceanic Science Letters recently.
“Our goal was to make regional forecasting smarter, faster, and more reliable, even in data-limited scenarios.” says Associate Professor Congqi Cao, corresponding author of the study. “This is especially valuable for areas where a dense network of meteorological observations is not available.”
The method was tested on the East China Regional AI Medium Range Weather Forecasting Competition dataset, which includes 10 years of reanalysis data from ERA5. The task involved using past atmospheric variables to predict five key surface weather indicators—including temperature, wind, and precipitation—every 6 hours for the next 5 days.
The results speak for themselves: the model achieved significant improvements in prediction performance, outperforming many mainstream global AI forecasting models. Specifically, the method reduced temperature forecast errors by 9.3%, improved the precipitation F1-score by 6.8%, and lowered wind speed errors by 12.5%.
“This is the first time semantic segmentation and learnable noise mechanisms have been used together for regional weather forecasting,” explains Prof. Cao. “It opens up new possibilities for accurate forecasting in other data-scarce regions.”
Looking ahead, the team plans to extend their method to real-time systems and apply it to more regions across China. They hope their work will eventually serve public safety, agriculture, and disaster prevention needs—delivering smarter, faster, and more local forecasts when they matter most.
Journal
Atmospheric and Oceanic Science Letters
DOI
Characteristics of the extratropical transition of tropical cyclones on different tracks in the western North Pacific
image:
Characteristics of four ETC clusters during transition and the environmental field configurations after transition
view moreCredit: Shanshan Li
As one of the most destructive types of natural disaster, tropical cyclones have long been a key focus in disaster prevention and mitigation efforts. However, when tropical cyclones migrate into the mid-to-high latitudes, they gradually lose their tropical cyclone characteristics, and transform into extratropical cyclones. People’s vigilance tends to decrease, believing that such storms pose less of a threat. In fact, this transformation expands the impact range of the storm, while combining the characteristics of tropical and extratropical cyclones. It leads to heavy rainfall, huge waves and strong winds, causing more serious damage than a single type of cyclone.
Currently, our level of understanding of the activity characteristics and transformation mechanism of these “extratropical transition tropical cyclones” (ETCs) in the western North Pacific remains incomplete, which poses significant challenges to disaster warning and numerical forecasting.
Recently, Associate Professor Lei Chen, from the Department of Atmospheric Science, School of Environmental Studies, China University of Geosciences (Wuhan), with her research team, conducted a classification of 386 ETCs that occurred in the western North Pacific from 1979 to 2022, into four clusters according to their track patterns: recurving ETCs, westward ETCs, northwestward ETCs, and abnormal track ETCs. In their study, published in Atmospheric and Oceanic Science Letters under the title “Climatological characteristics of the extratropical transition of tropical cyclones along different tracks in the western North Pacific (1979–2022)”, they reveal the transition characteristics and mechanisms of these four clusters of ETCs.
Results showed that the four clusters of ETCs possessed significant differences in transition duration, location, and mechanism. Recurving ETCs predominantly underwent transition during their track recurvature from northwest to northeast. Of the westward ETCs, 63.7% completed the transformation process rapidly after landing. Northwestward ETCs typically underwent transition in the baroclinic zone between 15°N and 20°N, characterized by the longest duration and slowest transition speed. Abnormal track ETCs mainly completed their transition over high-latitude oceans.
The transition process is closely related to the position and extent of the western Pacific subtropical high. “When the subtropical high extends westward, the frequency of ETCs increases, and the frequency decreases when it retreats eastward. Except for the westward ETCs, which are dominated by surface friction effects, the other clusters of ETCs mainly complete the transition in the northwest baroclinic zone of the subtropical high, and are affected by ocean thermal forcing or cold-air invasion,” concludes the corresponding author, Lei Chen.
Journal
Atmospheric and Oceanic Science Letters
No comments:
Post a Comment