基于3D卷积和自注意力机制的卫星云图预测研究
方巍, 李佳欣, 陆文赫

Research on satellite cloud image prediction based on 3D convolution and self⁃attention
Wei Fang, Jiaxin Li, Wenhe Lu
表2 3D SA?STLSTM与其他模型在卫星云图数据集上的测试结果
Table 2 Testing results of 3D SA?STLSTM and other models on satellite cloud image datasets
MethodMAEMSEPSNRSSIM
ConvLSTM[18]0.05280.01387.5610.194
PredRNN[20]0.01090.012714.7300.259
PredRNN++[34]0.01350.003720.7570.359
Crevnet[24]0.01440.000827.8910.461
3D SA⁃STLSTM0.00860.000728.7950.495