'NVIDIA's "Earth 2" and "Codip" Land in Taiwan, a "Typhoon Regular Course" ? Will South Korea Get Them Too?'

Typhoon Path and Impact Prediction Platform
Typhoons Form in the Philippines, Pass Through Taiwan, and Land in Korea
Using Taiwan Data Enables Korea to Prepare

Taiwan has started using a platform combined with NVIDIA's software 'Earth-2' and generative AI model 'Codex' from this year to predict the path and scale of typhoons. It is attracting attention as it is expected to have a considerable impact on South Korea as well. If Taiwan can analyze typhoons accurately, South Korea, which could be on the typhoon's next path, will be able to prepare appropriately based on these results.

A scene where Nvidia's generative AI Codif and software Earth2 are analyzing the shape of a typhoon located near Taiwan. Photo by Nvidia

NVIDIA recently unveiled Earth-2 and Codex at Computex 2024 held in Taipei, Taiwan. Jensen Huang, NVIDIA's CEO, also gave a keynote speech on the 2nd at the National Taiwan University Sports Center, where he explained Earth-2 and Codex in detail.

Earth-2 and Codex predict and visualize the typhoon's movement, cloud size, and rainfall amount before the typhoon actually arrives, providing simulation data. Earth-2 collects data, and the AI Codex receives it to analyze the expected path and other factors. The data includes detailed time-based movements of the typhoon, allowing comparisons between day and night conditions. Notably, Codex analyzes data 1,000 times faster than traditional methods used by national meteorological agencies and offers about 10 times higher resolution for the analysis target. It also requires 3,000 times less energy for inference needed to predict the path, reducing power consumption. Taiwan's Central Weather Bureau (CWB) under the Ministry of Transportation recently adopted Earth-2 and Codex.

NVIDIA's weather forecasting software Earth-2 and generative AI Codif provided simulation data visualizing the projected path of a typhoon expected to make landfall near Taiwan. Photo by Hyungmin Kim

Typhoon-related items that Nvidia's weather forecasting software Earth-2 can collect or predict. Photo by Hyungmin Kim

Globally, early detection and preparation for typhoons are becoming increasingly important. With abnormal climate phenomena caused by global warming becoming more frequent, typhoons have become more powerful. This can cause significant damage not only to people's daily lives but also to industrial sites. As the summer season approaches, when typhoons tend to pass intensively over the Korean Peninsula, concerns about typhoons in the industrial sector are growing. The memory of the severe damage caused by Typhoon Hinnamnor in September 2022 still remains.

If Taiwan's typhoon prediction capabilities improve, it is expected to benefit South Korea as well. By monitoring Taiwan's weather prediction trends and comprehensively judging them together with data analyzed domestically, better preparation for typhoons can be achieved. Taiwan's climate prediction information can also be accessed through the Asia-Pacific Economic Cooperation (APEC) Climate Center. Since its establishment in 2005, the APEC Climate Center has been producing and sharing optimal climate prediction information through a climate network among 21 APEC member economies. It maintains especially close communication with Taiwan's meteorological agency, which is geographically significant.

On the 5th, officials examined NVIDIA's weather forecasting software Earth-2 at the NVIDIA AI Summit held at the Grand Hi-Lai Hotel in Taipei, Taiwan. Photo by Kim Hyung-min

NVIDIA's Earth-2 and Codex show potential to become heroes saving not only Taiwan but the entire Asia region affected by typhoons. An NVIDIA official who explained the system on-site said, "By comprehensively reviewing data provided by Earth-2, Codex, and other institutions, not only Taiwan but several countries will be able to respond quickly to various natural disasters such as typhoons, earthquakes, and tsunamis."

This content was produced with the assistance of AI translation services.