2025 Volume 1 Pages 36-59
This paper provides an overview of recent advancements in river measurement methods. Non-contact methods, that have been developing and becoming popular over the past 20 years, are categorized into three types: 1) ground-based, 2) airborne, and 3) spaceborne. In-situ measurements are also discussed, by focusing on different approaches and the properties targeted for measurements. The paper additionally addresses how to share and pre-condition collected data. Recent developments in data processing, analysis, and uncertainty assessments are summarized. The paper finally outlines potential areas for future research and exploration.
hydrometry, non-contact method, flow rate, velocity, bathymetry, materials transport, uncertainty analysis, data share
Advancements and the widespread use of new measurement technologies in rivers allow researchers and engineers to gather large amounts of data that was, previously, difficult to acquire using conventional technologies, both in terms of quantity and quality. As climate change continues, runoff patterns and river morphology are likely to gradually change. Collecting high-quality data that helps us understand changing trends and develop evidence-based adaptation measures are currently underway. To give river researchers and engineers a clear understanding of the latest river measurement technologies and potential future developments, in this review, we first discuss the latest developments in traditional approaches for data collection, as well as the growing use of alternative remote sensing measurement methods. We, then, provide an overview of new trends in data acquisition and processing, as well as an uncertainty assessment for measured water flow and other variables. We, finally, address the issue of data accessibility for a wide range of users.
In this section, we provide an overview of the latest developments in river measurement methods and highlight growing trends in new approaches, most of which emerged from remote sensing technology. We, first, review the remote sensing approach that encompasses both near-field remote sensing (comprising ground-based and airborne methods) and far-field remote sensing (here, spaceborne methods). These techniques utilize a variety of wavelengths from visible light to microwaves. We also summarize emerging on-site measurement approaches used to assess various properties, including surface water flow using the remote sensing approach, as well as in-stream tracer methods, groundwater flow, river bathymetry, and sediment flux. Fig. 1 provides a schematic of the above methods. We, finally, introduce citizen participation for river measurements. Table 1 summarizes measurement methods, their corresponding spatio-temporal resolutions, and remarks. Below, we explain each method in detail.
Fig. 1. A schematic of the methods reviewed in this section.
Recent technological advancements in both close range (ground-based) and remote (airborne) sensing have enabled the precise measurement of surface water levels 1, topography, and water flow with unprecedented spatial and temporal resolution 2. LiDAR (Light Detection and Ranging) technology, utilized infrared or green laser pulses from both land and air, allows for the measurement of water surface elevation and river bathymetry with a substantially higher spatial resolution. LiDAR technology used for measuring river bathymetry from air is often referred to as Airborne LiDAR Bathymetry (ALB)3–5. These methods have become widely used in river research and practical applications6. The exponential growth of remote sensing technologies can be attributed to several factors:
The measurement of flow velocity and discharge using images can be broadly categorized in two ways: frame-comparing methods such as Large-Scale Particle Velocimetry (LSPIV) or optical flow (in computer vision) and frame-stacking methods represented by Space-Time Image Velocimetry (STIV). LSPIV (or other frame-comparing methods) quantifies the temporal evolution of the 2-dimensional, 2-component surface flow velocity distribution7, while STIV acquires the time-averaged, cross-sectional distribution of the streamwise velocity component8. Due to poor image quality, the latter technique is most suitable for stable flow rate measurements. These techniques, developed after 1980, have been extensively documented in many scientific studies9–15. Ground-based deployments are preferred for long-term continuous flow monitoring8,13,14.
Although image-based techniques have the advantage of yielding a large amount of data over a rather short measurement time, they may present shortcomings in sensitivity regarding measuring conditions (appropriate tracers and illumination conditions), as well as the selection of processing procedures (e.g. for the processing algorithm and cross-correlation parameters) needed to obtain accurate quantitative information7,10. Detert (2021)16 reviewed existing field data sets, pointed out several issues that can arise during field applications, and discussed how to mitigate such issues to ensure the accuracy of obtained data. Bodart et al. (2024)17 examined the sensitivity of data processing procedures and parameters, and then proposed a framework for identifying appropriate settings.
Aerial imagery using UAVs is highly flexible and safe; however, adverse weather conditions such as rain and strong winds can limit measurement capability using entry- to middle-class UAVs18,19. The use of smartphones for imaging has also advanced20–22. Smartphones have built-in sensors that enable the scaling of taken images. Image-based river velocity and discharge measurement techniques are relatively easy to deploy and use, holding promise for hydrological measurements in the future, especially in developing countries23.
Table 1. A list of methods reviewed in this section.
| Section and method | Measuring variables | Time Resolution | Spatial Resolution | Notes/Comments |
|---|---|---|---|---|
| Sec. 2.(1). Ground-Based Optical Imaging (e.g., LSPIV, STIV) | - Typically used for 2D surface velocity fields and can be used for discharge estimations | - Continuous, if cameras are permanently installed | - Sub-meter scale to a few meters, depending on camera resolution and distance to the river surface | - Requires surface features (e.g., tracers, surface foam, surface ripples) and suitable lighting |
| - Event-based measurements (e.g., floods) | - Susceptible to weather, illumination, and background noise | |||
| Sec. 2.(1). UAV-Based Optical Imaging (photos/videos for LSPIV, STIV, SfM) | - Captures 2D surface velocity fields (optical) or topography via Structure-from-Motion (SfM) | - On-demand flights; can be repeated daily/weekly | - Sub-decimeter to sub-meter resolution, depending on flight altitude and sensors | - Flexible and safe data collection |
| - Generally, short campaigns of minutes to hours | - Orthorectification can be challenging under wind or unstable flights | |||
| - Depth retrieval possible for very clear, shallow water | ||||
| Sec. 2.(1). Ground-Based Radar (Microwave/ UHF Doppler) | - Non-contact velocity measurements over water surfaces | - Continuous/real-time (when installed permanently) | - “Beam footprint” typically on the order of meters to tens of meters | - Good for continuous discharge monitoring |
| - Event-based (mobile deployments) | - May struggle under low-flow (weak signal backscatter) | |||
| - Affected by roughness/wind waves | ||||
| Sec. 2.(1). UAV-Based Radar | - Measures surface velocity along tracks | - On-demand flights, typically short campaigns | - Meter-scale footprint, depending on radar beamwidth and altitude | - Useful for hazard situations (floods) where ground access is difficult |
| - Requires stable UAV positioning | ||||
| Sec. 2.(1). Airborne LiDAR (infrared vs. green/ALB) | - Topographic mapping | - Campaign-based; repeated surveys every few months to years | - Infrared LiDAR: sub-meter on terrain but limited penetration in water | - ALB can penetrate water up to a certain depth (low turbidity) |
| - Green LiDAR (ALB): sub-meter (bathymetry possible in clear, shallow water) | - High cost, specialized sensors | |||
| - Provides dense, high-resolution topography | ||||
| Sec. 2.(2). Spaceborne Optical Imaging (e.g., SkySat, other multispectral) | - Satellite video for surface velocimetry. | - Revisit intervals vary by satellite, from near-daily (constellations) to weekly/monthly | - Meter to tens-of-meters scale (depends on sensor: e.g.,
|
- Highly dependent on cloud cover (passive sensors) |
| - Multispectral approaches for estimating depth | - Multispectral approaches can estimate depth in clear/ shallow rivers | |||
| - Satellite video (SkySat) allows for surface velocimetry but with limited scenes/frame lengths | ||||
| Sec. 2.(2). Spaceborne Radar (SAR, altimetry, SWOT) | - SAR for surface water extent/velocity (interferometry) | - Revisit intervals from days to weeks (e.g., 6 to 12 days for some SAR;
|
- Tens of meters to hundreds of meters (pixel size / altimeter footprint) | - All-weather, day/night capability |
| - Altimetry for water levels | - Resolution may be too coarse for small rivers or short-duration floods | |||
| - SWOT aims at
|
||||
| Sec. 2.(3). In-Stream Tracers (salt, fluorescent dyes, DNA) | - Mean velocity and flow rates | - Snapshot measurements from single or repeated injections | - Effectively “point-to-integrat- ed” measurements along a reach; resolution depends on sampling strategy | - Useful for diagnosing flow rates in complicated settings (low-flow, seepage) |
| - Time scale from hours to days (depending on travel time) | - DNA tracers allow ultra-low detection limits but may face degradation/adsorption issues | |||
| - Continuous injection + auto-samplers can yield near-real-time flow estimates (within
|
||||
| Sec. 2.(4). Ground Water Dynamics | - Ground water level | - Campaign-based | - Few meters to a few kilometers | -A single well is applicable to small domains |
| - Ground water velocity | -Combining with data assimilation can improve consistency (accuracy) and spatial and temporal resolution | |||
| Sec. 2.(5). GPR (Ground Penetrating Radar, ground or UAV-mounted) | - Provides subsurface profiles (e.g., depth to bed) | - Campaign-based; can be event-based | - Sub-meter to a few meters vertical resolution for subsurface/bathymetry | - Works under certain conductivity conditions |
| - Rapid data collection but post-processing needed | - UAV-based GPR is emerging but has radio-transmission regulations and range limitations | |||
| Sec. 2.(5). Acoustic Sounding (Single- and Multi-Beam Echo Sounders) | - Bathymetric mapping | - Campaign-based, repeated surveys as needed | - Sub-meter to decimeter for MBES (dense bathymetric coverage) | - Continuous profiles feasible (moving-boat ADCP) |
| - Real-time possible with ADCP-based boat surveys | - Limited by water depth, turbidity not usually an issue (compared to optical) | |||
| Sec. 2.(5). Suspended Sediment & Bedload (TDR, hydrophones, geophones, acoustic backscatters) | - TDR: high-concentration suspended load | - Typically continuous with installed sensors | - Local measurement points can be set in vertical arrays | - No universal way to measure all modes of sediment transport |
| - Hydrophones/geophones: detect particle impacts or vibrations | - Also event-based for flood sediment pulses | |||
| - In-grain accelerometers for gravel movement studies | ||||
| - Acoustic backscatters: Suspended sediment and bedload | ||||
| Sec. 2.(6). Citizen / Crowdsourced (Smartphone apps, low-cost gauges, social media videos) | - Not specific | - Opportunistic or continuous (if user frequently logs data) | - Highly variable; from single point (staff gauge) to spatial mapping (multiple users along a river) | - Offers broad spatial coverage at minimal cost |
| - Flood events commonly recorded | - Quality/metadata can vary widely | |||
| - Image velocimetry from YouTube or smartphone video is growing | ||||
| - Requires standardized protocols for robust data |
In addition to advances in image-based methods, the advancement and adaptation of microwave/UHF-Doppler-radar technology are also contributing to the proliferation of flow measurements24–28. Radar velocimetry is affected by errors, mainly during low flow when the backscatter of a beam can be compromised by high noise. Under this condition, the free-surface lacks the small perturbations needed to produce a detectable return signal29. Radar technology mounted on UAVs permits measurements of along-track surface velocity by spot dwelling within a river cross section at a location where the maximum surface velocity, which can be converted to a cross-sectional mean velocity assuming a cross-sectional velocity distribution, is recorded30.
Airborne SAR (Synthetic Aperture Radar) interferometry can provide the water-surface velocity component perpendicular to the flight direction. By combining remotely sensed river width with the channel slope, river discharge can be estimated31.
The remote sensing techniques described above measure free-surface flow velocity distributions. Recent applications of remote sensing techniques have revolutionized the safety of flood data acquisition and have led to the collection of vast amounts of river data32. To estimate discharge using the surface velocity distribution, the velocity-area method is widely used33. For estimating the area averaged velocity of each subsection of a river cross-section, we need to convert surface velocity to the depth averaged (or area averaged) flow velocity. In the past, this conversion was an interesting and important topic in river science and engineering34,35, and has been revisited in recent years36–38. One of the recent innovations in this field is the application of entropy theory for correlating surface velocity with the cross-sectional-mean-velocity39. This approach is later reviewed in Section 4.(2) in detail. Bathymetric data is also fundamental for both river management and research. The cross-sectional shape of a river section is essential data for discharge calculations. Remote sensing bathymetric survey methods represent another important recent innovation and will be reviewed in Section 2.(5).
One significant challenge that remains to date is to consider the influence of wind that alters the free-surface velocity40–42. During large floods, turbulence originating from the riverbed may dominate the velocity profile, including the water surface, so the impact of wind to surface flow velocity and wind wave generation is expected to be minor. However, further quantitative studies are required to estimate the effect of wind on the velocity distribution across the cross section.
Spaceborne methodsSatellites observe Earth’s surface water based on three methods:
Methods 1 and 2 are defined as passive observations and Method 3 is defined as an active observation40. The wavelengths used by Methods 1 to 3 increase as follows: visible light: order of 380 nm (Method 1)
Active observation using microwaves enables researchers to obtain surface information day and night. The active microwave method is particularly important for flood measurements where rainfall plays a crucial role. SAR can detect changes in the target surface height and conditions by actively emitting radio waves and by utilizing polarization and interference43. Interferometry can be implemented by simultaneously using multiple antennas or by observing the same location at different times using one device. The latter is useful for detecting changes over a large time scale.
Space-borne SAR can measure river surface velocity. This is accomplished using interferometric processing of SRTM’s X band SAR data40. Romeiser et al. (2005)44 explored the potential of TerraSAR-X for routinely measuring river water levels and velocities from a satellite, and concluded that a reach-length of more than 1 km is necessary to maintain accuracy40, although a spatial resolution less than 1 km for the velocity distribution is possible45. A recent simulation-based study reported that for effective measurements minimum river width should be 300 m, the flow direction should be within 60 degrees of available satellite look directions, water surface roughness should be sufficient to ensure a good signal-to-noise ratio of radar images, and the water surface should be free of ice46.
The coverage area of a single satellite over a given time is limited to the satellite ground track and is further constrained by Earth’s rotation, restricting the observation frequency for a specific target area. This restriction can be resolved by deploying multiple satellites across various positions, known as a satellite constellation. Advances in satellite manufacturing technology have enabled high-quality observations using small, lightweight satellites, making it feasible to achieve satellite constellations at realistic launch costs, even by private entities47.
The necessary time resolution for understanding the time evolution of fluvial wave propagation varies depending on the size and characteristics of the river basin targeted for investigation. A daily or longer time resolution is sufficient for large continental rivers, but a time resolution of one hour or less is necessary for floods occurring within small watersheds, steep-gradient rivers, or urbanized areas. Most active observations are obtained using one or very few satellites, which limits the applicability of space-borne methods for monitoring such short duration events. Therefore, satellite constellations are essential for river observations during short-duration floods. Kitajima et al. (2021)48 estimated that approximately 20 satellites are needed for flood observations in mountainous regions.
River measurements using Radar/Laser and Synthetic Aperture Radar (SAR) satellite altimetryThere are two types of satellite altimeters: laser altimeters and radar altimeters. Laser altimeters (or LiDAR) use light waves, while radar altimeters use lower frequency waves. From an operational perspective, both types of satellite altimeters measure the travel time between signal transmission and sensor reception49, which is converted in the distance by separating the instrument from Earth’s surface. SAR satellites and passive microwave sensors have been used to detect flooded areas50,51. Water surface elevations can be estimated by combining the detection of the edge of water covering a flooded area, if river morphology can be assumed as stable. Existing ground elevation data determined by other means can be used to estimate the elevation of the water’s edge as the difference between the two datasets. Examples of passive microwave applications are the Advanced Microwave Scanning Radiometer from Earth Observing System (AMSR-E) data used by Temimi et al. (2007)52 to monitor streamflow within the Mackenzie River basin in Canada, and Temimi et al. (2011)53 for determining river discharge during the 2008 flood in Iowa. Another noteworthy example is provided by Brakenridge et al. (2007)54 who used 37 Ghz AMSR-E data for globally inferring river discharge. Their approach is based on the ratio between the brightness temperature measured for a pixel unaffected by a river and a pixel centered over the river itself, respectively. Based on this approach, Tarpanelli et al. (2013)55 investigated the capability of Moderate Resolution Imaging Spectroradiometer (MODIS) for estimating river discharge at gauged and ungauged sites. The use of remotely sensed water surface elevation in rivers and streams for estimating river flow rates is an emerging research topic56. Both radar pulses and SAR have been used for water level measurements. Examples of recent radar altimetry technology including TOPEX/Poseidon (TP, radar altimeter, operation from 1992 to 2006), the European Remote-Sensing Satellite 2 (ERS-2, radar altimeter, operation from 1995 to 2011), and the Environmental Satellite (ENVISAT, radar altimeter, operation from 2002 to 2012) missions offered important information for water elevations in large rivers, lakes, and floodplains57.
The capability of coupling measurements of river velocity derived from MODIS and water levels derived from the ENVISAT Advanced Radar Altimeter (RA-2) for river discharge estimations was thoroughly investigated by Tarpanelli et al. (2015)58. Initiated in 2018, ICEsat-2 ATL03 Photon utilizes a laser altimetry for this purpose59. However, resolution in the direction orthogonal to the orbit was low, making water level observations for small lakes and rivers difficult60. The USA, France, Canada, and the UK collaborative project SWOT (Surface Water and Ocean Topography, SAR altimeter) launched a satellite with two Ka-band interferometric radars on a SpaceX Falcon 9 rocket in December 2022. The SWOT project enables global water surface height observations at an accuracy of approximately 10 cm, with a 21-day observation cycle, except for polar regions61.
Some regions on Earth lack sufficient conventional hydrological observations. In some areas, observational capacity is deteriorating62. The spread of spaceborne remote-sensing water level observations is anticipated to facilitate regular hydrological data acquisition for rivers wide enough to be observed from space during hydrologic events where water level changes are not sudden (due to the time constraint of satellite observations, refer to the discussion at the end of Section 2.(2)). Continuous and regular data acquisition made possible by satellite observations aids efficient water resource management by estimating river water-surface slopes and water depth estimations63, flow rate estimations based on rating curves64,65, flow rate estimations in river networks66–68, flood routing in large continental rivers69, and flood mapping in remote and poorly-gauged catchments70. Estimations of flow rates within river networks, obtained by assimilating satellite observations in canonical governing open-channel equations, promise a change in the streamflow monitoring nexus, as discussed in Section 4.(3).
River measurements using optical satellitesSkySat, launched in 2013 by Skybox Imaging (USA, California), was the first commercial optical satellite with video recording capabilities. A constellation consisting of 21 satellites was, subsequently, established71. The SkySat program has been transferred to Planet Labs (USA, California). SkySat offers video recording in a single wavelength band, with a spatial resolution of 1 meter, and multispectral images with a resolution of approximately 2 meters. The videos have a maximum length of 120 seconds and a frame rate of 30 Hz. Geometrically uncorrected video files of approximately 2.5K by 1K pixels and TIFF files are provided71. Legleiter and Kinzel, (2021)72 utilized satellite videos for a Particle Image Velocimetry (PIV) analysis. Obtained results were compared to flow velocity distributions from in-situ Acoustic Doppler Current Profiler (ADCP) measurements, and quantitatively confirmed agreement.
Flow tracking using in-stream tracersTracers in the present context are dissolved substances in surface water and groundwater that can be used for inferring flow rates by measuring and analyzing concentrations as the flow travels downstream73. Tracer methods are particularly useful for measuring flow rates in river sections where other methods are impractical, such as during low-flow conditions when flow velocities are low and significant water flows enter into the stream as seepage73,74.
Flow tracers can be categorized into five groups: natural ions, stable isotopes, radioactive isotopes, other solutes (such as heavy metals, organic compounds), and artificial tracers73. Natural environmental substances such as natural ions and stable isotopes are often used as flow tracers to monitor surface and subsurface flows73,75. Artificial tracers are also employed for stable and controlled measurements74. Commonly used artificial tracers include ions such as halogen ions (e.g. sodium chloride: salt) and fluorescent dyes. Halogen ions are easy to obtain and analyze, while fluorescent dyes have the advantage of lower detection limits. A combination of multiple tracers is often used to evaluate water budgets.
Automation is critical for practical use of this method. Clow & Fleming (2008)76 reported an automated system for injecting and measuring Rhodamine WT, and confirmed agreement with the weir flow rate within 6.3%. Sentlinger et al. (2019)74 utilized sodium chloride (salt) as an artificial flow tracer to measure river flow rate. They proposed a system for continuous and automated observations, confirming that uncertainties in their measurement were below 10% for flow rates within a range of 0 to 10 m
To prevent DNA concentration reduction during transport in groundwater, Mikutis et al. (2018)78 proposed silicon-coated silica spheres with adsorbed DNA fragments. This method increased the stability of DNA fragments under high temperature and non-neutral pH conditions compared to non-coated DNA tracers. Biodegradable materials are also used to stabilize DNA tracers. For example, alginate chitosan (a polysaccharide derived from the chitin found in the shells of crustaceans79) can be combined with DNA fragments to stabilize DNA and to prevent concentration reductions during surface and subsurface water flows80.
Groundwater dynamics characterizationsIn the previous sub-section, we reviewed the application of environmental flow tracers for surface and sub-surface flow measurements. This sub-section outlines methods for measuring groundwater without the use of flow tracers.
Groundwater level is critical for understanding the distribution and flow of groundwater. One straightforward method for measuring groundwater levels and changes is the drilling of observation wells. The method can be divided into two types: single-hole methods that analyze data from one observation well81, and multi-hole methods that involve multiple observation wells. Several flow measurement methods utilizing groundwater wells have been developed82. Subsurface flow can be estimated based on the slope of groundwater, which can be obtained using the multi-hole method. By measuring water flow in a well using local dye/particle tracing or mechanical velocimeters, we can estimate groundwater flow. Such an approach is considered unique to Japan83, possibly due to its suitability for detecting complex ground water flow patterns arising from the nation’s narrow yet complex geology.
Groundwater flow can be indirectly measured by detecting natural electric potential changes from the ground. This method has been used to identify spring water in boreholes up to 300 meters deep84. While the indirect method can detect groundwater flow on a large spatial scale, information regarding local geology and normal groundwater flow at the measurement site is necessary for quantitative analysis of the flow84.
Measuring groundwater distributions and flows from aircraft or artificial satellites is challenging. However, we can indirectly infer groundwater dynamics by observing the condition of vegetation cover85. When soil is exposed to open sky, we can evaluate soil moisture content using the received spectral distribution or the backscattering information obtained by SAR. A multispectral microwave is also required for estimating soil moisture from space86. Since the 2000s, when satellite data became more accessible and popular87, the analysis of variations in gravity fields obtained from the Gravity Recovery and Climate Experiment (GRACE) satellite, as well as changes in groundwater volume or groundwater flow from terrain shape changes obtained by Interferometric Synthetic Aperture Radar (InSAR) have been used in large-scale groundwater management88–90.
Measuring river bathymetry and sediment fluxBathymetric data (two-dimensional) is essential for both river management and research. The cross-sectional shape (one-dimensional) of a river section is crucial for calculating discharge using the velocity-area method. In both developed and developing countries, there is still a substantial need for indirect estimations of riverbed shapes, especially for small and medium-sized rivers23. However, obtaining these data using traditional point-wise measurements - using rods, weighted ropes, or single echo sounders deployed by humans, manned vessels, or unmanned vessels - can be labor-intensive and costly.
Measurements of river bathymetry using remote sensing methods shows promise despite the rapid attenuation of electromagnetic waves in water. Several techniques can be utilized for estimating river bathymetry using airborne technologies91.
Advancements in technology, such as LiDAR and UAVs combined with Structure from Motion (SfM), has deepened our understanding of river morphology92. Conventional LiDAR uses infrared wavelengths, that are quickly dissipated in water, making underwater measurements impossible. LiDAR utilizing green lasers, known as Airborne LiDAR Bathymetry (ALB), has been developed to remotely measure coastal and river bathymetry3–5 since green light penetrates deeper within the water column. ALB is effective under low water turbidity and small surface waves93. For shallow water depth with clear water conditions, bathymetry can be reconstructed from aerial photos taken from UAV or manned airplanes using the Structure from Motion (SfM) technique94,95. Multi-spectral imaging can also be employed to estimate river bathymetry96. SfM reconstructs bathymetry by matching the image pattern of the river bed taken from different viewpoints. This process requires a high image resolution to clearly resolve river bed textures, which necessitates capturing images at relatively low altitudes. In contrast, the multi-spectral method does not require such high image resolution, making it feasible to capture images from high altitudes, including space.
Ground Penetrating Radar (GPR) is an instrument that utilizes electromagnetic pulse emission, propagation, reflection, and reception to detect subsurface ground conditions97 and to apply them to remotely sensed river bathymetry. Initially developed for geological surveys, GPR has been applied in hydrometry. GPR has also been used for measuring ice-covered lakes using handheld equipment98,99, as well as for analyzing changes in river bathymetry24,97,100. Conventional GPR is moved over the ground to detect ground conditions. Some GPRs were designed to be used in the air. A recent, small GPR has been developed that can be used with Unmanned Aerial Vehicles (UAVs)101. Since alternate methods have difficulties in measuring bathymetry changes during high flow conditions102–104, the application of GPR for measuring bathymetry during floods is promising. In Japan, efforts have also been made to use GPRs to assess changes in a river’s cross-section during floods, when ALB and SfM approaches are not applicable; however, a regulation for radio wave emissions has hindered widespread adoption. High-water-conductivity influences radar signals and affects the accuracy of a channel cross‐sectional survey using ground-penetrating radar located above the water’s surface24.
Acoustic sounding is another emergent method for measuring river and lake bathymetry. Unlike conventional single sonar devices that only measure depth at a point below the sensor, Multi-Beam Echo Sounding (MBES) devices use a wide number of measurement points from one sitting. Moving the MBES over a pre-established area can create a detailed map of the river or lake-bed105. This method is particularly useful for capturing the shape of the riverbed in sandy rivers where morphological features, including the reach scale to grain scale, frequently change.
Sub-bottom profilers use acoustic sounds to detect the signal return below the channel bottom, enable to know the thickness of fluid mud layer formed in an estuary106 and alluvial sediment deposit layer in rivers107,108.
Hydroacoustic backscatter inversion techniques, initially developed for measuring suspended sediment concentration and grain size in marine environments109, have been increasingly applied in rivers using ADCPs or Acoustic Backscatter Systems (ABS). Sound beams from these systems were projected to either use side-looking110–114 or down-looking115–119 directions. While these techniques are promising for providing high-resolution measurements in both space and time, practical applications for riverine suspensions faces various practical and theoretical challenges that require further research119.
In addition to measuring suspended sediment flux, ADCPs have also been used for bedload measurements. The apparent bottom velocity can be assessed using ADCP data120. By analyzing the vertical velocity distribution measured by ADCPs, bed shear stress and depth averaged velocity can be estimated121. These two variables are widely used in bedload estimation models. Sassi et al. (2013)118 employed the van Rijn model122. Yorozuya et al. (2010)123 estimated bedload layer thickness using bed shear stress124 and evaluated the bedload rate by assuming a constant bedload concentration. Conevski et al. (2023)125 reported backscattering signals under various bedload transport conditions.
Time-Domain Reflectometry (TDR) was initially developed for identifying faults in electrical cables. TDR can measure soil dielectric properties, which are related to soil moisture content. As a result, TDR has been adapted for soil moisture measurements126. Since soil moisture content is the ratio of the volume or mass of water to soil, this technique can also be used to quantify soil density within a fluid. Specifically, TDR can measure suspended sediment concentrations within rivers127. The method is especially suited for continuous measurements of high suspended sediment concentrations. According to Chung & Lin (2011)128, the applicable concentration range spans from 2 to 300 g L
In gravel-bed rivers, researchers and engineers have used underwater sound and bed vibration to measure bed load transport129–132. These measurements often involve physical models for calculating the sediment transport rate based on mechanical principles rather than solely relying on empirical approaches133. Studies have also investigated how to separate vibrations caused by sediment transport from other factors such as turbulence and rainfall131,134. Tsubaki et al. (2017)135 attempted downsizing using a geophone device. Embedding accelerometers in individual gravel-like particles to monitor their movement in rivers has also been attempted136,137. Such recent studies have revealed that non-equilibrium sediment transport occurs during floods in gravel-bed rivers.
Crowdsourced measurements and revisiting easy-to-deploy stream gauging methodsCitizen participation in air temperature and weather measurements has a long history. Collaborative efforts connecting citizens and experts in hydrology and hydraulics are currently being explored138,139. For example, analyzing flood flows using videos captured by citizens that are stored in public repositories (e.g., YouTube) has been reported140. Beyond just taking images, citizens can be more actively involved in river measurements by using user-friendly tools such as smartphones1,20. Romano et al. (2016)141 summarized existing smartphone apps for emergency responses.
A proposal for a low-cost, easy to deploy stream gauging ruler142–144 also suggests a promising direction for enhancing citizen participation in measuring velocity and discharge in wadable streams. The rising bubble method is another low-cost stream gauging approach. Hilgersom & Luxemburg (2012)145 automatized this approach using an image processing method. Wilding et al. (2016)146 applied the rising bubble method in a weedy stream where other stream gauging methods are hard to implement. Hundt & Blasch (2019)147 compared the rising bubble method with three other low-cost methods in a laboratory flume, while King et al. (2022)148 compared three low-cost methods within small streams, reporting the uncertainty associated with each method.
Data related to rivers is critical for river research and management. Measuring water levels is relatively easy to conduct, while bathymetry measurements require more resources and incur costs, so they are typically conducted as public works. However, access to obtained data is sometimes limited for external users, including the public, as well as engineers149. While river managers have collected and revised data for cross-sectional shapes or two-dimensional bathymetry, the potential need of the public for access to updated bathymetric data is not well satisfied.
An online system for aggregating and sharing river data, including bathymetry, may be effective and important150. The U.S. Geological Survey (USGS) provides an online map, called Water Information from SPACE (WISP), which displays the water levels of rivers and lakes in the U.S. as measured by the SWOT satellite151. The River Morphology Information System (RIMOPRHPIS) was developed in the U.S. as a web-based platform for aggregating data and standardizing data formats, while incorporating data processing capabilities and promoting data usage150. A comprehensive and publicly accessible USGS streamflow measurement data set, called HYDRoSWOT, is available from a USGS National Water Information System archive of acoustic Doppler current profiler river discharge measurements collected from a wide range of rivers throughout the United States152. In Japan, efforts have been made to:
Another recent initiative by the Japanese government is the creation of a platform for river basin management, known as Plateau153.
On an international level, the Global Runoff Data Centre (GRDC) was established in 1988 to support research on global and climate change. GRDC manages a database that collects river discharge data from over 10,000 stations154,155. However, the availability and utilization of river flow velocity and topography data are not as widespread. Such limitations in data access may be partly attributed to a lack of sufficient monetary demand for such data compared to, for example, transportation networks and land-use data, both of which are useful for business. Nevertheless, with the proliferation of tools for analyzing and simulating topography and water flow, such as GIS and HEC-RAS, or iRIC software156, the demand for detailed river flow and topography data is abundant. River flow velocity data is precious for model validation but, to date, is not well shared.
Passalacqua et al. (2015)157 emphasized the importance of open data, standardized formats for easier data sharing, and open tools for data analysis. They anticipated progress in data openness related to rivers, both domestically and internationally. WaterML (Water Markup Language) is an XML-based information model designed for hydrological observation data. It aims to standardize data formats to promote better data utilization across information systems158. The Riverscapes Consortium is a collaborative framework aimed at improving and sustaining riverscape health159,160. Within this consortium, tools and data are shared based on FAIR principles161.
Data conditioningThe development of technologies such as artificial satellites, LiDAR, UAVs, and Structure from Motion (SfM) has made it possible to obtain higher density terrain data with increased frequency. However, these remote sensing methods have limitations in capturing ground elevation under buildings, in dense vegetation, in narrow valley bottoms, and in underwater terrain. These limitations pose challenges for hydrological and hydraulic analyses. Elevation data obtained from artificial satellites is useful for wide-area analysis, but is affected by inaccuracies in elevation. Obtaining global or large-scale topography without a remote sensing approach is difficult, thus, creating strong demand for correcting topography errors caused by remote sensing measurements. Efforts are being made to provide corrected elevation data by combining multiple data sources162. Estimating bathymetry using satellite-derived information with gradually varied flow equations is also promising63,163. However, the precision and resolution required for two-dimensional or three-dimensional hydraulic analyses have not been fully achieved.
The water depth distribution or cross-sectional shape can be estimated based on flow turbulence statistics calculated from the surface flow velocity distribution analyzed from video images164. This approach is expected to expedite the use of non-contact surface flow measurements for understanding processes occurring below the water’s surface, (e.g. bathymetry changes during floods, which are primarily a river engineering concern165,166. Further studies may be necessary to clarify applicable conditions for the approach167.
The accurate quantification of river surface flow turbulence characteristics requires a high-resolution water surface image that records the details of surface flow. Mid-wave infrared cameras show the high sensitivity of water surface temperatures, enabling the detection of small temperature differences caused by variations in air, water, and riverbed temperatures. This approach is particularly effective for visualizing flow velocity distributions during normal flow conditions when water surface turbulence is weak, so conventional visible light cameras cannot track flow without seedings164,166.
Non-contact flow measurements are easy to use when conducting continuous measurements, and provide a more detailed and reliable understanding of unsteady flow features. Continuous measurements may also lead to increased observations of hysteresis during the stage-discharge relationship (rating-curve). The limitations of flow discharge calculations using stage-discharge relationships have been documented168.
Modern topography measurements, such as LiDAR, ALB, UAV and MBES, provide data known as High-Resolution Topography (HRT). Exploring how to better utilize point cloud data obtained from LiDAR, ALB, and UAV-SfM for river management is an emergent topic for future research6,169. HRT data is often collected over time for major rivers in developed countries, leading to the standardization and widespread use of analysis software157,170,171. One application of HRT is Geomorphic Change Detection, in which elevation differences between multiple time periods are quantitatively analyzed. This approach can use data that consists of various sources including UAV-SfM, ground-based LiDAR172, and aerial LiDAR and paper maps173 to assess sediment transport budgets.
Other protocols for the practical application of HRT and Geomorphic Change Detection for river management consist of acquiring repeated MBES surveys over the same target area to produce a time-series of detailed bathymetry that can, subsequently, be used to estimate bedload amounts105,174–176.
The cross-sectional velocity distribution modelled using the entropy methodThe entropy method is a distinct approach and its application for reconstructing cross-sectional velocity distributions has become popular. Given such a time, this sub-section revisits the fundamental principles of the entropy method, incorporating a new figure (Fig. 3) to offer an updated interpretation of the physical meaning of negative entropy parameters. We believe this will assist researchers and engineers in utilizing the entropy method effectively and encourage further development of its applications.
In the field of streamflow measurements, entropy theory is used to identify the probability distribution closer to the real one,
Fig. 2. The velocity distribution within a river cross-section and the probability distribution of velocity.
Considering that a physical system,
therefore,
In other words, it is of note that if
with the constraints
Fig. 3. Examples of velocity distributions with different averaged velocity,
Based on the entropy defined as Eq. (3), Chiu derived a two-dimensional flow velocity distribution as a function of maximum flow velocity,
where
Therefore, the velocity distribution can be described by isovels,
Fig. 4. The curvilinear coordinate system for reproducing the velocity distribution.
Chiu (1988)182 also found a linear relationship between
with
The entropy method has been used in various applications in the field of hydraulic engineering for calculating the cross-sectional or depth-averaged flow velocity from a local velocity9,187–190, for estimating bed shear stress and sediment transport rates183 and for determining a cross-sectional shape for discharge calculations from water level data obtained by satellites91.
Based on entropy theory, a new methodology for estimating discharge beginning from the monitoring of surface flow velocity was developed by Moramarco et al. (2017)39. The method is particularly appropriate when advanced no-contact technology (as LSPIV or radar) is used for estimating discharge by monitoring surface flow velocity across a river. Bahmanpouri et al. (2024)188 reported applications of entropy theory to flow surrounding a bridge with piers in a curved bend reproduced by a 3D-CFD model, and confirmed that flow rate can be reasonably estimated in such a complicated river section. Moreover, using experimental lab tests in a meandering channel, Termini & Moramarco (2020)191 explored the possibility of evaluating the effect of secondary flow on the stream-wise velocity distribution, starting only from a knowledge of surface velocity and the entropy concept application. The research indicated that not only the shape of the velocity profile but also bed shear stress evolution along the bend could be estimated by the entropy model beginning only from knowledge of the surface velocity.
Therefore, the dip phenomenon, having a fundamental role in depth-averaged velocity assessments, is accurately identified by the entropy procedure for whatever flow conditions are present at investigated gauge sites, substantially improving the velocity index method, with
Data assimilation was initially applied in meteorological models, gained popularity, and then expanded its utilization to hydrological processes192,193. For example, the assimilation of hydraulic models with observed water level distributions based on satellites is used for estimating flow rates in river networks63. Field-measured water levels are used to assimilate two-dimensional flow fields194,195. Another example is offered by Kashiwada & Nihei (2018)196, who assimilated cross-sectional streamwise velocity distributions from measurements obtained at limited points. Data assimilation has also become popular in ground water flow analysis73. For example, Camporese et al. (2010)197 used observed pressure heads of ground water at six sites to assimilate a surface-subsurface flow model, leading to improved predictions of hydrological processes. Li et al. (2019)89 utilized GRACE (the Gravity Recovery and Climate Experiment) satellite data to assimilate ground water storage data on a global scale.
Considerations on measurement uncertaintyWhen measuring any flow property, it is important to minimize unnecessary uncertainty198. Uncertainty in a measured value is influenced by various factors such as measurement errors, limitations of reproducibility stemming from models and equations used for data analysis, and discrepancies between models/equations and actual phenomena199.
For uncertainty assessments, incorporating these influences as broadly as possible is desirable. Despax et al. (2016)200 analyzed the uncertainty involved in flow rate estimations using velocity-area methods with current-meters. An example of an uncertainty assessment in ADCP measurements based on a moving vessel is found in the work of González-Castro & Muste (2007)201. Furthermore, Despax et al. (2023)202 proposed a protocol for uncertainty in a moving vessel ADCP measurement. Estimated uncertainties were validated using empirical uncertainty assessments based on a large-scale dataset. However, since the true value is often unknown, uncertainty in each factor of assessment itself can only be treated as an uncertain quantity.
There are two major limitations in uncertainty analysis. The first is our limited understanding of phenomena related to a quantity to be assessed. The second is a limitation in the abstraction in models or equations. Kiang et al. (2018)203 discussed uncertainty in the rating curve equation. When assessing measurement uncertainty, we must be aware of these limitations.
When using a regression equation to represent the distribution of multiple observed values, there are two concepts related to uncertainty: the confidence interval, which indicates the extent to which observations deviate from the regression equation, and the prediction interval, which reflects how certain the obtained regression equation is in predicting values within the given observations. In addition to uncertainty arising from these variations, recognizing the presence of bias (systematic errors), which cannot be estimated in confidence and prediction intervals, is important198.
Le Coz (2012)204 provided a comprehensive summary of an uncertainty assessment for the rating curve. Moreover, Di Baldassarre & Montanari (2009)205 analyzed and quantified the uncertainty of river flow data. Although various factors related to rating curve uncertainty have been investigated, a comprehensive assessment framework has not yet been fully established. Muste et al. (2020)168 studied uncertainty in discharge estimations due to rating curve loops caused by hysteresis in flood wave propagation. Uncertainty analysis regarding river flow measurements using river flow visualization is also progressing206. Bayesian evaluation methods have been proposed to consider hydraulic factors, such as backwater and multiple hydraulic controls, along with the uncertainty of each measurement207, related to rating curves. Kiang et al. (2018)203 compared seven uncertainty assessment methods, including the Bayesian evaluation of Le Coz et al. (2014)208, and determined significant differences in uncertainty resulting from different assessment methods. Kiang et al. (2018)203 proposed a framework for selecting uncertainty assessment methods based on the aims of the uncertainty assessment. The challenges highlighted by Kiang et al. (2018)203 include how to effectively combine multiple uncertainty assessment methods and how to evaluate the performance of each uncertainty assessment method.
Improving hydrometry methods is challenging, particularly when assessing how measured results deviate from true values in field measurements. In laboratory settings of image velocimetry, improvement in the Particle-Image Velocimetry (PIV) and Particle-Tracking Velocimetry (PTV) methods have greatly benefited from quantitative performance evaluations using standardized image sets209,210. These evaluations employed artificially created image sets based on three-dimensional computational fluid dynamics (CFD) velocity fields to assess absolute error in the estimated velocity field. Sharing river flow videos is also beneficial for evaluating and comparing image velocimetry methods211; however, when using field data, the evaluation becomes relative since true values are often unknown199. Generating artificial river flow videos based on numerical flow simulations is important for evaluating image velocimetry algorithms for river flow measurements212.
Another approach for managing the indeterminacy of true values and the difficulty in quantifying bias in the standard uncertainty assessment, as outlined in the GUM framework213 involves conducting an empirical uncertainty assessment using repeated observations. An example of this empirical uncertainty assessment is provided by Le Coz et al. (2016)214 for the ADCP flow rate. Despax et al. (2019)215 additionally discussed sources of uncertainty based on the empirical uncertainty approach.
Advances in river measurement technologies require a timely response and collaboration amongst all stakeholders, including river management agencies, scientists, and the general public interested in rivers. To further advance river measurement and river data sharing, it is critical to share information, technological advancements, and data amongst river managers, engineers, and researchers. Creating an environment where all participant voices are genuinely heard and where discussions can take place is essential. Field engineers often possess valuable information regarding field conditions and methodologies, although their insights are not well utilized in hierarchical organizations.
Further research is needed to improve methods for estimating the depth-averaged and cross-sectional distribution of mean velocities based on surface velocity. Several factors significantly impact the correlation between surface flow and the cross-sectional velocity distribution. One major concern is how to account for the effect of wind and secondary currents. A preliminary step for addressing this concern involves collecting and sharing datasets that contain wind direction, wind speed, and the water flow velocity distribution within cross-sections under various field conditions. Although not yet fully understood, secondary flow within the cross-section impacts the relationship between surface and cross-sectional flow, and may not be negligible in certain river sections188.
The study of turbulence on the surface of river flows and an understanding of underlying water flow characteristics is an essential research topic in hydraulic engineering216. Advanced measurement methods and data analysis approaches enable researchers to easily connect obtainable information (e.g., water surface flow) with more difficult-to-obtain data (e.g., bathymetry and underwater flow during floods). The development and practical application of hydraulic methods, including data assimilation, for analyzing phenomena such as bathymetric changes during floods, also have great demand194.
Exploring strategies such as using artificial DNA fragments as environmental tracers for automating flow measurements in small and medium-sized rivers also holds value.
Further development and utilization of advanced sediment transport measurements, for example, utilizing topographical data and technologies such as geophones/hydrophones, and collecting relevant field data, would help quantify the sediment flux budget, which is critical for creating a sediment management plan on a basin scale.
Further development and improvement of uncertainty quantification methods is needed for applying such methods to various types of discharge measurements and rating curves. This is important but challenging given the complexity of measurement processes217 and rapid technological advancements in hydrometry.
While hydrological data such as water levels are relatively accessible through databases, the same level of accessibility is not always available for river topography and flow data. Further efforts to improve the aggregation, formatting, and sharing of river data, with the corresponding uncertainty, would contribute to disaster prevention and environmental management. The importance of disaster prevention and environmental management is growing due to climate change.
An invitation to provide a lecture at the summer seminar of the hydro-engineering division of JSCE motivated the first author to organize this review. The Japanese text of the lecture became the basis for the article. The first author acknowledges the support by the Foundation of River & Basin Integrated Communications, Japan, the River Fund of The River Foundation, Japan and JSPS KAKENHI Grant Numbers 23K17776 and 25K01332. The last author acknowledges the National Science Foundation, award NSF-EAR-HS 2139649, for contributing to this review paper.
