Numerous studies have demonstrated the effectiveness of CYGNSS (Cyclone Global Navigation Satellite System) data to detect inland water bodies. However, most of them focus on the detection of surface water extent, rather than the water levels and depths. Most of the existing studies on inland water level altimetry using CYGNSS data are based on the CYGNSS raw IF (raw Intermediate Frequency) data, and use either the time-delay or the phase methods. Although high accuracy can be obtained, raw IF data are currently not a CYGNSS standard data product, and thus it cannot be applied to emergency detection of sudden flood events, or long time-series monitoring of a single large inland water body. Based on these research gaps, this study presents an effective method to estimate the water levels and depths (on a 3 km grid) by combining CYGNSS L1 standard data with DTM (Digital Terrain Model) information. We compared the CYGNSS retrieved water levels with other water level reference data for eight case studies with different characteristics (3 floods, 3 lakes, and 2 reservoirs). The spaceborne LiDAR ICESat-2 and GEDI are used for the flood case studies, and water levels from DAHITI and Hydroweb databases are used for long-term changes of the lakes and reservoirs. All comparative validations yield encouraging results. For flood cases, comparison with ICESat-2 and GEDI showed strong correlation (mean R values for the three case studies were 0.98). The mean bias of CYGNSS retrieved water levels was - 0.26 m, and the mean RMSE was 1.34 m. For lakes and reservoirs, the comparison to DAHITI and Hydroweb showed a mean correlation value of 0.62 (R), while the mean bias and RMSE were - 0.63 m, and 2.61 m, respectively. Finally, the sources of uncertainty are discussed, including the effect of topography discretization, and the effect of uncertainty in the determination of surface water boundaries. This study demonstrates the feasibility of combining CYGNSS L1 standard data with DTM data to measure inland water levels, highlighting its suitability to monitor both flash floods and long-term changes of lake and reservoir water levels. This rapidly updatable water level information will contribute to further comparisons and hydrological researches.

中文翻译：

---

使用 CYGNSS L1 数据和 DTM 估算内陆水位的快速有效方法：应用于洪水、湖泊和水库监测


大量研究表明 CYGNSS（气旋全球导航卫星系统）数据在检测内陆水体方面的有效性。然而，它们中的大多数侧重于检测地表水范围，而不是水位和深度。大多数使用 CYGNSS 数据进行内陆水位测高的现有研究都是基于 CYGNSS 原始 IF（原始中频）数据，并使用时延或相位方法。虽然可以获得高精度，但原始 IF 数据目前不是 CYGNSS 标准数据产品，因此无法应用于突发洪水事件的紧急检测，或单个大型内陆水体的长时间序列监测。基于这些研究空白，本研究提出了一种通过将 CYGNSS L1 标准数据与 DTM（数字地形模型）信息相结合来估计水位和深度（在 3 公里网格上）的有效方法。我们将 CYGNSS 检索的水位与 8 个具有不同特征的案例研究（3 次洪水、3 个湖泊和 2 个水库）的其他水位参考数据进行了比较。星载 LiDAR ICESat-2 和 GEDI 用于洪水案例研究，来自 DAHITI 和 Hydroweb 数据库的水位用于湖泊和水库的长期变化。所有比较验证都产生了令人鼓舞的结果。对于洪水案例，与 ICESat-2 和 GEDI 的比较显示出很强的相关性（三个案例研究的平均 R 值为 0.98）。CYGNSS 检索水位的平均偏差为 - 0.26 m，平均 RMSE 为 1.34 m。对于湖泊和水库，与 DAHITI 和 Hydroweb 的比较显示平均相关值为 0.62 （R），而平均偏差和 RMSE 分别为 - 0.63 m 和 2.61 m。 最后，讨论了不确定性的来源，包括地形离散化的影响，以及不确定性对地表水边界确定的影响。本研究证明了将 CYGNSS L1 标准数据与 DTM 数据相结合来测量内陆水域水位的可行性，突出了其适用于监测山洪暴发以及湖泊和水库水位的长期变化。这些可快速更新的水位信息将有助于进一步的比较和水文研究。