Abstract
This study presents techniques based on the use of fluorescent quinine as a visual tracer for surface flows, to assess surface flow velocities in channels and streams under low luminosity conditions. Fieldwork was conducted in three open channels, with different hydraulic characteristics. A quinine solution, in both liquid and solid (ice cube) forms, was applied on the water flow surface and an Unmanned Aerial System (UAS) was used to record the movement of the fluorescent quinine. The results were compared to the velocities estimated using the thermal tracer technique and flowmeter-based velocity maps. The findings show that the quinine solution, in both liquid and solid forms, can be used to estimate open-channel surface flow velocities under low luminosity conditions. While the solid form of the quinine tracer was applied in a smaller volume than the liquid tracer, its fluorescence effect persisted longer. By comparison, the liquid tracer had the advantage of continual availability and was devoid of the constraint of melting.
HIGHLIGHTS
Using a UAS for surface flow velocity measurements can improve data recording in hard-to-reach survey sites.
The new quinine-based tracer allows us to observe the spatiotemporal water movement in open channels and to estimate surface flow velocities under low luminosity conditions.
Quinine tracer has high visibility under UVA light in low luminosity conditions.
INTRODUCTION
Accurate measurement of surface water flow velocities and discharges is essential for both engineering purposes and environmental monitoring. Precise estimations of flow velocities are particularly important for understanding and modelling runoff.
Tracers are commonly used in hydrological studies to estimate surface flow velocities. Various types of tracers are used, including non-fluorescent tracers such as dye (e.g., de Lima & Abrantes 2014; Abrantes et al. 2018), fluorescent dye tracers (e.g., Leibundgut et al. 2009; Zhang et al. 2010), buoyant particles (e.g., Tauro et al. 2012; Mujtaba & de Lima 2018), salt tracers and electrolytes (e.g., Lei et al. 2010; Schuetz et al. 2012; Abrantes et al. 2018), and thermal tracers (e.g., Schuetz et al. 2012; de Lima & Abrantes 2014; de Lima et al. 2015; Abrantes et al. 2018; Mujtaba & de Lima 2018; Abrantes et al. 2019).
Fluorescent dye tracers are common in hydrology for identifying connections between groundwater supply points (e.g., sinkholes and karst windows), discharge points (e.g., springs and wells), and estimating sheet flow velocities (e.g., Gilley et al. 1990; Buzády et al. 2006; Leibundgut et al. 2009; Zhang et al. 2010; de Lima et al. 2021; Zehsaz et al. 2022). Several environmental non-hazardous dyes such as uranine, rhodamine WT, eosin, CI Direct Yellow 96, as well as other optical brighteners, are widely accessible and used. The environmental impact of fluorescent tracers is considered tolerable since the quantity of fluorescent material used in hydrological studies is insignificant (e.g., Aley & Fletcher 1976; Buzády et al. 2006; Leibundgut et al. 2009).
In recent decades, remote sensing techniques using Unmanned Aerial Systems (UASs) have emerged as a valid and cost-effective alternative to in situ environmental measurements, including river discharge, streamflow velocities (e.g., Tauro et al. 2015; Fulton et al. 2020; Masafu et al. 2022), and water quality parameters (e.g., Su 2017). Some studies provide an overview of the contribution of UASs regarding natural and agricultural ecosystems monitoring (e.g., Manfreda et al. 2018; Tmušić et al. 2020). UAS, equipped with a gimbal-mounted camera, are now relatively inexpensive and able to record high-resolution images and videos.
de Lima et al. (2021) recently presented a proof of concept for using quinine solution as a fluorescent tracer for shallow overland flows over bare soil conditions, while Zehsaz et al. (2022) explored its potential for estimating sheet flow velocities over mulched, vegetated, and paved surfaces. The results from this technique were found to be positively correlated with the results of the dye and thermal tracer techniques in both studies. The results indicate that using quinine solution as a fluorescent tracer offers several advantages, such as high visibility under ultraviolet A (UVA) light and under low luminosity conditions, low cost, and neglectable environmental impact. However, the quinine tracer requires UVA artificial illumination during the experimental procedure, creating an inherent limitation to the fluorescent tracer technique.
This experimental study builds on the work of de Lima et al. (2021) and Zehsaz et al. (2022) by further exploring the capability of quinine-based tracers. Both liquid and solid forms of quinine solution were used as fluorescent tracers to estimate surface velocities in flowing surface water bodies, such as rivers, irrigation channels, and drainage channels. This study explores several application methods to test the tracer's effectiveness in estimating open-channel flow surface velocities under low luminosity field conditions using a UAS. This study's primary objectives are: (i) to evaluate the tracer's ability to estimate flow surface velocities in low luminosity conditions, (ii) to investigate different tracer application methods, and (iii) to compare results obtained using two quinine-based tracers: liquid and solid (ice cubes).
MATERIALS AND METHODS
Location of field study sites and experimental setup
Location . | Channel structure . | Channel shape . | Discharge (m3/s) . | Water surface width (m) . | Max water depth (m) . | Temperature (°C) . | |
---|---|---|---|---|---|---|---|
Water . | Ambient . | ||||||
Irrigation channel | Concrete | Trapezoidal | 4.25 | 7.0 | 2.0 | 19 | 14 |
Drainage channel – site 1 | Earth channel (soil) | U-shape | 0.59 | 3.9 | 0.7 | 20 | 19 |
Drainage channel – site 2 | Earth channel (soil) with a rocky bed | U-shape | 0.52 | 3.5 | 0.5 | 20 | 19 |
Location . | Channel structure . | Channel shape . | Discharge (m3/s) . | Water surface width (m) . | Max water depth (m) . | Temperature (°C) . | |
---|---|---|---|---|---|---|---|
Water . | Ambient . | ||||||
Irrigation channel | Concrete | Trapezoidal | 4.25 | 7.0 | 2.0 | 19 | 14 |
Drainage channel – site 1 | Earth channel (soil) | U-shape | 0.59 | 3.9 | 0.7 | 20 | 19 |
Drainage channel – site 2 | Earth channel (soil) with a rocky bed | U-shape | 0.52 | 3.5 | 0.5 | 20 | 19 |
To expose the fluorescent tracer, two UVA lamps with a Blacklight Blue (BLB) light bulb were used, with one lamp on each side of the measuring frame, aligned parallel to the flow direction (see Figure 1(b)). These lamps are characterised by a nominal power of 36 W, UVA irradiance at 20 cm, 315–400 nm: 350 mW/cm2; and peak emission wavelength of 0.354 μm. The two lamps, pointing to the water surface, were installed close to the centre of the channels, separated by 1.2 m (each lamp 0.6 m from the centre of the channel), and suspended from high-resistance strings attached to metal bars on channel margins. A measuring frame (i.e., observation area) with 1.0 m × 1.0 m was located between the lamps. A UAS was used to monitor (video record) the tracer application in the selected locations of the channels.
Tracers
Tracer application methods
The quinine-based tracers were tested under similar conditions (e.g., same straight channel section; approximately the same discharge, flow depths and velocities). Several methods for applying the liquid and solid quinine tracers into the flow surface were tested during the field experiments (Table 2), namely, point and linear applications. The liquid tracer volumes used with the point and linear application methods were 250 and 450 mL, respectively. The quantity of the liquid tracer solution used was the minimum that allowed tracking the tracer during the experiments. In the solid tracer release door container method, the ice cubes were released from approximately 0.20 m above the water surface. In the solid spill container, the ice cubes were spilt on the water surface at one point. For all cases, the application of the tracers was conducted carefully to minimally disrupt the flow.
Flow measurement
An electromagnetic flowmeter (Valeport Model 801) was used to measure flow velocities in the studied drainage open channels, in the selected cross-sections (sites 1 and 2) and draw the surface velocity fields. Due to access difficulties, the flowmeter was not used in the irrigation channel. The technical specifications of the flowmeter are as follows: measuring range of −5 to +5 m/s; accuracy of ±0.5%; sensing volume: cylinder of approximately 20 mm diameter × 10 mm height and minimum measuring water depth of 5 cm.
UAS and video recording systems
The real imaging video and photographs were recorded using a camera with a gimbal installed on a DJI Phantom 4 Pro quadcopter UAS controlled manually with the DJI GO 4 mobile application. The equipment specifications are presented in Table 3.
Aircraft . | Gimbal . | Camera . | |||
---|---|---|---|---|---|
Weight (battery and propellers included) | 1,375 g | Stabilisation | Three-axis | Sensor | 1-inch CMOS Effective pixels: 20M |
Diagonal size (propellers excluded) | 350 mm | Controllable range | −90° to +30° | Lens | FOV 84° 8.8 mm/24 mm (35 mm format equivalent) f/2.8 |
Max. speed | S-mode: 20 m/s A-mode: 16 m/s P-mode: 14 m/s | Max. controllable angular speed | 90°/s | Image size | 3:2 Aspect Ratio: 5,472 × 3,648 pixels |
Max. wind speed resistance | 10 m/s | Video recording mode | C4K: 4,096 × 2,160 pixels Max video bitrate: 100 Mbps | ||
Max. flight time | Approx. 30 min |
Aircraft . | Gimbal . | Camera . | |||
---|---|---|---|---|---|
Weight (battery and propellers included) | 1,375 g | Stabilisation | Three-axis | Sensor | 1-inch CMOS Effective pixels: 20M |
Diagonal size (propellers excluded) | 350 mm | Controllable range | −90° to +30° | Lens | FOV 84° 8.8 mm/24 mm (35 mm format equivalent) f/2.8 |
Max. speed | S-mode: 20 m/s A-mode: 16 m/s P-mode: 14 m/s | Max. controllable angular speed | 90°/s | Image size | 3:2 Aspect Ratio: 5,472 × 3,648 pixels |
Max. wind speed resistance | 10 m/s | Video recording mode | C4K: 4,096 × 2,160 pixels Max video bitrate: 100 Mbps | ||
Max. flight time | Approx. 30 min |
The thermal videos were recorded with the FLIR DUO PRO R infrared camera (resolution: 336 × 256 pixels; accuracy: ±5 °C or 5% of readings in the −25 to +135 °C range; spectral range: 7.5–13.5 μm). The cameras were installed with the sensor parallel to the water surface.
To gather the data needed for this study, the UAS was used to record real videos of quinine tracer movement in the drainage channel and the FLIR DUO R infrared camera was installed on a bridge over the irrigation channel to capture thermal and real video images simultaneously.
Field experimental procedure
At the selected cross-sections of the drainage channel, the water surface width was measured, as well as the depth of the water at regular spatial intervals (every 0.40 m) between the banks. The electromagnetic flowmeter was used to measure the flow velocities at several points in the cross-section; 23 and 25 measurements were taken, respectively, for drainage channel – sites 1 and 2. Cross-section velocity maps were created based on these measurements; for the wall and bed boundary layer of the channels, the flow velocity was assumed 0 m/s. A physical interpolation method, Thin Plate Spline (TPS) algorithm, was applied to the data to create the flow velocity maps.
The discharge in the drainage channels was estimated by spatially integrating the measured flow velocities. Tracers were applied carefully on the water surface, using an extension wooden bar, 0.2 m upstream of the measuring frame to minimise any interference in the stream. The movement of the tracers along the measuring frame was recorded with the FLIR DUO R infrared camera for the irrigation channel and with the real video camera installed on the UAS for the drainage channel. During the field experiments, three replicates of video recordings were conducted for each test (form and method of tracer application). While recording the fluorescent tracers in the drainage channels, the UAS was flying in P-mode, (e.g., maintaining a fixed GPS-controlled position) above the water surface with the camera's sensor parallel and 2.6 m above the water surface. The infrared camera was fixed to a tripod on the top of a bridge over the irrigation channel, recording real and thermal videos, with the camera's lenses 4 m above, and parallel to, the water surface. The water temperature in the selected cross-sections and the ambient temperature were measured (Table 1). The infrared camera provided data on the water temperature, while a cell phone application was used to collect information on the ambient temperature. The experiments were conducted without interference from the wind on the water surface.
Velocity estimation and image processing method
The surface velocity Vs was estimated by calculating the travel distance of the tracers' plume leading-edge (leading vertex) or leading-front (averaged discrete points, in the linear applications) over time Δt, within the measuring frame (de Lima et al. 2021; Zehsaz et al. 2022). When using solid tracers, the travel distance of each ice cube over time Δt was calculated separately, according to the procedure described in de Lima et al. (2023). The surface velocity Vs was considered as the average of the velocities estimated for a group of ice cubes, in both the point and linear application methods.
Methodology overview
RESULTS AND DISCUSSION
When tracing the leading-edge diffusion limits the time frame to perform the measurements, as well as introducing difficulties in the measurement. However, the movement of marked points on the tracer's leading front line was accurately monitored, as depicted in Figure 7(b)–7(d). Point and linear applications presented distinct difficulties with the former detection being considerably more straightforward than the latter.
Unlike the liquid form of tracers, the ice cubes kept their shape and fluorescent concentration during the experiments, facilitating the monitoring and measuring of the travel distance of the cubes over time within the measuring frame (Figure 7(e)–7(g). The differences between the spill and release door container methods are shown in Figure 7(e) and 7(f). When using the spill container method, the ice cubes moved closer to each other, resulting in a closer estimate of surface flow velocities at the point where the tracer was applied.
The flow surface velocities (Vs) obtained by applying a quinine tracer, and the procedure described in ‘Field experimental procedure’ section for all three experimental sites, are summarised in Table 4, where the mean and standard deviation (S.D.) for the three experimental replicates are also presented.
Location . | Discharge, Q (m3/s) . | Froude number . | Form of tracer . | Application method . | Surface velocity, Vs (m/s) . | |
---|---|---|---|---|---|---|
Mean (three replicates) . | S.D. . | |||||
Drainage earth channel – site 1 | 0.59 | 0.17 | Solid | Point-spill | 0.42 | 0.017 |
Point-release | 0.39 | 0.020 | ||||
Linear | 0.38 | 0.007 | ||||
Mean (three methods) | 0.40 | – | ||||
S.D. | 0.012 | – | ||||
Liquid | Point | 0.42 | 0.017 | |||
Linear-container series | 0.42 | 0.022 | ||||
Linear-box container | 0.41 | 0.019 | ||||
Linear-moving can | 0.39 | 0.007 | ||||
Mean (four methods) | 0.41 | – | ||||
S.D. | 0.012 | – | ||||
Drainage earth channel – site 2 | 0.52 | 0.37 | Solid | Point-spill | 0.71 | 0.015 |
Point-release | 0.71 | 0.019 | ||||
Linear | 0.74 | 0.004 | ||||
Mean (three methods) | 0.72 | – | ||||
S.D. | 0.012 | – | ||||
Liquid | Point | 0.75 | 0.01 | |||
Linear-container series | 0.78 | 0.023 | ||||
Mean (two methods) | 0.76 | – | ||||
S.D. | 0.016 | – | ||||
Irrigation concrete channel | 4.25 | 0.11 | Thermal | Point-spill | 0.5 | 0.013 |
Linear | 0.48 | 0.018 | ||||
Mean (two methods) | 0.49 | – | ||||
S.D. | 0.01 | – | ||||
Quinine-solid | Point-spill | 0.52 | 0.003 | |||
Linear | 0.48 | 0.029 | ||||
Mean (two methods) | 0.5 | – | ||||
S.D. | 0.02 | – |
Location . | Discharge, Q (m3/s) . | Froude number . | Form of tracer . | Application method . | Surface velocity, Vs (m/s) . | |
---|---|---|---|---|---|---|
Mean (three replicates) . | S.D. . | |||||
Drainage earth channel – site 1 | 0.59 | 0.17 | Solid | Point-spill | 0.42 | 0.017 |
Point-release | 0.39 | 0.020 | ||||
Linear | 0.38 | 0.007 | ||||
Mean (three methods) | 0.40 | – | ||||
S.D. | 0.012 | – | ||||
Liquid | Point | 0.42 | 0.017 | |||
Linear-container series | 0.42 | 0.022 | ||||
Linear-box container | 0.41 | 0.019 | ||||
Linear-moving can | 0.39 | 0.007 | ||||
Mean (four methods) | 0.41 | – | ||||
S.D. | 0.012 | – | ||||
Drainage earth channel – site 2 | 0.52 | 0.37 | Solid | Point-spill | 0.71 | 0.015 |
Point-release | 0.71 | 0.019 | ||||
Linear | 0.74 | 0.004 | ||||
Mean (three methods) | 0.72 | – | ||||
S.D. | 0.012 | – | ||||
Liquid | Point | 0.75 | 0.01 | |||
Linear-container series | 0.78 | 0.023 | ||||
Mean (two methods) | 0.76 | – | ||||
S.D. | 0.016 | – | ||||
Irrigation concrete channel | 4.25 | 0.11 | Thermal | Point-spill | 0.5 | 0.013 |
Linear | 0.48 | 0.018 | ||||
Mean (two methods) | 0.49 | – | ||||
S.D. | 0.01 | – | ||||
Quinine-solid | Point-spill | 0.52 | 0.003 | |||
Linear | 0.48 | 0.029 | ||||
Mean (two methods) | 0.5 | – | ||||
S.D. | 0.02 | – |
Mean velocity and standard deviation (S.D.) for three replicates. The discharge was estimated by applying the area-velocity method, based on flowmeter measurements.
The FLIR DUO R infrared camera's ability to record dual images (thermal and real video images simultaneously) facilitated the comparison of two tracer techniques under the same conditions (e.g., form of tracer, application method of tracer, and volume of tracer). Specifically, thermal imaging was exclusively employed in the irrigation channel, and the results for each tracer application method (point and linear) were obtained from the same set of experiments (form and volume of tracer), using both thermal and fluorescent tracing techniques.
On average, the flowmeter measurement velocities resulted between 3.5 and 8.3% higher than the estimations using the quinine tracer in drainage channel – site 1 and solid form of quinine tracer in drainage channel – site 2, respectively. However, the flowmeter yielded 2.6% lower velocities than the liquid quinine tracer in the drainage channel – site 2. This discrepancy is likely caused by experimental and instrumental measurement errors.
CONCLUSIONS
In this study, a UAS was used to assess the movement of fluorescent quinine tracers in different forms and application methods and their ability to estimate open-channel surface flow velocities under low luminosity conditions. The conclusions driven by the fieldwork in different sites are:
Quinine solution tracer exposed to UVA light can be used to estimate open-channel surface flow velocities at night or under low luminosity conditions.
The fluorescent-based approach for estimating surface flow velocities is a simple and low-cost technique. Essentially, all that is required are UVA lamps and a regular camera to make the observations. However, the installation of the setup, such as fixing the lamps or camera, may require considerable effort, depending on the fieldwork and local conditions. The use of a UAS can facilitate some of the setup installations, especially in cases where accessing the site is difficult. Installing cameras on a UAS does not require carrying or fixing any heavy supports, such as tripods.
The advantage of using the solid (ice cubes) form of the quinine tracer over the liquid form was that, although the volume of the solid tracer used in the experiment was smaller than the liquid tracer, the tracer in solid form was easier to track due to diffusion of the liquid tracer in the channel flow. The liquid tracer plume is untraceable in a few seconds after applying the tracer into the flow, whereas the ice cubes take time to melt, thus maintaining the same concentration of tracer during the measuring time (the time lapse between the instants when the leading edge of the tracer enters and exits the measuring frame). However, in some conditions (e.g., high ambient temperatures), it might not be easy to have access to or maintain the ice cubes solid in field conditions.
When using the liquid tracer, the leading edge of the tracer plume is difficult to distinguish after a short while of the tracer application. On the other hand, the linear application of the liquid tracer led to some difficulties when analysing the images for estimating the surface velocities because it is challenging to track the marked points on the leading front line of the tracer plume separately, within the measuring frame and time.
Based on the findings of the study and experimental conditions, the authors recommend using the solid form of the quinine solution tracer as it is easier to track and keep a constant concentration of the tracer throughout the measuring time. For measuring flow surface velocities at a specific point, the point application method is recommended, while the linear application method is suggested for obtaining surface velocities across the channel or to estimate the average velocity within the channel width.
AUTHOR CONTRIBUTIONS
S.Z., J.L.M.P.d.L., and J.M.G.P.I. conceived and designed the experiments; S.Z. performed the experiments; S.Z., J.L.M.P.d.L., and M.I.P.d.L. analysed the data; S.Z. and J.L.M.P.d.L. wrote the draft paper; J.L.M.P.d.L., J.M.G.P.I., M.I.P.d.L., and R.M. revised the manuscript. All authors have read and agreed to the published version of the manuscript.
FUNDING
This research was partly funded by the Portuguese Foundation for Science and Technology (FCT), through projects MEDWATERICE (PRIMA/0006/2018), supported by national funds (PIDDAC), projects UIDB/04292/2020 and UIDP/04292/2020 granted to MARE – Marine and Environmental Research Center, University of Coimbra (Portugal), strategic project UIDB/04450/2020 granted to RISCO – Research Centre for Risks and Sustainability in Construction, and project LA/P/0069/2020 granted to the Associate Laboratory ARNET – Aquatic Research Network, supported by national funds. The author S.Z. was granted a PhD fellowship from FCT (Ref. 2020.07183.BD).
ACKNOWLEDGEMENTS
The authors express their gratitude to Bruno Matos, Jean Tavares at the Instituto Federal de Educação, Ciência e Tecnologia do Rio Grande do Norte (IFRN) and Rui Pedroso de Lima at the Innovative Dynamic Monitoring for Water Quality and Ecology (INDYMO) for their help in conducting the fieldwork. Additionally, the authors would like to acknowledge Hydrology Research, the Euromediterranean Network of Experimental and Representative Basins (ERB) – 18th Biennial Conference 2022 and Daniele Penna, the current coordinator of the ERB, for the sponsorship in publishing this research.
DATA AVAILABILITY STATEMENT
All relevant data are included in the paper or its Supplementary Information.
CONFLICT OF INTEREST
The authors declare there is no conflict.