Abstract
<jats:p>Unmanned aerial vehicles (UAVs) are increasingly used for non-contact hydraulic measurements in rivers and streams, but reported accuracy varies across methods, sites, and flow conditions. Here, we present a quantitative review of published UAV applications for water surface elevation (WSE), bathymetry, surface velocity, and discharge measurements. A total of 96 peer-reviewed studies with independent in-stream or field-based reference measurements were compiled and analyzed. In the compiled literature, mean WSE uncertainty was 0.15 m for photogrammetry, 0.18 m for Light Detection and Ranging (LiDAR), and 0.036 m for radar, with radar showing the lowest reported uncertainty but the smallest evidence base. Bathymetry results were more variable, with compiled mean uncertainties ranging from 0.08 to 0.25 m depending on method and validation setting. Ground Penetrating Radar (GPR) showed the lowest mean uncertainty, but this result was based on a single study, while bathymetric LiDAR showed a mean uncertainty of 0.13 m from two studies. For UAV-based surface velocimetry, the compiled mean absolute percentage error (MAPE) was 16.8% and the mean percentage error (MPE) was −3.1%. Among the methods reviewed, Particle Tracking Velocimetry (PTV) showed the lowest MAPE (∼9.6%), whereas Particle Image Velocimetry (PIV) showed the lowest systematic bias (MPE ∼ +0.8%). Surface velocimetry errors mainly reflected image quality, surface visibility, and validation setup. Discharge estimates derived from surface velocimetry based methods showed lower error (MAPE ∼12.9%) than Manning equation based methods (MAPE ∼20.6%) in the current dataset. However, discharge accuracy also depended on cross-section data, bathymetry, and velocity conversion. Plain Language Summary Unmanned aerial vehicles, commonly known as drones, are increasingly used to measure river water levels, channel shape, surface flow speed, and discharge without placing instruments directly in the water. These approaches can improve safety and make it easier to collect measurements during floods or at sites that are difficult to access. However, their accuracy varies depending on the sensing method and on local conditions such as water clarity, turbulence, vegetation, lighting, and channel shape. In this study, we compiled 96 published field studies and quantitatively compared the performance of drone-based methods for water surface elevation, bathymetry, surface velocity, and discharge. We found that water surface elevation can often be measured with errors on the order of centimeters to decimeters. Radar-based studies reported the lowest average uncertainty in the compiled literature, but only a small number of such studies are currently available. Bathymetry is generally more variable because accuracy depends strongly on water clarity, depth, and bottom visibility. Surface flow velocity measurements are often reliable, with an overall compiled mean absolute percentage error of about 17%. In the current dataset, discharge estimates based on surface-velocity methods showed lower error than estimates derived primarily from Manning-equation-based approaches. Overall, this review provides quantitative guidance for selecting UAV-based hydraulic methods and for understanding their expected accuracy and limitations in river monitoring and management.</jats:p>