2.1. Study Site and Sampling Points

Yeo-cheon River flows through Suwon city, Gyeonggi province, South Korea before emptying its content into Woncheon Reservoir. The reservoir (37°16’51” N and 127°03’46” E), which has surface area of 8.96 km² and a volume of 198.8*103 m³, was established for the purpose of securing water for agricultural use (^{Park et al., 2006}). But now a days, it has become a major recreation center after the development of park and apartment complexes in the vicinity. The sampling positions were determined relative to the jet streamer location (37°17’11” N and 127°03’41” E) in the river at stretch in the mouth to the Woncheon reservoir and in the reservoir itself (Fig. 1(a)). Specifically at 50 and 25 m upstream of the device position and at 5, 25, 50 and 100 m downstream of the device location. The upstream locations were considered to represent the area not subjected to the unit effect.

Fig. 1. (a) Location of the six sampling points in the Yeo-cheon River at the mouth of the Woncheon Reservoir, Korea; (b) a schematic diagram of the combined device and its working mechanism.

The data and sample collections were made near the middle of the specific site water column at all the sampling points. Except the 100 m position, samples were also taken at the top (5 cm below the water surface) and at the bottom (10 cm above the sediment) layers in the downstream points. The river quality was monitored for short term effect of the device operation at the six different points in August, September and October 2015, comprising eleven water quality parameters. The period was chosen as it corresponds to the season of cyanobacteria bloom in Korea (^{Ahn et al., 2007}). The parameters were selected due to their implication to water quality problems related to algal blooms. The selected water quality parameters were temperature (T, °C), depth (m), flow rate (V, m/s), turbidity (NTU, Nephelometric Turbidity Units), dissolved oxygen (DO, mg/L), chlorophyll-a (Chl-a, mg/m³), Dissolved organic carbon (DOC, mg/L), total dissolved nitrogen (TDN, mg/L), ammonia nitrogen (NH₃-N, mg/L), total dissolved phosphorus (TDP, mg/L) and phosphate phosphorus (PO₄^3--P, mg/L).

Temperature, DO and pH were measured in the field during sample collection using M-100 and Chl-a using chlorophyll probes (Technology and Environment corp., Korea). The flow was also determined in-situ using MiniAir 20 flow sensor (Schiltknecht, Germany) and Turbidity was measured in the lab by turbidimeter (HACH Company, USA). The NH₃-N, TDN, PO₄^3--P and TDP were all determined in the laboratory using DR/5000 spectrophotometer (HACH Company, USA), after filtering the water sample through 47 mm glass microfiber filters (Whatman Ltd, UK) since turbidity and color may cause inconsistency of results. DOC concentration was determined by high-temperature combustion on a TOC-V CPH (Shimadzu, Japan) analyzer.

Total nutrient concentration (phosphorus and nitrogen) were considered as the trophic status of a waterbody is based on them, instead of just dissolved inorganic phosphate or dissolved inorganic nitrogen. Ammonia is often primary form of N dissolved in aquatic system and it is major form used for algal growth. Similarly, orthophosphate represents soluble reactive phosphate, which is the only phosphorus compound readily available for algal uptake without further breakdown. Table 1 gives the average initial values of some water quality parameters before the first operation of the device on August 31, 2015.

2.2. The In-Situ Water Quality Control Device and Its Operational Conditions

The main components of the device were water driving pump equipment, water flow generator, ultrasonic irradiation unit, ozone generator and equipment floating body (Fig. 1(b)). The water flow generator created jet flow of 70, 000 m³/day from a PVC made nozzle rectifier tube of overall length 2,000 mm and inner diameter 300 mm. An electrical centrifugal pump provided driving water at discharge of 3.0 m³/min. The ultrasonic irradiation device had 6 transducers oscillating at 200 kHz frequency for 30 s of irradiation contact time with water; while the ozone generator had a quartz glass double discharge tube to generate ozone to the unit duct at concentration of 3*10^-3 mg/L or at the rate of 9 g/h. The water was subjected to ultrasonic irradiation at the mouth of the duct and then mechanically mixed while drugged by the flow generator before discharged to the downstream. A floating body with overall length of 5,457 mm and width of 2,350 mm kept the whole device system at the middle section of the river. Operation control and electrical equipment were also installed on the floating body having handrails and slip prevention floor.

The water circulation was supposed to expose the phytoplankton escaped untreated in the device during the early laps, making the system more efficient. Oxygen rich water layer is formed over sediments, subjecting the dysfunctional algae sunken to the bottom to strong oxidization and aerobic digestion processes (^{Herald, 2011}; ^{Yoshinaga and Kasai, 2002}).

3.1. pH and Temperature

One day before the sample collection, the jet streamer, which had been running daily for study other than described here, was switched off for the short term analysis. The water pH was limited within the narrow range of 7.6 to 8.8 before the experiment, showing continues increase with time following the flow operation. However, there was pH decline at upstream and at the furthest point in the downstream when the other components were not integrated to the flow. The flow caused water column mixing which could be followed by release of acidic gases, such as carbon dioxide and hydrogen sulfide, in to the atmosphere, hence increasing the pH. The average increase in pH by the water flow and by the ultrasonication was almost similar. Even though ^{Ahn el al. (2003)} reported decrease of pH with ultrasonic radiation in an enclosure study, ^{Yu et al. (2013)} emphasized that in an open system the complex reactions may lead to increase in pH.

Carbon dioxide is produced from ozone oxidation of organic matters to form bicarbonate ion by combining with a hydrogen ion contained in water, releasing a hydrogen ion, thus increasing the pH (^{Weinberg and Narkis, 1992}). This effect of ozone on pH was recognized only at 5 m distance after 3 hr. But, extended ozone oxidation lowered the pH after 3 hr. The limit in amount of ozone used was the likely cause for longer time taken at locations further downstream to experience the same effect.

Even though the water was shallow (maximum depth 2 m), Fig. 2 shows that there was temperature variation by more than 2°C between the water surface and the layer above the sediment before the jet streaming started to run. This has the potential of chemical stratification in the water column, stimulating rapid decline of oxygen concentrations with depth, and then enhancing nutrient release from sediments which may promote algal blooms. The slight decrease of pH vertically to the sediment layer (result not shown here) supports this assumption. The temperature of the water column increased and became more uniform due to the water circulation which was triggered by the flow driving pump and the flow generator. Diurnal heating and cooling played a role in spread of the vertical lines horizontally, showing temperature variation. The 6 hr data of the water pump operating alone was taken late during the day that the temperature decreased.

Fig. 2. Vertical variation of water temperature downstream from the device under jet streaming (JS), ultrasonication (US), and ozone (O3) application.

3.2. Flow Rate and Dissolved Oxygen (DO)

The flow rate of the river caused by the jet streamer before and during the device operation is shown in Table 2. The further a point was located from the device, the flow strength got diminishing. Thus, the efficiency of the device on the water quality improvement was consistent in the vicinity close to the device, where the flow was relatively strong. Coupling ultra-sonication with the pump flow, stronger water current was formed, but limited to 50 m. After 3 hr, the flow velocity grew to 0.2 m/s from the initial rate of 0.01 m/s. In the ozone incorporating system, the magnitude of the flow increased even more, with highest value of 0.39 m/s at 5 m after an hour; however, the stretch it covered was shorter. The change in the water quality parameters followed more or less the water flow velocity pattern.

Before the unit operation, the overall mid-depth DO in the water varied from 4.5 to 8.1 mg/L. Fig. 3 illustrates that after the operation commenced, the DO values of the water column were greatly improved. The pump induced flow by itself enhanced the DO until 3 hr of operation, especially at 25 m downstream (the lines are widely spaced). There were increase in DO upstream and downstream, except for the 100 m point downstream where there was almost no effect. The pump caused movement of the upstream water, and to the downstream direction, the discharge got attenuated to value of zero at 100 m, diminishing the effect.

Fig. 3. Depth wise variation of dissolved oxygen (DO) with time in the downstream, on 8/31 (jet streaming, JS only), 9/16 (JS and ultra-sonication, US) and 10/14 (JS, US and ozone, O3).

Based on enclosure experiment, ^{Ahn et al. (2003)} found that DO declined by ultra-sonication treatment. Generating ultra sound waves in the flow, the effect of water circulation on DO dominated that of ultra-sonication in the early stage of the operation, and later on vise-versa. That is why the DO values are more after 3 hr than 6 hr at the 5 m distance. But as the flow effect got stronger at further distance, the overall DO was improved.

The application of ozone to water body results in increase of DO concentration in the water due to decomposition of molecular ozone to oxygen; however, the amount was very small compared to that added by air or oxygen (^{Weinberg and Narkis, 1992}). Moreover, due to the limited dose of ozone applied, significant change was observed only at 5 m from the device. Fig. 4 presents the percentage increase of DO at the layer immediately above the sediment, which is of paramount importance in terms of oxidation of micropollutants, biodegradability and nutrient release.

Fig. 4. Percent increase of DO with time at bottom (sediment) layer and at 5 m, 25 m, and 50 m downstream from the device.

Chlorophyll-a and turbidity

3.3.

Initially, the Chl-a varied from 1.5 to 46 μg/L in the water column, with values as high as 33 μg/L recorded at the surface. It is shown in Fig. 5 that there were dramatic decline of the Chl-a concentrations under the three operation conditions. Running the jet streamer alone, there was no significant change until the 1 hr mark. However, decline in Chl-a content as high as 80% was achieved after 3 hr at locations further from the unit. The concentration increased momentarily upstream before it showed limited improvement. The increase in oxygen levels influenced redox reactions in the bottom layer, in turn this partly determined the availability of nutrient (especially phosphorus) from sediment surface. At first, the change was small may be due to cell distribution caused by incomplete mixing (^{Pastorok et al., 1981}). Detailed investigation is required to explain the real mechanism.

Fig. 5. The change inchlorophyll-a(Chl-a) concentration at both the upstream and downstream sampling points during the short-term operation, under (a) Jet Streaming (JS) only, (b) JS and ultrasonication (US), and (c) JS, US, and ozonation (O3).

Combining the flow with ultra-sonication, shorter time was taken to remove Chl-a from the water column at downstream points. The significant decrease of Chl-a was the result of sonication, which facilitated sedimentation (sinking loss) after collapsing the gas vacuoles (^{Ahn et al., 2003}). However, there had been continuous increase of the concentration at the upstream, throughout the operation time as the sonication did not address upstream of device.

When ozonation was combined to the system, Chl-a was completely removed from the water column up till the 25 m stretch just after 1 hr operation. The removal was also substantial at the remaining reaches, even though there was slight fluctuation after an hour. The application of ozone enhanced effectiveness of ultra-sonication in algal inactivation as the powerful oxidizing nature of ozone raptured more cells. Generally, integrating the three system components showed instant impact at all the sampling points. However, in the operation of ultra-sonication, probably due to the water clarity following sedimentation helped more light penetration, sedimented pigments settled down as the result of device operation in the previous days recovered and suspended. Time of the day also likely affected efficiency of ultrasound on cyanobacteria control (^{Ahn et al., 2003}; ^{Leclercql et al., 2014}). The possibility of cell distribution due to water circulation should also be mentioned.

On the other hand, the short term jet streaming showed continues decrease of turbidity in the downstream direction up to 50 m (Fig. 6). There was increase of turbidity shortly after the device started operation and then decrease, at both the upstream and the furthest point of 100 m. The increase of turbidity in the upstream is likely related to sediment suspension caused by the water pump. Operating ultra-sonication with the flow, there was slight fluctuation of turbidity in the closest vicinity. The turbidity magnitude has generally been so low. Further downstream and to the upstream, there was mixed trend of increasing and decreasing.

Fig. 6. The change in water turbidity in Nephelometric Turbidity Units (NTU) concentration at both the upstream and downstream sampling points during the short-term operation, under (a) Jet streaming (JS) only, (b) JS and ultrasonication (US), and (c) JS, US, and ozonation (O3).

When ozone was added to the system, turbidity declined promptly until the 3 hr of operation, after which it increased at all the points. Turbidity indirectly refers to suspended matter, which are removed by ozone from water, presumably because ozone induces flocculation and then sedimentation (^{Davidson et al., 2011}). Turbidity may fluctuate rapidly in waters to which ozone is injected due to the flocculating effect of ozone on colloidal organic material (^{Park et al., 2013}). The turbidity changing trend matched with that of Chl-a, showing that sedimentation of cells resulted in clarity of the water column.

3.4. Phosphorus, nitrogen and dissolved organic carbon (DOC)

The concentration of TDP decreased at the surface and middle layers of the 5 m point when the water was allowed to flow and other system components non-operational. But, there were mixed features at the other locations. Although oxidation of sediment-water interface is supposed to form a barrier to the release of dissolved phosphorus ions from sediments, the increase in temperature of the sediment might simulate decomposition and suspension of phosphate to overlaying water (^{Pastorok et al., 1981}). The water flow over the sediment may also increase phosphate release.

In the sonicated system, generally the TDP decreased by more than 50% at all the layers after 6 hr. However, likely due to dissociation and dissolution of loosely adsorbed phosphate from the sediment particles, TDP increased at the bottom layers of the 25 m and 50 m locations in the meantime. The effect is correlated with turbidity and Chl-a concentrations. The TDP concentration at the middle of the water column is given in Table 3.

There is slight increase of the TDP concentration when ozone was applied, especially in the middle and bottom layers. Ozone was clearly playing role of dissociating phosphate from sediment particles. Even though DO got improved over time, the increase in TDP showed that oxygen was not alone responsible for phosphorus control. Probably, different phosphorus mobilization mechanisms such as decomposition of organic matter and pH centered phosphorus release exceeded the importance of oxygen-controlled release of iron-bound sedimentary phosphorus.

^{Koski-Vahala and Hartikainen (2001)} demonstrated that high pH, associated with intensive resuspension, may significantly increase the internal P loading risk. But, the sediment of shallow water body is the site for numerous highly dynamic processes that it may have very substantial effects on the total P budget and other water quality parameters. In deep water column, a redox dependent accumulation of P occurs in the anoxic hypolimnion during stratification. However, in well mixed and whole water column oxidized shallow water, the sediment has often been confirmed to release P to oxic water body, suggesting that other factors than redox conditions at the sediment-water interface are involved (^{Sondergaard et al., 2003}).

On the other hand, when the water was allowed to flow in this study, the reactive phosphate showed mixed trends. With joined ultra-sonication and flow operation, the reactive phosphate decreased more or less in the same way as the TDP (Table 3). Finally, when ozone was added to the system, there was increase of reactive phosphate up to 3 hr and then slightly decreased. Overall, there is almost no change between the initial and final operation time.

Concerning nitrogen, there was no any remarkable trend of total nitrogen and the initial amount itself was not significantly high (average 1.0 mg/L) to pose considerable quality problem. When water flow was combined with ultra-sonication, the ammonia nitrogen increased intensely and then decreased, at all the sampling points. Ultra-sonication can kill bacteria and hence might have affected the bacteria mediated nitrogen cycle (^{Laliberte and Haber, 2014}). Applying ozone to the system, the ammonia nitrogen was gradually suppressed with time to mean value of 0.02 mg/L, in the downstream direction. Some of the ammonia likely reacted with the ozone to form nitrate. Dissolved organic carbon (DOC) is parameter which includes active chemical matter. Clear contribution of the device operation on DOC could not be recognized, even with ozone application, within the short time of operation considered.

Finally, it worth mentioning the climatic condition which likely affected the flow, ultrasonic and ozone combined system. The average rainfall values a week before sampling for case 1 (water flow), case 2 (ultrasound and water flow combined) and case 3 (flow, ultrasound and ozone all together) were 3.4, 2.5 and 10 mm, respectively and the corresponding average wind speed two days prior to operation were 1.2, 1.0 and 2.4 m/s, respectively. Hence, the intensive rainfall and strong wind speed prior to the operation having ozone as component could have some impact on the result.

The rate of oxygen diffusion is affected by the mixing of water layers caused by artificial circulation and mixing due to wind action. Thus, wind action has the possibility of affecting the water quality through increased diffusion. Besides, windinduced resuspension is reported to be a mechanism that frequently causes increased concentrations of suspended solids in shallow lake water (^{Sondergaard et al., 2003}). However, this mechanism is suggested as a possibility in the present study.