Precipitation Spatio-Temporal Forecasting in China via DC-CNN-BiLSTM DOI Open Access

Peng Shu,

Xiaoqi Duan, Changliang Shao

и другие.

Water, Год журнала: 2025, Номер 17(9), С. 1381 - 1381

Опубликована: Май 4, 2025

Accurate and reliable precipitation prediction remains a significant challenge due to an incomplete understanding of regional meteorological dynamics limitations in forecasting routine weather events. To overcome these challenges, we propose novel model, DC-CNN-BiLSTM, which integrates dilation causal convolutional neural network (DC-CNN) with Bidirectional Long Short-Term Memory (BiLSTM) network. The DC-CNN component, by fusing dilated convolutions, extracts multi-scale spatial features from time series data. In parallel, the BiLSTM module leverages bidirectional memory cells capture long-term temporal dependencies. This integrated approach effectively links localized inputs broader hydrological responses. Experimental evaluation demonstrates that DC-CNN-BiLSTM model significantly outperforms traditional models. Specifically, improves Root Mean Square Error (RMSE) 9.05% compared ConvLSTM 32.3% ConvGRU, particularly medium- precipitation. conclusion, our results validate benefits incorporating advanced spatio-temporal feature extraction techniques for forecasting, ultimately improving disaster preparedness resource management.

Язык: Английский

Precipitation Spatio-Temporal Forecasting in China via DC-CNN-BiLSTM DOI Open Access

Peng Shu,

Xiaoqi Duan, Changliang Shao

и другие.

Water, Год журнала: 2025, Номер 17(9), С. 1381 - 1381

Опубликована: Май 4, 2025

Accurate and reliable precipitation prediction remains a significant challenge due to an incomplete understanding of regional meteorological dynamics limitations in forecasting routine weather events. To overcome these challenges, we propose novel model, DC-CNN-BiLSTM, which integrates dilation causal convolutional neural network (DC-CNN) with Bidirectional Long Short-Term Memory (BiLSTM) network. The DC-CNN component, by fusing dilated convolutions, extracts multi-scale spatial features from time series data. In parallel, the BiLSTM module leverages bidirectional memory cells capture long-term temporal dependencies. This integrated approach effectively links localized inputs broader hydrological responses. Experimental evaluation demonstrates that DC-CNN-BiLSTM model significantly outperforms traditional models. Specifically, improves Root Mean Square Error (RMSE) 9.05% compared ConvLSTM 32.3% ConvGRU, particularly medium- precipitation. conclusion, our results validate benefits incorporating advanced spatio-temporal feature extraction techniques for forecasting, ultimately improving disaster preparedness resource management.

Язык: Английский

Процитировано

0