Radar Echo Maps Prediction Using an Improved MIM with Self-Attention Memory Module

Table 1

Spatialtemporal models hyperparameter settings.

Model	Number of layer	Number of kernel	Kernel size	Optimizer
Causal LSTM [28]	4	16 × 16 × 16 × 16	5 × 5	Adam
ConvLSTM	4	16 × 16 × 16 × 16	5 × 5	Adam
ST-LSTM	4	16 × 16 × 16 × 16	5 × 5	Adam
MIM	4	16 × 16 × 16 × 16	5 × 5	Adam
SAMM-MIM	4	16 × 16 × 16 × 16	5 × 5	Adam