3.1. Fluctuation of Phytoplankton Cell Abundance in the Raw Water Source
The fluctuations in the cell abundance and dominant species of phytoplankton in the second water intake structure of Paldangho Lake (PIF) caused by changes in water temperature and precipitation during the study period are illustrated in Figure 3
and Table 2
. The cell abundance of phytoplankton before 7 July was below 1500 cells/mL, and the diatom species Cyclotella meneghiniana
, Fragilaria crotonenesis
, and Aulacoseira granulata
were the most dominant. The cell abundance increased to 23,136 cells/mL between 21 July and 24 August when the water temperature was beyond 25 °C, while three cyanobacteria (Microcystis aeruginosa
, Dolichospermum crassum
, and Merismopedia tenuissima
) became dominant with a relative abundance of over 90%. This increase was due to increased high water temperatures (Figure 3
A) and total phosphorus concentration and rainfall in mid-August (Figure 3
B). After September, the percentage of diatoms increased to 30%–80%, with a similar species composition as that in April–June, whereas the percentage of green algae and cyanobacteria decreased to below 30%. After September, the dominant species was Aulacoseira granulata
, a diatom. Analyzing the fluctuation of phytoplankton in the water intake facility demonstrated that the growth of cyanobacteria increased exponentially in the presence of nutrients and under high water temperature conditions. A previous study showed that nuisance cyanobacteria can grow under high water temperatures [20
]. Therefore, the use of a pretreatment system such as ABM during high temperature periods is exclusively required to prevent the inflow of nuisance cyanobacteria into water purification plants through water intake structures.
3.2. ARE of the ABM at Each Water Depth
For an in-depth analysis of the algal removal efficiency (ARE) of the ABM during the study period, the water level was divided into three layers: surface layer (0–2 m), middle layer (3–5 m), and deep layer (>5 m).
In the surface or SBM layer, the ABM removed approximately 38.7% of the total phytoplankton. The ARE of the ABM averaged as 63.8% in the high water temperatures over 25 °C when cyanobacteria density was high. As the cyanobacteria under high water temperatures were concentrated in the surface layer, the cell abundance that flows to the inside of the ABM was effectively reduced by the SBM, leading to a high ARE (Figure 4
A). In the middle or AFM layer, the mean ARE of the ABM was 15% with a wide range of −98% to 65%. The negative value indicates higher phytoplankton density in the water of IN passing through the ABM compared with the reservoir. The mean ARE was 42.1% during high water temperature periods (Figure 4
B). In contrast, in the bottom or no-ABM layer, the ARE averaged 31.7% with a wide range of 0%–91%, with an ARE of approximately 50% during high temperature periods (Figure 4
C). As the bottom layer had a low density of total cyanobacteria, even during the high water temperature periods, the ARE was meaningless. These results indicated that the algal removal ability of the ABM is more effective for cyanobacteria concentrated in the surface layer during high temperature periods. In general, the use of algal blocking mats or fences is more effective for dense algal blooms like scum [16
] or stagnant water with a long water residence time [21
]. In this study, the simple blocking mat, corresponding to the surface layer, showed the highest ARE for high density cyanobacteria such as Microcystis aeruginosa
3.3. ABM Effects over Time
To monitor the daily fluctuation of phytoplankton and their removal by the ABM, sampling was conducted three times between 2:00 p.m. on 24 August and 8:00 a.m. on 25 August. In this study, the mean water temperature was 28.3 °C and the cyanobacterial dominance occupied over 90%. Among the phytoplankton in both the PIF and reservoir water during the study, cyanobacterium Microcystis aeruginosa
was found to be the dominant species (Figure 5
, Table 2
At 2:00 p.m. on 24 August, the cell abundance of phytoplankton was 66,000 cells/mL in the surface layer, but they sharply decreased as the water deepened. The ARE of the ABM was the highest (92.0%) in the one meter layer, with an average of 57.7%, which gradually decreased with water depth (Figure 5
A). At 20:00 on 24 August, the phytoplankton was over 45,000 cells/mL in the surface layer. In the layers below, unlike the results obtained at 2:00 p.m., the phytoplankton was maintained at over 30,000 cells/mL between two and five meters. The ARE was the highest (68.5%) in the three meter layer, with an average of 60.8%, but there were no large differences among water depths (Figure 5
B). At 8:00 a.m. on 25 August, the cell abundance of phytoplankton at all water depths was below 15,000 cells/mL. The ARE was the highest (12.3%) in the one meter layer, averaging 6.5% over all depths, showing a lower algal removal efficiency (Figure 5
C). Such changes in ARE with daily fluctuations in phytoplankton biomass can be explained by the vertical migration of Microcystis
]. In particular, cyanobacterium Microcystis aeruginosa
can control their buoyancy with light; they rise to the surface layer for photosynthesis during the day and sink to the deeper layers at night due to the carbohydrates produced via photosynthesis [23
3.4. Algae-Related Odor Removal Efficiency of the ABM
Concentration changes in the odor and taste compounds 2-methylisoborneol (2-MIB) and geosmin showed different tendencies over the study period. 2-MIB increased with increasing water temperatures, and was the highest on 8 September, then gradually decreasing after 23 September (Figure 6
A). The odor and taste removal activities of the ABM were different from the algal removal efficiency; the 2-MIB concentration at the PIF site on 23 September and 6 October was slightly higher than those of the OUT reservoir. We could not explain why the 2-MIB increased in IN. Of the cyanobacteria, two other genera, Oscillatoria
, were more abundant than Microcystis aeruginosa
in PIF on the same day. The cellular abundance of Pseudanabaena limnetica
showed a similar pattern to the 2-MIB concentrations. Generally, the production of 2-MIB in the freshwater environment is characterized by aquatic organisms such as cyanobacteria and actinomycetes [25
]. Further studies are needed to understand the species-specific contribution of cyanobacteria to MIB production.
The concentration of geosmin was highest on 24 August, rapidly decreasing after September (Figure 6
B). The geosmin concentration showed a significant correlation (r
= 0.70) with the biomasses of Microcystis aeruginosa
and Dolichospermum crassum
, which are cyanobacteria commonly found in the PIF [4
]. The mean removal efficiency of the ABM against odor and taste compounds was 26%, and the highest occurred in September when the cyanobacteria were most prevalent. Although phytoplankton content in the ABM was 38% after September, the concentrations of odor and taste compounds in PIF were high, demonstrating poor removal efficiency. Therefore, both 2-MIB and geosmin need to be controlled between August and September when the water temperature rises and the concentrations of both these odor compounds are high. However, 2-MIB should also be controlled in September when the water temperature drops. To reduce the concentration of odor and taste compounds that are produced inside the structure even when the water temperature drops, ABMs need to be predominantly used when cyanobacteria appear, or a depth adjustment within the ABM is needed to stabilize the water flow inside the structure.
3.5. Effects of the ABM and Its Operation
When the PIF maintains conditions with enough nutrients and solar radiation, it leads to the prevalence of Microcystis
species. The light intensity was 3000–30,000 lx inside the ABM and 100–20,000 lx outside. Analyzing the ARE using the difference between the light intensity inside and outside the ABM showed an ARE of over 35% (Figure 7
). However, the light intensity measurements gradually decreased 10–14 days after the installation of a photometer, which complicated the analysis. As biofilms grew on the surface of the photometer installed underwater, the measured light intensity declined. Measuring the ARE of the ABM using a photometer is fast, whereas measuring the ARE through biomass measurements takes much longer as the biomass needs to be analyzed under a microscope. Additionally, information on the cycle of biofilm formation obtained due to the decrease in light intensity provided data for determining when the ABM needed to be washed to maintain its efficiency. The data suggested that the ABM should be washed once every other week in midsummer as a baseline when biofilm formation is most active to ensure stable operation of the structure.
The raw water flowing into the PIF changed the phytoplankton community and water quality parameters daily due to the ABM and hydraulic flow. The effects of ABM on pH, electric conductivity, dissolved oxygen (DO), Chl-a
, and turbidity related to phytoplankton growth were analyzed during the experimental period (Table 3
). The overall water temperature was slightly lower inside the PIF than in the reservoir. This suggested that the low water temperature inside the PIF was due to the pumping of water into the deep portion of the reservoir. The ABM was originally designed to entirely block the inflowing reservoir water of 0–2 m depth, called the SBM layer. Over the study period, the electric conductivity and pH inside the PIF were lower than those of the reservoir, from 25 August to 23 September, inside the PIF the electric conductivity and pH was temporarily higher than in the reservoir, where diatoms flourished with decreasing water temperature. The turbidity was higher inside the PIF than in the reservoir due to the floating of bottom sediment by the pumping of the reservoir water below the two-meter depth, consistent with the water temperature trends. Dissolved oxygen concentration was lower inside the PIF than in the reservoir during the study period, which occurred in the process of accepting the deep layer water of the reservoir, in the case of turbidity and water temperature. Another reason is due to the higher density of phytoplankton outside than PIF rather than inside [29
]. Therefore, the lower reservoirs around the PIF are undergoing anaerobic or anaerobic decomposition [30
]. As mentioned in the ARE, relative to phytoplankton, Chl-a
was lower inside the PIF than in the reservoir when the cyanobacteria dominated, but showed the opposite phenomenon when diatoms dominated. In summary, the ABM we applied and tested can effectively block the inflow of cyanobacteria by blocking the surface water of flowing rivers, as mentioned for phytoplankton. However, when raw water from the reservoir is introduced into the drinking water purification plant using a pump, it less effectively improves the water quality due to the inflow of turbid low temperature water from the SBM to the bottom layer.
In this study, the incidence of destroyed cyanobacterium Microcystis and colonies by ABM was not confirmed. Within the ABM, the SBM layer can effectively block the surface cyanobacteria, but AFM can also destroy Microcystis and colonies and introduce microcystin into the PIF. Therefore, to confirm this, in vitro testing of the destruction of Microcystis and colonies, and the production and penetration of microcystin using AFM is required.
In terms of biological methods [31
], various physical and chemical treatment methods for blocking cyanobacteria have been reported worldwide, including in Korea [3
]. Researchers have demonstrated that the dissolved air flotation method (DAF) is one of the most effective physical methods for removing 93%–98% of Chl-a
. As a chemical treatment, yellow clay or loess is often used in Korea to control both the cyanobacterial bloom in fresh water and even red tide in seawater. Although these technologies are effective in suppressing cyanobacteria in a small-sized laboratory and on a mesocosm scale, they are difficult to apply to water purification plants that supply a large amount of water resources, generate secondary pollution sources, and are inefficient in terms of processing costs. The ABM method did not generate large amounts of by-products or secondary pollutants such as algal biomass, 2-MIB, geosmin, and toxins like microcystin after treatment. In addition, ABMs can be sufficiently applied to block cyanobacteria in drinking water purification plants that require large amounts of water supply, and to block suspended matter such as high concentration turbid water due to rainfall. However, periodic washing must be performed and physical damage to the ABM device must be avoided during typhoons or abrupt weather changes depending on the season.
In order to block nuisance cyanobacterium Microcystis and its colonies in water purification plants like PIF, the application of the ABM model should be considered based on the following points:
1. Above all, a physically robust structure is required in a location where the residence time is short and the flow rate is high, such as in Paldangho Lake. Although a heavy weight was attached to the bottom of the AFM in this study, warping was observed due to the strong pumping pressure. This may cause turbid water inflow even though the phytoplankton density lowers as water depth increases. In this study, the maximum installation depth of the PIF is 9 m, but since fluctuations occur due to rainfall or Paldangho Lake discharge, when the low water capacity decreases, the inflow of sediments or turbid water from the bottom layer is inevitable. Therefore, a location less influenced by water depth is a prerequisite.
2. During the biofilm formation and cleaning of the AFM film, determining the exact cleaning cycle using factors such as water temperature or phytoplankton succession is difficult. Although the 1 m underwater photometric method attempted in this study is one alternative, the frequent washing required due to rapid biofilm formation is difficult when the density of phytoplankton or attached organisms increases. Therefore, the exact cleaning cycle must be determined and in-field AFM washing technologies must be developed.
3. ABM is more effective than other curtains or simple blocking membranes [14
] and is economical compared to chemical treatment methods [33
]. The structure of a simple membrane is difficult to maintain when the flow rate increases due to rainfall or the flow rate is strong due to a strong bending of the structure due to strong pumping pressure. In the case of Paldangho Lake, since cyanobacterial blooms are frequently generated, chemicals such as activated carbon and loess must be used in the water purification process. In addition, various side effects are expected. However, simple comparisons of the short-term ABM model with other technologies is complicated. Thus, long-term field application and technology improvement studies on the ABM model are required for economic evaluation.