We developed a refined gravimetric geoid model for Japan on a 1 × 1.5 arc-minute (2 km) grid from a GOCE-based satellite-only global geopotential model and a regional gravity field model updated in this study. First, we have constructed a regional gravity field model for Japan using updated gravity datasets together with a residual terrain model: 323,431 land gravity data, 77,389 shipborne marine gravity data, and Sandwell’s v28.1 altimetry-derived global marine gravity model. Then, the geoid was determined with the gravity field model. The methodology for gravimetric geoid determination was based on the remove–compute–restore technique with Helmert’s second method of condensation of topography (Stokes–Helmert scheme). Here, the hybrid Meissl–Molodensky modified spheroidal Stokes kernel was employed to minimize the truncation error under an appropriate combination of different kinds of gravity data. In addition, a high-resolution GSI-DEM on a 0.4 × 0.4 arc-second (10 m) grid, together with the SRTM-DEM on a 7.5 × 11.25 arc-second (250 m) grid, was utilized for precisely applying terrain correction to the regional gravity field model. Consequently, we created a gravimetric geoid model for Japan, consistent with 971 GNSS/leveling geoid heights distributed over the four main islands of Japan with a standard deviation of 5.7 cm, showing a considerable improvement by 2.3 cm over the previous model (JGEOID2008). However, there remain some areas with large discrepancies between the computed and GNSS/leveling geoid heights in northern Japan (Hokkaido), mountainous areas in central Japan, and some coastal regions. Since terrestrial gravity data are especially sparse in these areas, we speculated that the largeness of the geoid discrepancies there could be partly attributed to the insufficient coverage and accuracy of gravity data. The Geospatial Information Authority of Japan has started airborne gravity surveys to be covered over the Japanese Islands, and in future, we plan to develop a geoid model for Japan further accurately by incorporating airborne gravity data to come.

中文翻译：

---

使用 GOCE 和更新的区域重力场模型改进日本的重力大地水准面模型

我们在 1 × 1.5 弧分（2 公里）网格上根据基于 GOCE 的纯卫星全球位势模型和本研究中更新的区域重力场模型为日本开发了一个改进的重力大地水准面模型。首先，我们使用更新的重力数据集和残差地形模型构建了日本的区域重力场模型：323,431 个陆地重力数据、77,389 个船载海洋重力数据和 Sandwell 的 v28.1 测高衍生的全球海洋重力模型。然后，用重力场模型确定大地水准面。重力测量大地水准面的方法基于去除-计算-恢复技术和 Helmert 的第二种地形凝结方法（Stokes-Helmert 方案）。这里，在不同类型的重力数据的适当组合下，采用混合 Meissl-Molodensky 修正的球体斯托克斯核来最小化截断误差。此外，还利用 0.4 × 0.4 弧秒 (10 m) 网格上的高分辨率 GSI-DEM 以及 7.5 × 11.25 弧秒 (250 m) 网格上的 SRTM-DEM 来精确应用地形对区域重力场模型的修正。因此，我们为日本创建了一个重力大地水准面模型，与分布在日本四个主要岛屿上的 971 GNSS/水准水准面高度一致，标准偏差为 5.7 厘米，比之前的模型（JGEOID2008）提高了 2.3 厘米。然而，在日本北部（北海道）计算的和 GNSS/水准水准面高度之间仍有一些地区存在较大差异，日本中部的山区和一些沿海地区。由于这些地区的地面重力数据特别稀少，我们推测大地水准面差异较大可能部分归因于重力数据的覆盖范围和准确性不足。日本地理空间信息管理局已经开始对日本列岛进行空中重力测量，未来我们计划通过结合未来的空中重力数据，进一步开发日本的大地水准面模型。