ABSTRACTAtmospheric motion vectors (AMVs) are estimated by selecting and tracking the cloud and water vapour features from geostationary satellite images. Satellite-derived AMVs are extremely helpful in understanding synoptic-scale atmospheric dynamics, circulations, and also potential input parameters of the numerical weather prediction (NWP) models. In India, till now, operational derivation of low-level visible (VIS) AMVs are retrieved using a coarser spatial resolution (4 km) images from INSAT-3D and INSAT-3DR satellites. In this study, high-resolution (HR) low-level visible AMVs are derived from INSAT-3DR HR (1 km) images. The operational VIS AMVs retrieval technique, with some fine-tuning in the algorithm for the use of HR images, is followed for HR VIS wind derivation. The accuracy of the new HR VIS winds is improved significantly compared with corresponding coarser resolution operational VIS winds against numerical model analysed winds. The statistical comparison has been carried out by calculating root mean square vector difference (RMSVD), speed bias, and percentage improvement. These newly retrieved high- and operationally available coarse-resolution VIS winds are further assimilated in the numerical model to assess their impact on short-range weather forecast during 1 to 15 July 2017. Assimilation of HR VIS winds has improved the forecast, especially at upper levels.