A gridded salinity dataset with high resolution is essential for investigating global ocean salinity variability and understanding its role in climate and the ocean ecosystem. In this study, a new version of the Institute of Atmospheric Physics gridded salinity dataset with a higher resolution (0.5° by 0.5°) is provided by using a revised ensemble optimal interpolation scheme with a dynamic ensemble. The performance of this dataset is evaluated using “subsample test” and independent observations and synthetic datasets. Compared with the previous 1° by 1° resolution IAP product, the new dataset is more capable of representing regional salinity changes with the meso-scale and small-scale signals (i.e., in the coastal and boundary currents regions), meanwhile, maintains the large-scale structure and variability. Therefore, the new dataset complements the previous data product. Besides, the new dataset is compared with in situ observations and several international salinity products for the salinity multiscale variabilities and patterns. The comparison shows the smaller magnitude of mean difference and RMSD in basin scale for the new dataset, some differences in strength and fine structure of the “fresh gets fresher, salty gets saltier” surface and subsurface salinity pattern amplification trends from 1980 to 2017, a broad similarity for the salinity changes associated with El Niño-Southern Oscillation (ENSO) and a consistent salinity dipole mode in the tropical Indian Ocean (S-IOD). These results support the future use of gridded salinity data and the research of the ocean salinity changes under climate change.