Brightness of point application of fluorescent quinine tracer for surface waters

Abstract Fluorescent tracers have been widely used in hydrology. Recently, quinine started to be used as a fluorescent tracer for estimating the velocity of surface sheet flows over various soil surface conditions and environments. In the present work, the visibility of the fluorescent tracer (quinine) was assessed for various applications’ forms of the tracer (liquid, ice cube with quinine and soaked sponge). The brightness intensity of all tracer forms was estimated for different hydraulic conditions (hydrostatic, linear, and rotational flows) and for clear water, and water with medium and high suspended sediment loads. Results show that, when used as a flow velocity tracer, liquid quinine solution has to be applied carefully into the water and should better be used on sheet flows, shallow overland flows or shallow still waters. Its visibility in deep and muddy flows is insufficient for surface velocity estimations. The sponge soaked with quinine solution, which partially floats, is better visible in clear waters or low-medium suspended sediment loads, regardless of the water depth. However, for high turbulence and rotational flows, the soaked sponge sinks and is no longer visible. The ice cubes showed better visibility in all tested flow water depths and suspended sediment loads, although, in very shallow depths (of millimetres), ice cubes cannot be used because they might not follow adequately the fluid motion, which also applies to the sponge.


INTRODUCTION
The study of flow velocities is fundamental for a variety of hydrological and engineering applications.There is often the need to assess the velocity in situations where there are linear and rotational flows, which are common in, e.g., rivers, channels, and effluent treatment systems (VON SPERLING, 1997) or when there is a combination of different surface velocity vectors, a common situation in large reservoirs and also in the oceans (KOSHEL et al., 2019).
The most common velocity estimations are associated with surface flows in rivers and channels.In recent years, many studies have been presented regarding various applications of methods for estimating velocity distribution in the water column for different hydraulic conditions (NIKORA; NIKORA; O 'DONOGHUE, 2013;KUMBHAKAR et al., 2021;ZHANG et al., 2023).Moreover, some studies have scrutinized the velocity distribution profiles on specific sections (STEPHAN; GUTKNECHT, 2002).
Hydraulic tracer studies assume that the chosen tracer is inert and thus should represent the water flow.Salt tracer solutions are common, but normally have a higher density than fresh water; therefore, they may lead to tracer settling at the bottom.This happens, for example, in wetlands (WANG; JAWITZ, 2006;SPEER et al., 2009;WAHL et al., 2010).
Flow visualisation using tracers is a useful tool in the study of fluid dynamics as it provides a quick qualitative and quantitative description of the flow field, both of steady and unsteady regimes (GRAHAM, 1984;SMITS;LIM, 2012).
Also, it has the advantage to have reduced interference with the flow.Other frequently used techniques, such as invasive velocity probes, are only able to provide information at discrete points within the flow field (BOITEN, 2008).
Visualisation techniques might use different types of tracers (e.g., particles, fluorescent or coloured dye, and thermal tracers) added to the flow to signal the fluid motion, which can be defined using optical or thermographic techniques, such as LASER systems, high-speed cameras, and infrared devices.
Whereas common dyes, which are less expensive and normally safer to use, such as food colouring, are visible under naked eye (given that there is sufficient ambient light) when they are applied into the water, fluorescent dyes depend on ultraviolet (UV) or LASER light to help clearly expose the tracer from the fluid (i.e., the dye fluoresces/glows under those lights).This property allows one to use fluorescent dyes under specific ambient conditions, in particular, low ambient light conditions.Of the commonly used fluorescent dyes, fluorescein has been used since the end of the 19 th century due to being detectable in low concentrations, although it has very poor stability under sunlight (DOLE, 1906;SMART;LAIDLAW, 1977).
Recently, de Lima et al. (2021) presented a proof of concept for using quinine as a fluorescent tracer for estimating the velocity of surface sheet flows over various soil surface conditions and environments.This technique, where quinine is injected in liquid form into the flow, was compared with other similar tracer techniques that use dye and thermal tracers, also in liquid form.It was reported that the quinine fluorescent tracer offered flow velocity estimates that were comparable to the other tracers' estimates, and that the main advantages of using the quinine tracer were the high visibility of this tracer under ultraviolet A (UVA) light for low luminosity conditions, small environmental impact, and low-cost.Zehsaz et al. (2022) further explored the functionality of quinine as a tracer to estimate runoff velocities over mulched, vegetated, and paved surfaces.The results from that study showed that the quinine fluorescent tracer can be used in different rural and urban areas to estimate sheet flow velocities.
This study further explores the applicability and performance of the quinine tracer by systematically testing its brightness intensity for various hydraulic conditions: (i) suspended sediment loads (turbidity); (ii) velocities; and (iii) water depths, using laboratory reservoirs and flumes, under controlled conditions.Pixel classification techniques were used to evaluate the tracer movement in time-lapse imagery to capture the dynamics of the fluorescent tracers.

Laboratory setups and materials
Different laboratory setups and conditions were used in this experimental work  (FERREIRA et al., 2006;ABRANTES et al., 2018;ISIDORO et al., 2021), at the University of Coimbra, Portugal.
Because the quinine fluorescent tracer is only visible in darker ambient light conditions under UVA light, an ambient light-shielded environment was provided for conducting the experiments.The quinine solution was obtained by mixing water with quinine powder at a concentration of 80 mg•L -1 (for higher brightness, see Zehsaz et al., 2022).The tracer was applied in the following forms To generate a suspension of fine materials in the water, a non-organic soil was used.This soil (used in, e.g., Oliveira, Correia and Cajada, 2018) has a high content of clay/silt particles (64%), shows low plasticity (dry unit weight: W = 9.9 kN•m -3 ; plasticity index: PI = 2%; liquid limit: wL = 44.0%;plastic limit: wP = 42.0%) and is classified as a low plasticity silty soil (ML), according to ASTM (1998).Sediment load suspensions of 2 g•L -1 and 5 g•L -1 were used during the experiments (for both hydrostatic and non-hydrostatic conditions), which were considered medium and high suspended loads, respectively, for the purpose of these experiments (in the high suspended load, visibility under water was very limited).
During the experiments, a camera with a 12.80 mm × 9.60 mm ISOCELL 2L2 (S5K2L2) CMOS sensor was used to capture images with a resolution of 4290 × 2800 pixels, and 1920 × 1080 pixels for the video frame.The camera was fixed in a position parallel to the water surface and at a height of 1.20 m from the water surface (Figure 1).These images were analysed to estimate the brightness of the quinine tracer, as explained next.

Laboratory procedure
The assessment of the brightness intensity of all quinine tracer forms (liquid, ice, and soaked sponge -Figure 2) was carried out for all hydraulic conditions (hydrostatic, linear/1D flow and rotational movement -Figure 1) and for clear water and water with medium and high suspended loads, for the purpose of comparison.

Brightness estimation of the quinine tracer
Every image consists of a set of pixels, which constitutes the raw building cells of the image.The pixels are defined in two categories: grayscale/single channel and colour.The pixels in an RGB colour space (colour image) have three colour channels: red, green, and blue.However, a grayscale image has just one channel and it usually uses an 8-bit representation for each pixel.A RGB image can be converted to a grayscale one by taking the weighted average of the red, Source: elaborated by the authors.Where R, G, and B are the red, green, and blue colour values of a pixel, respectively.
In this case, each pixel corresponds to a single scalar value ranging from 0 to 255.This number represents thus the brightness (lightness) intensity value of each pixel.Zero corresponds to "black" and 255 corresponds to "white" (STONE, 2003).
Analysis of the recorded image frames was conducted using the MATLAB image processing toolbox.All snapshots of the videos and images were converted to grayscale; therefore, the brightness intensity range for the fluorescent tracer varied from 0 to 255.The brightness intensity of the pixels that correspond to the tracer plume/stain in each image was extracted.

Hydrostatic measurements
For evaluating the brightness of the quinine tracer, different water depths (0.01, 0.10, and 0.65 m) and suspended sediment loads (clear water and medium and high loads) scenarios were tested for hydrostatic conditions (Figure 1a).Three replicates of each of the experiments, which correspond to three independent photographs for each scenario, were undertaken.

Linear flow measurements
A hydraulic flume was used for linear (1D) flow conditions (closed circuit -Figures 1b and 3).Experiments were performed for clear water and water with medium and high suspended sediment loads (2 and 5 g•L -1 , respectively), like the no-flow (i.e., hydrostatic) experimental conditions.
For this hydraulic setup, the surface flow was fed at the upstream end of the flume through run-on from the feeder tank, which was controlled manually from the water pumping system using a valve until steady state flow was achieved.Once discharge became stable, the velocity measurements were undertaken with the installed video camera using the methodology described in de Lima et al. (2021).The movement of the tracer along the scanned area (Figure 1b) was recorded in separated videos, for each tracer form and settings.
The area scanned by the camera was established starting at 0.50 m downslope of the upstream end of the flume, with a dimension of 0.15 × 0.50 m 2 over the flume.Three replicates of the experiment were conducted for each flow discharge rate, which varied between 0.035 L•s -1 and 1.530 L•s -1 .

Rotational flow measurements
The rotational flow measurements' setup is presented in Figures 1c and 4. The water movement was induced by rotating a cylindrical reservoir.The rotational velocities tested in this experiment were 5, 10, and 20 rpm, simulating vorticity in natural systems.Similarly to the other hydraulic conditions tested, all experiments were performed for clear water and water with medium and high suspended loads.

Still water conditions
The difference between the visibility of the fluorescent (quinine) liquid solution, ice, and soaked sponge is illustrated in Figure 5 for the hydrostatic condition (i.e., horizontal water surface; no flow; no wind).The quinine liquid solution quantity used was the same, and the size of the tracers in ice and sponge forms were initially the same in all experiments; however, due to the melting of the ice, some snapshots might reveal different sizes for these two tracer forms.Brightness of fluorescent quinine tracer and soaked sponge form of application of the quinine tracer.The relatively low brightness intensity of the liquid quinine stain is due to dispersion and the fact the tracer sinks in the water column; many times, it attained values that were 50% less than the brightness of ice or sponge quinine tracers.In some cases, the tracer in liquid form could not even be detected.
For the experiments carried out for still water (hydrostatic) conditions, the brightness intensity found along transects defined for each of the plumes of the three tracer forms (liquid, ice, and sponge) was plotted in Figure 8.These transects contain the brightest pixel of each of the tracers' plumes; for the ice and sponge tracers, the transects are aligned with the longest dimension of their rectangular shape and, for the liquid tracer, imaginary circles were superimposed to the plumes and the transects contained the circles' centre and the brightest pixel in each image.Figure 8 shows that the liquid form of the tracer application had the biggest and the ice had the smallest plume area, visible from the top-view images of the water surface.The smallest size of the ice in comparison to the sponge at the time of measurements is due to the melting of the ice (the ice and the sponge were prepared for the experiments having the same size, see Figure 2).Nevertheless, although the ice plume had the smallest brightness areal coverage, it had a much higher average brightness intensity in comparison to the other two quinine tracer forms.In addition, although in the present experiments the average brightness of the sponge's plume is higher than that of the liquid tracer, the brightness is not homogeneous over the whole sponge surface due to the different pore sizes in the sponge structure.Figure 8 shows the mean values of three replicates of the experiments.

Linear flows
The    Source: elaborated by the authors.Note: The zero of the x-axis contains the brightest pixel of each of the tracer plumes, with this axis supporting the representation of the variation of the tracer brightness intensity across transects identified for each of the plumes of the various tracer forms (liquid, ice, and sponge).Regardless of the actual positioning of the tracer plume in the image, for the ice blocks and sponges, the transects were defined along the longest dimension of their rectangular shape, containing the brightest pixel; for the liquid tracer, imaginary circles were superimposed to the plumes, and the transects contained the circles' centre and the brightest pixel in each image.The graph shows the mean values of three replicates.the soaked sponge brightness was lower than the brightness of the ice and liquid tracer forms, whereas 1 s after the tracers' application time the quinine solution's liquid plume manifested already a much smaller brightness than the sponge.The ice cubes have the tendency to melt over time, which might significantly reduce their visibility, depending on the original size of the ice cubes, the water temperature of the recipient, and the duration of the experimental tests.
In flowing waters with a high suspended sediment load, the quinine liquid solution plume could not be detected (Figure 11).Sponges soaked with quinine solution were observed to work well at the beginning of the tests but, after some time, as the sponge gets more soaked with the flowing water, it sinks in the water column and can no longer be detected.Also, for the type of sponge used, it was observed that, after using the same sponge for several times, it did not perform well and sunk rapidly.The sponges should be changed after one or two experimental trials.
Table 1 summarizes the data for surface flow velocity estimations using ice and sponge (average and standard deviation).
The analyses of results presented in Table 1 show that the applied tracer velocities, estimated by both techniques (ice and sponge), are similar under different suspended loads and similar discharges.The liquid form of the tracer had poor results in presence of sediments loads.It was not possible to visualise or track the leading edge of the liquid tracer in such turbid conditions.
Therefore, no comparisons of velocity estimation were made between liquid and other tracer techniques.

Rotational flows
The comparison of the brightness intensity of the tracer in the form of ice cubes for clear water and medium and high suspended sediment loads is shown in Figure 12 for a rotational movement with a velocity of 20 rpm. Figure 12 shows that the suspended sediment load had no visible effect on the brightness of the quinine ice cubes, due to buoyancy.Several rotational velocities (5, 10, 20 rpm) were tested.The tracer in liquid and soaked sponge forms were also tested but these tracers did not deliver good visual results, since both were pulled down into the vortex.Only the ice cubes remained visible for all tested rotations.Source: elaborated by the authors.Note: Two superimposed images are represented in each panel, for a time lapse Δt = 1 s.In the graphs, the water movement is from left to right.Flow velocity was approximately 0.02 m•s -1 .In the horizontal axes, the "distance" is expressed in pixels.Vertical scales are not the same in the three graphs, but, overall, the colour scale used is the same.de Lima, J.L.M.P. et al

CONCLUSION
Certain studies have aimed at forecasting velocity profiles in vegetated open channel flows (NIKORA; NIKORA; O'DONOGHUE, 2013), while others have investigated the whole velocity profile, encompassing both vegetation canopy and overlying flooded flow regions (KATUL; POGGI; RIDOLFI, 2011).

(
Figure 1): (a) hydrostatic conditions in acrylic reservoirs (parallelogram of base 0.40 × 0.65 m 2 for low water depths and 0.85 × 0.85 m 2 for higher water depths); (b) linear (1D) uniform flow in a laboratory flume sloping surface (0.15 m wide and 4.00 m long); (c) rotational movement on an acrylic rotating cylindrical reservoir (diameter of 0.25 m).The main purpose of the experiments was to appraise the brightness of the quinine fluorescent tracer when applied into the water in different forms and for different water turbidity and hydraulic conditions.The experimental laboratory tests were conducted in the Laboratory of Hydraulics, Water Resources and Environment of the Department of Civil Engineering, of the Faculty of Science and Technology of the University of Coimbra (Portugal).The flume identified in Figure 1b has been used in other laboratory studies into the flow: (1) liquid solution injected with a syringe; (2) quinine solution ice blocks (hereinafter called cubes) placed at the water surface; and (3) sponge soaked with quinine solution also placed at the water surface.The liquid tracer volume was 10 mL (commonly used volume injected with a syringe, for similar hydraulic conditions and class of experiments) and the size of the ice and sponge was approximately 0.040 × 0.035 × 0.010 m 3 in all experiments.The tracers were added carefully into and on the water surface to have a minimum disturbance Brightness of fluorescent quinine tracer of the hydraulic conditions.Figure 2 shows the different materials and forms of the quinine tracer applications in the present set of experiments.

Figure 1 -
Figure 1 -Illustration of the laboratory setups for the three hydraulic conditions: a) hydrostatic; b) linear flow; c) rotational movement.

Figure 2 -
Figure 2 -Different materials and forms used for the quinine tracer application.
J.L.M.P. et al green, and blue components of each pixel in the image.The equation to convert RGB to grayscale is: I = (0.2989 × R) + (0.5870 × G) + (0.1140 × B)(1)

Figure 5
Figure5shows that the visibility of the liquid form of application of the tracer diminishes as the height of the water column increases, since the liquid applied into the water sinks in the water column, moving away from the surface.Higher suspended sediment loads (i.e., muddy waters) further add difficulty to the visualization of the tracers.In this experiment, the tracer is only still visible from the surface for a water column depth of 0.65 m (the deepest water column tested) for clear water.The brightness intensity (i.e., visibility) of the three forms of quinine tracer was plotted for the hydrostatic condition, and different water column depths and suspended sediment load (Figures6 and 7).These figures show that the tracer liquid plume had the lowest brightness intensity in comparison to the ice

Figure 3 -
Figure 3 -Illustration of the fluorescent tracer application for evaluating the brightness of quinine tracer in liquid, ice, and soaked sponge forms for linear uniform flow conditions: a) 3D view and b) top view.

Figure 4 -
Figure 4 -Illustration of using quinine ice cube as a tracer for rotational flow movement in a rotating cylinder.
brightness intensity of the three forms of quinine tracer applied into the water surface was explored for linear (1D) flow conditions, on the laboratory flume.For a clear water flow depth of 0.01 m, Figure 9 shows photographs of the applied tracers taken at time intervals Δt = 0.5 s and Figure 10 shows, in each panel, two superimposed brightness intensity graphs corresponding to images captured with a time lapse of 1.0 s.It becomes clear that the ice had the highest brightness intensity in comparison to the other two forms of tracers and it remained approximately constant during the measurement time window, which is of a few seconds.For very shallow/sheet flows, small ice cubes should be used because, in these conditions, large ice particles can get stuck to the flume bed and eventually give wrong estimations of the flow velocity.The same applies with respect to the size of the soaked sponges to be used.The quinine liquid solution had a good brightness intensity (i.e., good visibility) at the beginning of the experiment but that intensity tends to fade out over time, due to the dispersion of the tracer in the flow mass.After a Δt = 1.0 s time lapse, the brightness level decreased significantly (i.e., 55%; Figure 10 -top).The sponge soaked with quinine solution maintained better its brightness level during each of the experiments' measurement time.At the beginning of each experiment, Source: elaborated by the authors.Note: The images are top views of the water surface, where the tracers were applied.

Figure 5 -
Figure 5 -Illustrative comparison of the visibility of three types of application of quinine tracer (liquid, ice, and soaked sponge) in still waters, for water column depths of 0.01 m (top), 0.10 m (middle) and 0.65 m (bottom) and for clear water and two levels of suspended sediment loads.

Figure 6 -
Figure 6 -Brightness intensity of three fluorescent quinine tracer forms applied in still water conditions for different water column depths and levels of suspended fine materials loads (medium and high suspension loads correspond here to 2 and 5 g•L -1 , respectively).

Figure 7 -
Figure 7 -3D illustration of the brightness intensity of all three forms of tracers at the same time and condition for the application of the tracers in still water conditions: clear water (top), medium suspension load (2 g•L -1 ; centre) and high suspension load (5 g•L -1 ; bottom), for the water depth of 0.65 m.

Figure 8 -
Figure 8 -Brightness intensity of the three tracer forms applied into still water, for a water column depth of 0.65 m.
With respect to the brightness of point application of fluorescent quinine tracer into surface waters, the following conclusions could be drawn from this set of experiments: i. Liquid quinine solution, when applied as a flow velocity fluorescent tracer, should better be used on sheet flows and shallow overland flows and still water, in clear water environments.Its visibility in deep and muddy flows is many times insufficient for surface flow velocity estimations.Liquid quinine Brightness of fluorescent quinine tracer Source: elaborated by the authors.Note: The water depth was 0.01 m and the flow velocity was approximately 0.10 m•s -1 .

Figure 9 -
Figure 9 -Comparison of the visibility of three forms of quinine tracer applied into clear water, for linear (1D) flow.

Figure 10 -
Figure 10 -3D illustration of the variation in brightness intensity of three tracer forms of liquid (top), ice cube (centre) and soaked sponge (bottom) applied in linear flow condition for a clear water depth of 0.01 m.
Source: elaborated by the authors.Note: The liquid and soaked sponge quinine tracer forms were not visible.The water column depth in the initial static condition was 0.10 m.

Figure 12 -
Figure 12 -Comparison of the brightness intensity of quinine tracer ice cubes in a circular water movement with a velocity of 20 rpm, and different suspended sediment loads.

Figure 11 -
Figure 11 -Comparison of the visibility of two forms of quinine tracer (ice and sponge) for medium and high sediment suspended loads (linear movement condition with a flow depth of 0.10 m).

Table 1 -
Surface flow test results for ice and sponge with linear flow conditions under different suspended loads, water depths and discharges.
Source: elaborated by the authors.Note: Mean values and standard deviations (SD) are for four replicates.