Comparative study on typhoon's wind speed prediction by a neural networks model and a hydrodynamical model

Area of Science:

• Meteorology and Atmospheric Sciences
• Computational Science
• Environmental Modeling

Background:

• Accurate prediction of natural phenomena like typhoons is crucial but challenging.
• Existing weather forecasting models often encounter errors in configuration and prediction accuracy.
• Uncertainty is inherent in all weather predictions, necessitating improved modeling techniques.

Purpose of the Study:

• To review and compare the performance of two distinct models for predicting typhoon wind speed.
• To evaluate the strengths and weaknesses of dynamical (WRF) and data-driven (ANFIS) approaches.
• To identify potential complementary aspects and suggest improvements for future typhoon prediction research.

Main Methods:

• Utilized the Weather Research and Forecasting (WRF) dynamical model.
• Employed an Adaptive Neuro-Fuzzy Inference System (ANFIS) for prediction.
• Assessed model performance using statistical parameters: Root Mean Square Error (RMSE) and Correlation Coefficient (CC).

Main Results:

• The Adaptive Neuro-Fuzzy Inference System (ANFIS) model demonstrated higher accuracy in typhoon intensity prediction compared to the WRF model.
• ANFIS achieved a lower RMSE and a higher CC, indicating superior performance.
• Both models have distinct advantages and disadvantages, with ANFIS showing greater predictive power for typhoon intensity.

Conclusions:

• The ANFIS model offers a more accurate approach for typhoon intensity prediction than the WRF model in the South China Sea.
• Further research should focus on refining model concepts and exploring how different methods can complement each other.
• Improving the conceptualization and integration of modeling techniques is essential for advancing typhoon forecasting capabilities.