Abstract Water Surface Elevation (WSE) is an important hydrometric observation, useful to calibrate hydrological models, predict floods, and assess climate change. However, the number of in-situ gauging stations is in decline worldwide. Satellite altimetry, including the recently launched satellite missions (e.g. the radar altimetry missions Cryosat 2, Jason 3, Sentinel 3A/B and the LIDAR mission ICESat-2), can determine WSE only in rivers which are more than ca. 100 m wide. WSE measurements in small streams currently remain limited to the few existing in-situ stations or to time-consuming in-situ surveys. Unmanned Aerial Systems (UAS) can acquire real-time WSE observations during periods of hydrological interest (but with flight limitations in extreme weather conditions), within short survey times and with automatic or semi-automatic flight operations. UAS-borne photogrammetry is a well-known technique that can estimate land elevation with an accuracy as high as a few cm, similarly UAS-borne LIDAR can estimate land elevation but without requiring Ground Control Points (GCPs). However, both techniques face limitations in estimating WSE: water transparency and lack of stable visual key points on the Water Surface (WS) complicate the UAS-borne photogrammetric estimates of WSE, while the LIDAR reflection from the water surface is generally not strong enough to be captured by most of the UAS-borne LIDAR systems currently available on the market. Thus, LIDAR and photogrammetry generally require extraction of the elevation of the “water-edge” points, i.e. points at the interface between land and water, for identifying the WSE. We demonstrate highly accurate WSE observations with a new radar altimetry solution, which comprises a 77 GHz radar chip with full waveform analysis and an accurate dual frequency differential Global Navigation Satellite System (GNSS) system. The radar altimetry solution shows the lowest standard deviation (σ) and RMSE on WSE estimates, ca. 1.5 cm and ca. 3 cm respectively, whilst photogrammetry and LIDAR show a σ and an RMSE at decimetre level. Radar altimetry also requires a significantly shorter survey and processing time compared to LIDAR and especially to photogrammetry.

中文翻译：

---

无人机系统 (UAS) 对小溪中水面高程的观测：雷达测高、激光雷达和摄影测量技术的比较

摘要 水面高程 (WSE) 是一项重要的水文观测，可用于校准水文模型、预测洪水和评估气候变化。然而，全球现场测量站的数量正在下降。卫星测高，包括最近发射的卫星任务（例如雷达测高任务 Cryosat 2、Jason 3、Sentinel 3A/B 和 LIDAR 任务 ICESat-2），只能在超过约 100 米的河流中确定 WSE。100 m 宽。小溪流中的 WSE 测量目前仅限于少数现有的原位站或耗时的原位调查。无人机系统 (UAS) 可以在具有水文意义的时期获取实时 WSE 观测值（但在极端天气条件下有飞行限制），在很短的调查时间内和自动或半自动飞行操作。UAS 机载摄影测量是一种众所周知的技术，可以以高达几厘米的精度估算陆地高程，类似地，UAS 机载 LIDAR 可以估算陆地高程，但不需要地面控制点 (GCP)。然而，这两种技术在估计 WSE 方面都存在局限性：水的透明度和水面 (WS) 上缺乏稳定的视觉关键点使 UAS 对 WSE 的摄影测量估计复杂化，而来自水面的 LIDAR 反射通常不够强目前市场上大多数 UAS 搭载的 LIDAR 系统都可以捕获。因此，LIDAR 和摄影测量通常需要提取“水边”点的高程，即陆地和水界面处的点，用于识别 WSE。我们使用新的雷达测高解决方案展示了高度准确的 WSE 观测，该解决方案包括具有全波形分析的 77 GHz 雷达芯片和准确的双频差分全球导航卫星系统 (GNSS) 系统。雷达测高解决方案显示了 WSE 估计值的最低标准偏差 (σ) 和 RMSE，约。1.5 厘米和约。分别为 3 cm，而摄影测量和激光雷达在分米级显示 σ 和 RMSE。与激光雷达（尤其是摄影测量）相比，雷达测高还需要明显更短的测量和处理时间。雷达测高解决方案显示了 WSE 估计值的最低标准偏差 (σ) 和 RMSE，约。1.5 厘米和约。分别为 3 cm，而摄影测量和激光雷达在分米级显示 σ 和 RMSE。与激光雷达（尤其是摄影测量）相比，雷达测高还需要明显更短的测量和处理时间。雷达测高解决方案显示了 WSE 估计值的最低标准偏差 (σ) 和 RMSE，约。1.5 厘米和约。分别为 3 cm，而摄影测量和激光雷达在分米级显示 σ 和 RMSE。与激光雷达（尤其是摄影测量）相比，雷达测高还需要明显更短的测量和处理时间。