Contactless river discharge surveying with UAS hydrometry: Performance evaluation using a large and diverse set of river cross sections

River discharge plays a crucial role in hydrologic studies and water resource management. Accurate discharge estimations enable significant advancements in scientific research and water-related decision-making processes. Given that discharge is the product of flow area and flow velocity, traditional in situ measurements of river discharge typically require detailed data on river water level, river bathymetry, and the bulk velocity of the river cross-section. However, such manual measurements are time-consuming and impractical in certain situations, such as rivers in remote, hard-to-access regions or those experiencing extreme high-flow events. With the increasing availability of technical and computational resources, remote sensing offers significant potential for improving discharge estimation. Among these technologies, Unmanned Aerial Systems (UAS) have emerged as a valuable solution for improving discharge estimation. Their low cost, high accuracy, and ability to cover large areas make them particularly effective for monitoring in remote or hard-to-access locations. While numerous studies have developed and demonstrated the feasibility of UAS-based discharge estimation algorithms, their evaluations are often limited to specific sites. Thus, questions remain regarding the adaptability of these algorithms across diverse river systems.

Funded by European Union's Horizon Europe project UAWOS (Unoccupied Airborne Water Observing System), This work focuses on evaluating the performance of UAS-based discharge estimation algorithms across a diverse set of cross-sections to enhance their generalizability. Specifically, this work seeks to address the following key questions: 1, how well the discharge algorithm performs based on UAS velocimetry, bathymetry and water surface elevation across different cross-sections? 2, which input datasets and river characteristics may limit the performance, and how sensitive they are? 3, how can we improve the discharge estimation strategies?

To answer the questions above, we applied several bulk velocity estimation models on UAS hydrometry parameters to calculate the river discharge among various types of rivers: Rønne Å River (Sweden), Isar River (Germany), Po River (Italy), Orco River (Italy), and Torne River (Sweden). We utilized various in situ discharge measurements to assess the accuracy of our algorithms and investigated how specific cross-sectional properties affect performance. We further systematic analysed the uncertainty from the inputs and models, and discovered strategies to optimize the discharge estimation results by utilizing Bayesian inference. Overall, this study shows that advanced UAS hydrometry technique is an accurate and reliable way for river discharge estimation, providing valuable insights for hydrological studies and water resource management.