Removal of Algae and Cyanobacterial Toxins during Slow Sand and Bank Filtration Ingrid Chorus, Hartmut Bartel, G. Grützmacher, G. Böttcher Umweltbundesamt, Versuchsfeld Marienfelde, Schichauweg 58, D-12307 Berlin, Germany (E-mail: ingrid.chorus@uba.de) Abstract The German Federal Environmental Agency (Umweltbundesamt UBA) operates an experimental field on the southern outskirts of Berlin for the investigation of special aspects of environmental hygiene. One part of the field consists of a storage pond with subsequent bank filtration, artificial recharge and slow sand filtration where experiments on the behavior of hazardous substances during bank filtration, artificial recharge or slow sand filtration can be carried out on a field scale without any adverse environmental impacts under nearly natural conditions. Experiments on the elimination of cyanobacteria and their toxins are currently being conducted in the plant as part of the research project of the KompetenzZentrum Wasser Berlin (KWB) named Natural and Artificial Systems for Recharge and Infiltration (NASRI). Keywords: Field scale experiments; bank filtration; slow sand filtration; cyanobacteria I. Introduction For NASRI s experiments on the elimination of a variety of substances present in Berlin s surface waters, a field site operated by the German Federal Environmental Agency (Umweltbundesamt UBA) on the southern outskirts of Berlin is a central facility. One part of the field consists of a storage pond with subsequent bank filtration, slow sand filters and infiltrations ponds (Figures 1 2), which is operated by the Agency s water pollution control section (Section II 4.3). This plant as well as the results of preliminary experiments on slow sand filters using cyanobacterial toxins are slow sand filters & infiltration ponds surface area (water): 3300 m² 45.5 m total area: 5290 m² bank filtration 88.3 m storage pond 55.4 m > 2 m 1,5-2 m 1 1,5 m 17.5 m 23 piezometers inlet Conference Wasser Berlin 2003 51
Berlin Centre of Competence for Water presented below. Figure 1: Areal view of the storage pond system (supplemented from Bartel & Grützmacher 2002). infiltration ponds bank filtration storage pond piezzometers inlet fine sand 0.8/2 mm gravel 2/8 mm coarse gravel 32/56 mm middle sized gravel 8/20 mm 18 m 10 m 20 m 88 m Figure 2: Cross section through the storage pond system. II. Description of the site and possible modes of operation The storage pond system is fed with groundwater abstracted from the surrounding quaternary aquifer. Before passage into the pond, the water is subjected to biological treatment for removal of iron and manganese. From the treatment plant the water flows into the storage pond, which has a capacity of 7,000 m 3 and a surface area of about 5,300 m 2 (Figure 1). From the storage pond, the water is conveyed into two slow sand filters and two infiltration basins, each with a surface area of about 75 m 2 (Figure 1). The slow sand filters have drainage pipes which collect the water following passage through the sand and conduct it to a measuring station. In addition, there are two subsurface drains which collect the water at a depth of about 5 m over the entire width of the storage pond at points 14 m and 36 m behind the bank line, and likewise convey it to the measuring station. There, the parameters ph, O 2, electrical conductivity, temperature, redox potential, TOC, TN b and fluorescence are measured and analysed continuously in flow-through mode controlled and registered by a PC. The water from the drainage systems can be either pumped back to the inlet of the storage pond (recirculation) or discharged to the sewage system (flow-through). The intake, and hence the flow, can be regulated via the pumping rate. Minimal flow rates are a few hundred liters per hour, the achievable maximum is 40 m 3 /h. The subsoil within the area of the bank filtration plant as well as the filling of the slow sand filters and infiltration ponds consists of medium- or fine-grained sands and gravel of good permeability (average k f value: 2*10-3 m/s). Accordingly, flow rates of between 0.2 m/d and 6 m/d can be set depending on the output of the pumps. 52 Conference Wasser Berlin 2003
infiltration pond piezometers storage pond about 4 m sand (0.8 / 2 mm) gravel (2 / 8 mm) 36 m gravel ( 8 / 20 mm) concrete gravel (32 / 56 mm) with drainage pipe Figure 3: Schematic cross section through the bank filtration passage (supplemented from Bartel & Grützmacher 2002). The major advantage of this plant is that experiments on the behaviour of hazardous substances and their metabolites during bank filtration, slow sand filtration or artificial recharge can be carried out on a field scale, but with full budgeting of amounts added to the system and amounts found in the filtrate. Also, as the system is sealed to the subsurface, work with hazardous agents is possible without any adverse environmental impacts. In spite of these tightly controlled conditions, external conditions scale factors (e.g. weather conditions) largely match those found in the real environment. Thus, the sites fill the gap between lab experiments (full control, but poor representation of real life conditions) and field observations (full representation of real life conditions, but poor control). III. Experiments on the Elimination of cyanotoxins during Sediment passage Cyanobacteria, some of which contain hepatotoxic polypeptides so-called microcystins (MC)-, can be found in surface waters worldwide. When present in water abstracted either directly or indirectly (via bank filtration) for use as drinking water, these cyanotoxins, (toxic in concentrations as low as a few micrograms per liter), can jeopardize the supply of drinking water. As a part of the research project of the KompetenzZentrum Wasser Berlin investigations are being carried out on the elimination of microcystins during bank filtration. This includes field experiments that will be performed using the experimental plant described above. In a precursor project funded by the German government (BMBF) preliminary experiments were carried out on microcystin elimination by slow sand filtration. The aim was to find out which processes govern microcystin elimination during sand and soil passage. In healthy cyanobacterial populations more than 90% of the microcystins are cellbound (Sivonen and Jones 1999), so physical filtration of the cyanobacterial cells is the most important elimination process. This was shown in an experiment with live cells of Planktothrix agardhii, a filamentous cyanobacterial species, on a slow sand filter with well developed schmutzdecke (clogging layer). The cyanobacteria were added to the water body of the filter resulting in maximum cell-bound microcystin concentrations of 12 µg/l. After 0.8 m of sand passage (middle to coarsely grained Conference Wasser Berlin 2003 53
Berlin Centre of Competence for Water sand, d 50 = 0.518 mm) no significant amounts of particulate microcystins were detected. 99.99% were eliminated in the uppermost 10 cm (Table 1). Details can be taken from Grützmacher et al. (in prep.). Table 1: Physical filtration of cyanobacterial cells in a field scale experiment. Filtration velocity (m/d) 0.4 Duration of experiment (d 49 in water body 12 Maximum cell-bound MCconcentration in 0.1 m depth of filter 0.003 (µg/l) in 0.8 m depth of filter < 0.001 There are however situations when high shares of extra-cellular microcystins occur. This was the case in a previous experiment with live cells of Planktothrix agardhii from one of Berlin s lakes that were applied to a slow sand filter during autumn 1999. As shown in Figure 4 relatively high concentrations of dissolved microcystins (1.5 µg/l maximum 85% of total microcystins) were observed towards the end of the experiment. During this time the biovolume declined sharply due to low temperatures (Grützmacher et al. 2002). There are further indications that the cyanobacterial cells accumulated on the sediment surface act as a source of dissolved microcystins after cyanobacteria has ceased to be present in the water body. These observations were made during the final period of an experiment with continuous dosing of live cells of Planktothrix agardhii from a mass culture onto one of the slow sand filters (Grützmacher et al. in prep.). After the dosing had been shut off on day 54 of the experiment microcystin concentrations in the water body declined rapidly. On day 65 of the experiment no more cyanobacteria or microcystins were detected in the water body. In sampling points 10 cm below the filter surface dissolved microcystins were detected in traces until day 78 (Figure 5). 54 Conference Wasser Berlin 2003
total microcystins in water body (µg/l) 100 10 total microcystins dissolved microcystins 2,0 1,5 1,0 0,5 dissolved microcystins in water body (µg/l) 1 0 5 10 15 20 25 days after toxin application Figure 4: Total and dissolved microcystins during an experiment with cell-bound toxins (modified after Grützmacher et al. 2002). 0,0 MC (µg/l) and biovolume (cm³/m³) 2 1,8 1,6 1,4 1,2 1 0,8 0,6 0,4 0,2 end of dosing of cyanobacteria biovolume in water body total MC in water body total MC 10 cm below filter surface 0 50 55 60 65 70 75 80 days aber beginning of experiment Figure 5: Biovolume of Planktothrix agardhii and microcystins in water body and 10 cm below filter surface in an experiment with cell-bound microcystins during November 2001 (supplemented from Grützmacher et al. in prep.). In order to quantify elimination of dissolved toxins, cyanobacterial cells previously lysed by freeze-thawing were introduced into one of the slow sand filters (filtration velocity: 1.8 m/d), resulting in a microcystin concentration (dissolved) in the water of Conference Wasser Berlin 2003 55
Berlin Centre of Competence for Water 4.4 µg/l. Simultaneously, Na-fluorescine was added as tracer. Sampling of the effluent after 0.8 m of sand passage was conducted as one hour composite samples, with simultaneous continuous measurement of temperature, ph, redox potential, specific electrical conductivity and fluorescence. The results of the tracer and toxin analyses are shown in Figure 6 and Table 2. The average arrival time (when 50 % of the substance added is recovered in the effluent) were 318 and 551 minutes for the tracer and microcystins, respectively. This yields very little retardation (R = 1.73) and a low k d -value of 0.14 cm³/g for microcystins. Thus microcystins are adsorbed only slightly more than tracers, showing that this process plays a negligible role in cyanotoxin removal in sandy material as usually used for bank filtration. Sum MC (% of the applied amount) 16% 14% 12% 10% 8% 6% 4% 2% 0% Microcystins (ELISA) Tracer 100% 90% 80% 70% 60% 50% 40% 30% 20% 10% 0% 0 1000 2000 3000 4000 5000 Time (minutes) Sum tracer (% of the recovered amount ) Figure 6: Breakthrough curves of tracer and microcystins. On the other hand only 15 % of the toxin applied were recovered in the effluent. Modeling yielded half lives of 1 h, so this result indicates biodegradation taking place inside the sand filter to be the main process for eliminating dissolved microcystins. Table 2: Results of the field experiment with dissolved microcystins. Input-concentration of microcystins 4.4 µg/l Maximum concentration after filtration 0.46 µg/l Microcystin recovery after 3 days 15 % Avarage arrival time (tracer) 318 min Avarage arrival time (microcystins) 551 min Retardationfactor (R) 1.73 Distribution coefficient (k d ) 0.14 cm³/g The results of these experiments suggest that sand or soil filtration are secure treatment methods for source water contaminated by microcystins under most conditions. The main elimination processes are (1) physical filtration of mostly cellbound toxins and (2) biodegradation of the released, dissolved toxins. 56 Conference Wasser Berlin 2003
There are, however, still uncertainties about the reliability of biodegradation under a wide range of conditions met in practice. Are there limits to the extent and the rates of degradation when - a cyanobacterial population dies (and lyses) due to insufficient light and / or nutrients, low temperatures or application of algicides (when there is a high amount of extra-cellular microcystins)? - the sediment is sandy material with low shares of clay and silt (so there is poor adsorption of dissolved microcystins)? - low temperatures and / or anoxic conditions prevail (so rates of degradation decrease)? - there is a missing clogging layer or schmutzdecke (so biodegrading microflora is lacking)? - the sediment had no previous contact to microcystins (so adapted bacteria can not have developed)? These questions can not be satisfactorily addressed by laboratory experiments as the main elimination processes take place at the water-sediment interface due to macro- and microbiological processes and these are difficult to simulate adequately under laboratory conditions. Field observations at bank filtration sites are also limited to sites where and to times when cyanobacterial blooms take place and encompass great uncertainties regarding input concentrations, dilution, geological setting etc.. The storage pond system on the UBA s experimental field offers a unique opportunity to gain a thorough understanding of natural processes by field scale experiments under well defined conditions. References: [1] Bartel, H. and Grützmacher G. (2002): Elimination of microcystins by slow sand filtration at the UBA experimental field. In: Ray, C. (ed.): Riverbank Filtration: Understanding Contaminant Biogeochemistry and Pathogen Removal, p. 123 133. Kluwer Academic Publishers. [2] Grützmacher, G.; Böttcher, G.; Chorus, I. and Bartel, H. (2002): Removal of microcystins by slow sand filtration. Environmental Toxicology 17: 386 394. [3] Grützmacher, G.; Böttcher, G.; Chorus, I. and Bartel, H. (in prep.): Cell-bound and extra-cellular microcystins during slow sand filtration. [4] Sivonen, K. and Jones, G. (1999): Cyanobacterial toxins. In: Chorus, I. And Bartram, J. (eds.): Toxic cyanobacteria in water, a guide to their public health consequences, monitoring and management, p. 41 112. E & FN Spoon. Contact: Dr. Ingrid Chorus Umweltbundesamt, Versuchsfeld Marienfelde Schichauweg 58 D-12307 Berlin Tel.: +49-30-8903 1346 e-mail: ingrid.chorus@uba.de Conference Wasser Berlin 2003 57