Real-time centimetre-level precise positioning from Global Navigation Satellite Systems (GNSS) is critical for activities including landslide, glacier and coastal erosion monitoring, flood modelling, precision agriculture, intelligent transport systems, autonomous vehicles and the Internet of Things. This may be achieved via the real-time kinematic (RTK) GNSS approach, which uses a single receiver and a network of continuously operating GNSS reference stations (CORS). However, existing CORS networks have often been established simply by attempting regular spacing or in clusters around cities, with little consideration of weather, climate and topography effects, which influence the GNSS tropospheric delay, a substantial GNSS positional error and which prevents homogeneous RTK accuracy attainment. Here, we develop a framework towards optimizing the design of CORS ground infrastructure, such that tropospheric delay errors reduce to 1.5 mm worth of precipitable water vapour (PWV) globally. We obtain average optimal station spacings of 52 km in local summer and 70 km in local winter, inversely related to the atmospheric PWV variation, with denser networks typically required in the tropics and in mountainous areas. We also consider local CORS network infrastructure case studies, showing how after network modification interpolated PWV errors can be reduced from around 2.7 to 1.4 mm.

中文翻译：

---

从对流层效应的角度优化全球导航卫星系统网络实时运动基础设施以实现均匀定位性能

全球导航卫星系统 (GNSS) 的实时厘米级精确定位对于滑坡、冰川和海岸侵蚀监测、洪水建模、精准农业、智能交通系统、自动驾驶汽车和物联网等活动至关重要。这可以通过实时动态 (RTK) GNSS 方法实现，该方法使用单个接收器和连续运行的 GNSS 参考站 (CORS) 网络。然而，现有的 CORS 网络通常只是通过尝试规则间隔或城市周围的集群来建立，几乎没有考虑天气、气候和地形影响，这些影响会影响 GNSS 对流层延迟、大量 GNSS 位置误差并阻止实现同质 RTK 精度. 这里，我们开发了一个优化 CORS 地面基础设施设计的框架，以便在全球范围内将对流层延迟误差减少到 1.5 毫米的可降水蒸汽 (PWV)。我们获得了当地夏季 52 公里和当地冬季 70 公里的平均最佳站间距，与大气 PWV 变化成反比，热带和山区通常需要更密集的网络。我们还考虑了本地 CORS 网络基础设施案例研究，展示了在网络修改后如何将插值 PWV 误差从大约 2.7 毫米减少到 1.4 毫米。与大气 PWV 变化成反比，热带和山区通常需要更密集的网络。我们还考虑了本地 CORS 网络基础设施案例研究，展示了在网络修改后如何将插值 PWV 误差从大约 2.7 毫米减少到 1.4 毫米。与大气 PWV 变化成反比，热带和山区通常需要更密集的网络。我们还考虑了本地 CORS 网络基础设施案例研究，展示了在网络修改后如何将插值 PWV 误差从大约 2.7 毫米减少到 1.4 毫米。