Application of a hybrid algorithm of LSTM and Transformer based on random search optimization for improving rainfall-runoff simulation
2024

Improving Flood Forecasting with a New Hybrid Model

Sample size: 98 publication Evidence: high

Author Information

Author(s): Li Wenzhong, Liu Chengshuai, Hu Caihong, Niu Chaojie, Li Runxi, Li Ming, Xu Yingying, Tian Lu

Primary Institution: School of Water Conservancy and Transportation, Zhengzhou University, Zhengzhou, 450001 China

Hypothesis

Can a hybrid model combining LSTM and Transformer improve rainfall-runoff simulation for flood forecasting?

Conclusion

The RS-LSTM-Transformer model significantly enhances the accuracy and stability of flood forecasting compared to traditional models.

Supporting Evidence

  • The RS-LSTM-Transformer model achieved an NSE of 0.970 at a 1-hour lead time during calibration.
  • The model maintained RMSE and MAE below 36 m3/s and 27 m3/s, respectively, during validation.
  • The study validated the model using 98 real flood events, demonstrating its effectiveness.

Takeaway

This study created a new model to predict floods better by combining two smart computer techniques, making it easier to forecast when floods might happen.

Methodology

The study used a hybrid model that integrates Random Search optimization with LSTM and Transformer architectures to simulate flood processes.

Limitations

The model's performance depends on high-quality data and faces challenges related to computational resource demands.

Digital Object Identifier (DOI)

10.1038/s41598-024-62127-7

Want to read the original?

Access the complete publication on the publisher's website

View Original Publication