Integrating an attention-based deep learning framework and the SAFY-V model for winter wheat yield estimation using time series SAR and optical data

•Few previous studies have explored the estimation of LAI using SAR and optical data with large differences in spatial resolution.•The UpSc-AConvGRU model is proposed to improve the LAI estimation accuracy and fill gaps in the optical image time series.•Time-series estimated LAIs and VTCIs are used as inputs of the SAFY-V model for yield estimation.•The yield estimation accuracies are validated at irrigated and rain-fed farmlands, also under different levels of disease.•The proposed approach has low calculation cost and satisfactory yield estimation accuracy at the regional scale.