Separation of permanent water body and temporal water body is essential for the grasp of global water distribution. Global 3-second Water Body Map (G3WBM) is a global scale high-resolution water body map recently-developed by the authors, which separates water body types by the idea of water frequency, using multi-temporal Landsat data. In this paper, we further improved the accuracy and computational efficiency of G3WBM algorithms, and refined the resolution from 90m to 30m. The validation against U.S. National Hydrography Dataset revealed that, although some small water bodies could not be visualized, more than 85% of permanent water bodies were classified correctly. Comparison between the water body mask constructed in this paper and G3WBM in the U.S. showed that water judgment accuracy was improved and over 2600km2 of temporal water has been newly separated from permanent water.