Precipitation forecasting based on deep learning strategy using empirical wavelet transform, Markov chain-incorporated long-short term memory network
Accurate and stable precipitation forecasting can better reflect the changing trend of climate and also provide timely and efficient environmental information for management decision, as well as prevent the occurrence of floods or droughts. This paper proposes a hybrid model for precipitation forecasting and demonstrates its efficiency. In the study, the empirical wavelet transform (EWT) was firstly introduced to decompose and pre-analysis hidden characteristics of the precipitation data. Secondly, the long-short term memory (LSTM) network are improved after incorporation of Markov chain (MC) based algorithm with which the percentage of rainless and rainy months is forecasted perfectly, thus generation of any extreme and non-physical precipitation is eliminated. Thirdly, the multi-step prediction was explored to improve the reliability and flexibility of rainfall. To verify the performance of the proposed model, monthly precipitation data were used as illustrative cases. Parallel experiments using non-decomposing models, other traditional machine learning approaches optimized by the mind evolution algorithm have been designed and conducted to compare with the proposed model. Results obtained from this study indicate that the proposed hybrid model can capture the nonlinear characteristics of the precipitation time series and thus provides more precise forecasting results.