Fusing Landsat 8 and Sentinel-2 data for 10-m dense time-series imagery using a degradation-term constrained deep network

•A novel deep network is proposed for Landsat 8 and Sentinel-2 data fusion.•Residual dense blocks and attention modules are combined to extract features.•A degradation constraint term is used to ease the temporal variation issue.•The data fusion network can generate 10-m dense time-series imagery.•Two test datasets are provided for future model validation.