A spatiotemporal attention-augmented ConvLSTM model for ocean remote sensing reflectance prediction

Gaoxiang Zhou, Jun Chen, Ming Liu, Lingfei Ma

Abstract: Highlights•An augmented ConvLSTM model was proposed for ocean Rrs 7-day-ahead prediction.•An attention-augmented convolution was used to extract spatiotemporal features.•The proposed model achieves a satisfactory performance at 443, 488 and 555 nm bands.•The proposed method outperformed the CNN, LSTM, CNN-LSTM, and ConvLSTM models.