Application of the deep learning for the prediction of rainfall in Southern Taiwan

Meng-Hua Yen,Chu-En Lin,Ding-Wei Liu,Yi-Chia Hsin,Chii-Chang Chen

doi:10.1038/s41598-019-49242-6

Abstract

Precipitation is useful information for assessing vital water resources, agriculture, ecosystems and hydrology. Data-driven model predictions using deep learning algorithms are promising for these purposes. Echo state network (ESN) and Deep Echo state network (DeepESN), referred to as Reservoir Computing (RC), are effective and speedy algorithms to process a large amount of data. In this study, we used the ESN and the DeepESN algorithms to analyze the meteorological hourly data from 2002 to 2014 at the Tainan Observatory in the southern Taiwan. The results show that the correlation coefficient by using the DeepESN was better than that by using the ESN and commercial neuronal network algorithms (Back-propagation network (BPN) and support vector regression (SVR), MATLAB, The MathWorks co.), and the accuracy of predicted rainfall by using the DeepESN can be significantly improved compared with those by using ESN, the BPN and the SVR. In sum, the DeepESN is a trustworthy and good method to predict rainfall; it could be applied to global climate forecasts which need high-volume data processing.

Highlights

Taiwan is located in the subtropical monsoon climate zone, year-round rainy, and surrounded by seas, and has highly changing terrain elevation
By comparing the results based on deep echo state network (DeepESN) model with the echo state network (ESN) model, we found that the evaluations items are able to be improved by the DeepESN with a slight increase of computing time
We demonstrate that rainfall prediction can be achieved by means of the neural networks

Summary

Results and Discussion

After training with the data since the beginning of 2002, we can input the test data (i.e, the remaining data) into the network in the established model to forecast rainfall and RMSE, NRMSE, and γ are further calculated from the observed precipitation and the predicted value. Excluding optimized training length of 20,000 hours from the data of the Obszen, the remaining data length of 82,500 hours will be adopted to predict rainfall in the southern Taiwan and used to examine the performances of the ESN/DeepESN model. The meteorological data from the Obsyuj were further adopted as the input for the ESN/DeepESN model testing to examine whether the ESN/DeepESN model trained by the Obszen can be directly applied to make rainfall prediction at other observatories. By comparing the statistical quantities between the two stations

ObsZen Obsyuj ObsZen Obsyuj Trainlm trainbfg

Conclusion

Data and Methods

Additional Information