Typhoon Track Forecast Service

                                                        

Global Typhoons Track. ©Wikipedia

  Every summer, South Korea experiences significant damage from typhoons. With global warming causing an increase in typhoon intensity, accurate prediction of the typhoon's path and strength at the early stages is crucial. Providing sufficient lead time for response measures can help prevent potential damages.

  ECMWF's prediction of Typhoon Kong-rey was a major societal issue indeed. There were significant discrepancies between the results, and this caused unprecedented damages in both Korea and Japan. This caught our attention and led us to choose our subject for the hackathon: "Predict Typhoon Kong-rey and potentially other typhoons using FCN.

  Our project targets this issue by constructing a typhoon track prediction using FourCastNet(FCN), a weather forecasting model based on the Fourier Neural Operator (FNO). Our team focused on studying the integration of FCN, which was trained on European Centre for Medium-Range Weather Forecasts Reanalysis v5 data(ERA5 data), with the JMA Best Track data. This effort aims to harness the robust predictive power of FCN and the highly accurate historical records of the JMA to enhance the accuracy and efficiency of our weather forecasting model.

  If you are interested in typhoon prediction using deep learning models, visit our website. It provides not only fundamental information about typhoon path prediction but also detailed insights into how our model operates. This information will enhance the understanding of typhoons and raise awareness of how we can protect ourselves from their impacts.

 

You can find more information about our presentation explanation here


Data analysis: How the FourCastNet model works and how it predicts typhoons movement. We used GT data as input for each time step in the auto-regressive predictions. if you need detail information, please see Details.

Future prediction : Provides information on the current location of the typhoon, the expected damage to each region of South Korea. You can change the date, region, and time, and the displayed warning message will vary according to the date and region. The prediction image of the typhoon will also change its location and radius based on the date and time you set.

Details : Describes the traditional weather forecasting methods and their limitations, and explains why AI technology was introduced. In addition, it provides a detailed process of how the FCN model works and the data and variables used. You can see the improved forecast performance on this page.

About us : For contact us, Provides information and contribution about the members in this project. 

 

NIMS, ABLEARN, NVIDIA

Team Typhoon Sensei