Knowledge journal / Edition 2 / 2023


From removing micropollutants to inundation trial to combat peat subsidence

This is the seventeenth edition of Water Matters, the knowledge magazine of specialist journal H2O. This edition consists on a variety of topics, written by water professionals based on solid research.

The editorial board, consisting of experts from the sector, made a selection based on a clear relationship with daily practice in the water sector, which is the purpose of Water Matters. Research, results and findings form the basis for articles that describe new knowledge, insights and technologies with a view to practical application.
In this edition of Water Matters, you will read that BAKF ozone technology is a serious option for meeting the requirements of the new European Urban Wastewater Directive and is ready for practical application.
Furthermore, you will find articles on:
- nutrient and micropollutant removal with floating helophyte filters
- a study on the accuracy of methods determining the health risks of blue-green algae
- sewage water measurements to monitor spread of hepatitis A virus at a school
- an inundation trial that effectively inhibits peat subsidence
- a trial showing that sand transplantation and biostimulation can effectively remove organic micropollutants
- a study showing that in-line inspection pays off to identify the degradation of asbestos cement pipes in the drinking water distribution network
- a study showing that grass herb mixtures in sloping banks contribute to a climate-resilient water system
- an initiative to take a first step towards standardisation of data processing from water quality sensors.

Water Matters, like the journal H2O, is an initiative of Royal Dutch Water Network (KNW), the independent knowledge network for and by Dutch water professionals. KNW members receive Water Matters twice a year free of charge as a supplement to their journal H2O.
The publication of Water Matters is made possible by leading players in the Dutch water sector. These Founding Partners are Deltares, KWR Watercycle Research Institute, Royal HaskoningDHV and Stichting Toegepast Onderzoek Waterbeheer (STOWA). With the publication of Water Matters, the participating institutions aim to make new, applicable water knowledge accessible.

You can also read the Dutch version of Water Matters digitally on H2O-online.

Would you like to respond? Let us know via

Monique Bekkenutte Publisher (Royal Dutch Network)
Huib de Vriend Chairman of the editorial board Water Matters


Knowledge journal / Edition 2 / 2023

New technology for far-reaching micropollutant removal ready for implementation

The upcoming revision of the European Urban Wastewater Treatment Directive is creating a high demand for new technologies for the removal of micropollutants from wastewater. This must not only be effective, but also sustainable (without the use of chemicals, and low-energy, so a low CO2 footprint). The new BACF-ozone technology, also referred to as BO3 technology, appears to be a good candidate. At the Amstel, Gooi & Vecht Water Authority's Horstermeer wastewater treatment plant, it has been tested on a pilot scale. Is this technology, developed by Wageningen University and engineering consultancy firm Royal HaskoningDHV, ready for implementation?

The proposal by the European Union to revise the Urban Wastewater Treatment Directive tightens up the requirements for the removal of organic micropollutants from wastewater, such as drug residues. These requirements are expected to come into force in the near future, in the period 2030-2040. The same Directive also prescribes, for instance, energy neutrality at wastewater treatment plants (WWTPs). The current technologies for the removal of organic micropollutants, such as activated carbon treatment or ozonisation, consume a great deal of energy: coal and natural gas for the production and regeneration of activated carbon and (a great deal of) electricity for the production of ozone. In addition, they are based on only one removal mechanism: either absorption of activated carbon, or oxidation by ozone. Micropollutants that are less susceptible to one of these two mechanisms are not removed or are barely removed. The width of the spectrum of removed substances is therefore limited.

Combination technology

Wageningen University's Environmental Technology chair group and Royal HaskoningDHV have together developed a new sustainable technology for the removal of micropollutants from wastewater, BACF-ozone technology, also referred to as BO3 technology. This method is based on a combination of several removal mechanisms. By combining them, a wide range of micropollutants is removed. This ties in with the objective of the national government’s ‘Chain approach to Drug Residues from water’ (Ketenaanpak Medicijnresten uit water), reducing the amount of micropollutants in water.
Partly financed by the TKI Water technology scheme (TKI-regeling Watertechnologie), BACF-ozone technology has been researched for more than six years at laboratory scale [1]. Within the STOWA Micropollutants Removal Innovation programme (STOWA Innovatieprogramma Medicijnresten Verwijdering (IPMV) ), a feasibility study of the BACF-ozone technology was carried out in 2022 [2]. The results of this study led to pilot tests. This article describes this pilot, carried out at the Amstel, Gooi & Vecht Water Authority's Horstermeer WWTP (October 2022 – April 2023).

How does BACF-ozone work?

The technology consists of two process steps (illustration 1). The first step is treatment of the water in a biological activated carbon filter (BACF), an aerobic biological purification process with granular activated carbon. In the second step, the ozone is dosed. With conventional activated carbon treatment, the carbon filter is reactivated on a regular basis. This does not happen with the BACF-ozone technology.
From the start of the pilot tests, the activated carbon in the BACF was already fully saturated and functioned as a carrier material for micro-organisms in the BACF. These specialised micro-organisms break down organic matter (measured as DOC, COD or UVA254), nitrite, ammonium, and many of the micropollutants.
The laboratory tests had already shown that, as a result, a very low ozone dose of approximately 2-2.5 mg O3/L is sufficient in the ozonisation step to remove the majority of the remaining (bio-recalcitrant) micropollutants. The energy consumption of this ozonisation step is therefore a factor 3 to 4 lower than that of a conventional ozone installation.

Set-up of the pilot study

The pilot installation at the Horstermeer WWTP was fed by the outflow from the secondary clarifiers of the WWTP, with a flow rate up to 2.5 m3/hr. The BACF pilot installation (Nijhuis Saur Industries) was equipped with a filter bed of approximately 800 L. Granular activated carbon from the BOCAC reactors of the NieuWater UltraPuurWaterfabriek water purification plant in Emmen was chosen as the filter material. This material had already been operational for more than twelve years and had never reactivated. It is therefore saturated and it serves mainly as carrier material for micro-organisms. The BACF is operated with an ‘empty bed contact time’ (EBCT) of 22, 30, and 60 minutes. The EBCT is a measure for the hydraulic contact time in the filter bed; a higher EBCT means more time for biodegradation of organic matter. In the inflow of the BACF, pure oxygen was dosed, up to a concentration of dissolved oxygen in the inflow of approximately 10 to 45 mg/L. When the filter bed became polluted (high resistance in the reactor), the BACF installation was backwashed with its own filtrate. The backwash frequency was once every 3 to 7 days.

Illustration 1. Diagram of the BACF-ozone pilot installation. The numbers 1, 2, and 3 are the three sampling points. BACF = biological activated carbon filtration and O3 = ozone reactor.

From January to March 2023, five ozone test days took place in order to research the BACF-ozone configuration. Prior to an ozone test day, the filtrate from the BACF was collected over a period of 48 hours and stored in a 30 m3 buffer tank. On an ozone test day, ozonisation tests were carried out with a flow rate of 6-7 m3/hr with a PureBlue mobile ozone installation. Ozone was fed with a diffusor system at ozone doses of 0.1, 0.2, and 0.4 g O3/g DOC (DOC in the outflow of the BACF).

Results of the pilot study

The BACF was operational continuously for 6 months. After an adaptation period of a few days, the removal of organic matter (DOC and UVA254) was the same as in the full-scale BODAC reactors in Emmen, namely 15-30%. The same applies for ammonium and nitrate removal: from the start-up, an almost complete removal (nitrification) was observed. The ammonium concentration therefore dropped from 1.0 to <0.015 mg N/L, and that of nitrite from 0.3 to <0.015 mg N/L.
After the adaption period, the removal of micropollutants was studied over the BACF. With EBCTs of 22, 30, and 60 minutes, a yield of approximately 75%, 80%, and 90% respectively was obtained for the average removal of the 7 best of the 11 guide substances, see illustration 2 (left). Looking at a wider range of substances (all 11 guide substances and 8 candidate guide substances) the removal was significantly lower (illustration 2, right): 45%-55%. For several substances, including a few European guide substances, the concentration over the BACF even increased, a phenomenon that is also known from activated sludge systems. A metabolised form of the micropollutants (for example, excreted by the body) is then converted into the original substance (carbamazepine), as a result of which it appears as if a production of carbamazepine occurs.

The addition of the ozone step to BACF improves the removal yield. This effect is clearest when we look at the 11 guide substances (illustration 2, right) and the 8 candidate guide substances (not illustrated). For an EBCT of 30 minutes and ozone dose of 0.3 g O3/g DOC, the average removal yield was approximately 80% for both the 11 guide substances and 8 candidate guide substances. Illustration 2 (left) shows that the removal yield of the 7 best of the 11 guide substances with an ozone dose of 0.4 g O3/g DOC and EBCTs of 22, 30, and 60 minutes improves to approximately 95%. This, while the maximum yield to be determined is approximately 97% (because after purification several substances fall under the laboratory reporting limit).

Illustration 2. Removal yield of BACF-ozone technology on water from wastewater treatment plant secondary clarifier. Removal yields for guide substances with various contact times (EBCT in minutes) in the BACF and with ozone doses (0.1, 0.2, and 0.4 g O3/g DOC). The removal yield is the averaged yield of the removal yields for the 7 best removed (left) and all 11 (right) guide substances averaged over the measuring days. Error bars indicate the variation between measuring days.

In addition, a so-called biological effect measurement was carried out with various bioassays (microtox, daphniatox, and CALUX tests Era, GR, P53, PAH, and PXR). It appears from this that the ecotoxicity of the water decreased by more than 50% due to the treatment with BO3 technology.

A consideration is the potential formation of bromate during ozonisation. Up to an ozone dose of approximately 0.3 g O3/g DOC, the concentration of bromate remained under the current quality requirement for surface water of 1 µg/L. For higher ozone doses, the bromate concentration was above 1 µg/L. Bromate formation mainly depends on the local water matrix, the change of the water matrix in the BACF, the ozone feeding system and the concentration of bromide.

A look ahead to practical application

The pilot tests of BACF-ozone technology can be regarded as successful, in the sense that a high removal of a wide range of micropollutants is achieved whereby the concentration of bromate remains under the surface water norm of 1 µg/L. The tests also took place without any technical or technological problems.

Furthermore, BACF-ozone technology does not use continuous activated carbon dosing or reactivation, and very low ozone doses. With a CO2 footprint of 49 to 56 g CO2/m3 WWTP flow rate, BACF-ozone technology is therefore almost a factor 2 to 4 more sustainable than the IPMV reference technologies (ozone with sand filtration, PACAS, and GAC filtration), which have a respective CO2 footprint of 90, 122, and 228 g CO2/m3 WWTP flow rate.
The costs of BACF-ozone technology per cubic metre of WWTP flow rate are comparable with those of the ozone with sand filtration reference technology. Thanks to the low consumption (of electricity and oxygen), the operational costs are low, and the impact of increasing energy and raw material prices is therefore limited.

In anticipation of the European Urban Wastewater Treatment Directive revision, BACF-ozone technology is therefore a very suitable option, with a high removal yield for a wide range of micropollutants, a low CO2 footprint, and costs that are comparable with those of reference technologies. However, just as for other ozone technologies, attention to potential bromate formation remains necessary.

Manon Bechger
(Water authority Amstel, Gooi en Vecht)
Koen van Gijn
(Wageningen University/Water authority Rijn en IJssel)
Alette Langenhoff
(Wageningen University)
Huub Rijnaarts
(Wageningen University)
Laura Piai
(Royal HaskoningDHV)
Arnoud de Wilt
(Royal HaskoningDHV)

Background picture:
The BACF-ozone pilot installation at the Horstermeer wastewater treatment plant


BACF ozone technology was developed for the removal of micropollutants at wastewater treatment plants. After successful lab tests at Wageningen University, last year the technology was researched at pilot level at the Amstel, Gooi & Vecht Water Authority’s Horstermeer wastewater treatment plant. In the pilot study, a high removal yield of approximately 80% was measured for a wide range of micropollutants (11 guide substances). Here, the concentration of bromate remained under 1 µg/L, the quality norm for surface water. The CO2 footprint of BACF ozone technology is a factor 2 to 4 lower than that of the reference technologies (PACAS, ozone with sand filtration, and GAC filtration), the costs per cubic metre are comparable. BACF ozone technology is a serious option for achieving the requirements in the planned revision of the European Urban Wastewater Treatment Directive and is ready to be deployed on a practical scale.


1. Van Gijn, K. (2023). The BO3 process for removal of micropollutants from wastewater treatment plant effluent: exploring the synergies between biological treatment and ozonation. PhD-thesis, Wageningen University.

2. STOWA 2022-41. Feasibility study BO3 technology (Haalbaarheidsstudie BO3-technologie)

^ Back to start


New BACF-ozone technology

Knowledge journal / Edition 2 / 2023

Floating treatment wetlands locally remove nutrients and micropollutants from polluted surface water

Many waters are coping with eutrophication, poor ecological condition, and pollution by pesticides and drug residues. Together, the Vallei & Veluwe Water Authority and Wageningen University researched floating treatment wetlands as a potential solution. They appeared to be particularly suitable for slightly polluted water.

Constructed wetlands (helophyte filters) are nature-based systems that are used to remove nutrients and micropollutants from treated wastewater. Moreover, they reduce the ecotoxicity of the water and are known as sustainable, environmental-friendly, inexpensive, and easy to manage [1, 4]. This way, besides purifying water, they contribute to many of the current climate, energy, and biodiversity challenges.


The Zijdewetering in Veenendaal is a slow flowing drainage canal on sand that is mainly fed by effluent from the wastewater treatment plant in Ede. Despite the high removal efficiency of this wastewater treatment plant for nitrogen, the water remains too nutrient-rich, which causes excessive plant growth. The nocturnal oxygen concentrations are therefore low, and the ecological status is poor. Constructed treatment wetlands would be ideal for this location to treat the effluent water for multiple contaminants and improve local ecology at the same time [4]. However, physically there is not enough land area for this.

Figure 1. Removal of nutrients and micropollutants by floating treatment wetland.

Floating helophyte filters

Where there is a shortage of land area, floating treatment wetlands can be used in existing waterways. The choice of the substrate in these wetlands determines their treatment performance, and the use of a bio-based substrates contributes to their sustainability. However, there is not enough research into the effect of floating wetlands with bio-based substrates on water quality.
Within this study, floating treatment wetlands with jute or mycelium substrate were studied. This was done in a controlled mecocosm setup with water from the Zijdewetering. Oxygen dynamics were monitored together with the removal of nutrients and micropollutants. The results were scaled up for the whole Zijdewetering with calculations. In addition to the mesocosms, a field test was carried out with floating treatment wetlands in the Zijdewetering to monitor their development (in light of water currents and the weather).


The research setup consisted of 39 48-litre mecocosms, each with a miniature floating wetland and 12 plants in various combinations (reed, reed mannagrass, and yellow iris). The substrates used were mycelium, jute or ‘no substrate’ (control). The mycelium was made from Ganoderma fungus cultivated on wood waste. Every 5 days, half of the water was replaced.
The removal efficiency was assessed 24 hours after replacing the water (equal to the retention time of water in the Zijdewetering). Ammonium, nitrogen, and phosphor were measured using Hach Lange tests; nitrite, nitrate, phosphate, and sulphate using ion chromatography; and for micropollutants, we used liquid chromatography. For every sample, the pH, temperature, electric conductivity, and oxygen concentration were determined.


Looking at nutrient removal in constructed wetlands with jute, mycelium, and ‘no substrate’, it is evident that mycelium has a lower purification yield for phosphate, ammonium, and total nitrogen. Furthermore, for half of the mycelium wetlands, we see an increase in phosphate (figure 3). This was caused by gradual breakdown of the mycelium, which released nutrients. The mycelium breakdown also has negative effects on nitrogen removal: mycelium wetlands removed nitrogen significantly less compared to jute- and the control wetlands.
Jute wetlands, however, remove phosphate efficiently compared to control mecocosms. Jute absorbed phosphate within 8 hours, which increased the removal efficiency from 70% (control) to 100%.

Figure 2. Sample mecocosm with mycelium substrate planted with reed mannagrass (photo Hazel van Waijjen)


Of the twenty micropollutants studied, sixteen were present above the detection limit of 50 nanograms per litre. Their removal was better than expected: for ten substances, more than 70% was removed (figure 3). The removal efficiency of mycelium wetlands was again relatively low. This is likely caused by the presence of easily degradable organic material in these wetlands. Bacteria that break down organic micropollutants prefer to ’eat’ mycelium first. removal efficiency in jute wetlands was comparable to the control treatment.


The breakdown of mycelium and jute in the mecocosms required a great deal of oxygen. All mecocosms contained less than 1 milligram of oxygen per litre, which also led to lower ammonium removal with jute and mycelium, since complete nitrification is not possible without oxygen.
Healthy oxygen dynamics are essential for healthy water ecology as well as for ammonium removal. Submerged water plants use oxygen at night, which can lead to almost anaerobic conditions in case of too much vegetation. Border plants such as reeds, however, give off oxygen via their roots [1], also at night. The hypothesis is that floating wetlands with emergent water plants outcompete part of the submerged water plants and thus realise more robust oxygen dynamics. This will be further researched.

Seasonal dependency

The removal efficiency for micropollutants had halved by the end of the summer, compared with mid-summer. During the growing season, plants contribute to pollutant removal by direct absorption and by stimulating the microbiome, e.g. by giving off oxygen and signal substances. Aging of plants reduces these contributions. In addition, the breakdown of plants at the end of the summer leads to a surplus of easily digestible plant residue, which decreases the breakdown of micropollutants. Just like with the mycelium substrate, the bacteria then prefer ‘eating’ the plant residue.
The addition of activated carbon or biochar to the wetlands can compensate for this lower removal efficiency. Biochar and activated carbon have a considerable absorbing capacity [2, 4], as a result of which micropollutants are retained during the winter to be broken down during the growing season. As a result, a natural version of a BODAC system emerges: a biologically enhanced activated carbon filter that absorbs pollution, which is then broken down aerobically by microbes [4].
The seasons also have a considerable influence on the removal of nutrients. In spring, plants absorb most nutrients, and release them again in autumn. In order to guarantee a high removal, biomass must be harvested twice per year, in early summer and early autumn. The great advantage of floating wetlands is that roots, where most nutrients are stored, can also be harvested [1, 3].
For that matter, problems with water quality are also seasonal. Most problems occur during the summer due to high biological activity and poisonous algal blooms.

Figure 3. Removal of phosphate, ammonium, total nitrogen, and micropollutants by containing floating treatment wetlands after 24 hrs of treatment (23 August). Removal efficiency of nutrients shown per mecocosm (njute= 14, nmycelium= 10, ncontrol= 15) . Removal of 16 micropollutants per mecocosm (njute= 224, nmycelium= 160, ncontrol= 240).

Stability of floating helophyte filters

In the field test with planted jute and mycelium wetlands in the Zijdewetering, the mycelium started to break down within a month. The jute, which was fastened around a floating PVC frame, also teared within a month. In a later six-month test, floating PVC frames covered with gauze and filled with jute were used. These appeared to be sturdier and provided excellent support for plant growth.

Added value

Floating wetlands are multifunctional; they not only remove nutrients and micropollutants, but also remove pathogens, reduce ecotoxicity, and work as an ecological buffer. In addition, they are adaptive, robust, and aesthetic, as a result of which they are suitable for application in urban areas. There, they contribute to mental health, have a cooling effect, and create natural habitats for city wildlife [5]. In short, floating wetlands are an integral ‘no regret’ solution to several problems[1, 4].

Conclusions and advice

Floating treatment wetlands can be an enrichment to (urban) ponds thanks to their positive effect on water quality and ecology. However, the removal efficiency for nutrients is too low for considerable eutrophic water or for controlling (blue-green) algae nuisance. Too many wetlands would be needed for an adequate effect. They are extremely suitable for the treatment of slightly polluted water, for the revitalisation of urban water bodies, or for ecosystem services and additional nature value.

The advice is to choose a robust and sustainable design of recycled materials (such as second-hand PVC or gauze), filled with the bio-based substrates jute and biochar (which retains micropollutants during winter). Our research shows that the type of plant species have less effect on the water quality (data not presented). Therefore we advise endemic fast-growing flowering plants, thanks to their positive effect on insects. This vegetation must be harvested twice yearly for efficient nutrient removal, whereby the harvested material can serve for the production of energy or materials (biogas, fibres, reed roofs, etc.)


This research was carried out as an internship at the Vallei & Veluwe Water Authority, and was made possible by SIGN (thanks to Peter Oei, Laila Kestem, and Dewi Hartkamp), and Wageningen University Environmental Technology.

Hazel van Waijjen
(Vallei & Veluwe Water Authority)
Frans de Bles
(Vallei & Veluwe Water Authority)
Anita Buschgens
(Vallei & Veluwe Water Authority)
Katarzyna Kujawa-Roeleveld
(Wageningen University)

Background picture:
Field study with floating treatment wetland in the Zijdewetering (photo by Richard Huinink).


In this study, we researched the effect of floating treatment wetlands on the water quality of surface water. Two bio-based substrates were applied: mycelium (made from wood waste) and jute. Mycelium showed to break down quickly, and nutrients were released instead of removed. However, jute resulted in excellent phosphate removal and good nitrogen removal. During the summer, the jute wetlands were very effective in removing micropollutants, but this effect decreased considerably towards autumn. This can be remedied by adding biochar or activated carbon in order to retain contaminants for breakdown during the following summer. All in all, floating treatment wetlands are robust and very suitable for cleaning up and revitalising slightly polluted (urban) water bodies.


1. Shen, S., X. Li, and X. Lu, (2021). Recent developments and applications of floating treatment wetlands for treating different source waters: a review. Environ Sci Pollut Res Int, 28(44): p. 62061-62084. 26

2. Lei, Y., et al., (2021). Sorption of micropollutants on selected constructed wetland support matrices. Chemosphere, 275: p. 130050

3. Kadlec, R., J. Vymazal, (2005). Vegetation effect on ammonia reduction in treatment wetlands, in Natural and Constructed wetlands: Nutrients, Metals and Management. Leiden. p. 233-260

4. Van de Beuk, J. et al., (2022). Study of natural purification systems for removal of organic micropollutants (Verkenning natuurlijke zuiveringssystemen voor verwijdering van organische microverontreinigingen). STOWA: Amersfoort

5. Semeraro, T. et al., (2021). Planning of Urban Green Spaces: An Ecological Perspective on Human Benefits. Land, 10(2)

^ Back to start


Cleaning lightly polluted waters

Knowledge journal / Edition 2 / 2023

Risk assessment of blue-green algae: fast, accurate or both?

Water authorities and provinces wish to know the health risk posed by toxic blue-green algae in their bathing waters. The ideal method for this is accurate and fast, as well as affordable. The available range varies from not-very-specific-but-fast biomass assessments to precise-but-expensive toxin measurements. How good do these methods assess the health risk for recreational swimmers, and is there an optimum method or combination of methods?

Every summer, many recreational swimmers happily make use of the various bathing water locations in the Netherlands, which is exactly the time when blue-green algae can occur in high densities. Various species of blue-green algae can produce toxins, which forms a risk for the health of people and animals as with densities of blue-green algae, also the concentrations of toxins can increase considerably. Not all species of blue-green algae produce toxins, and even within a species toxic and non-toxic strains exist. The non-toxic strains lack a gene for producing toxins.


In 2018, during the bathing season (1 May to 1 October), we carried out various methods for assessing health risks posed by blue-green algae at bathing water locations in eleven Dutch lakes. They were selected based on their distribution throughout the country (see map), and historical data showing that blue-green algae were found in the past. The assessment involved measurements of:
1. the densities of all blue-green algae (fluorescence);
2. the volume of potentially toxic blue-green algal species (microscopy);
3. toxin genes (DNA test, quantitative PCR);
4. toxin concentrations (immunology test (ELISA) or chemical analysis (LC-MS/MS)).
(See [1] and [2] for details, stakeholders involved and a complete overview of the results).

The bathing water locations studied.

Blue-green algae everywhere

We found blue-green algae at all bathing water locations. The amounts varied considerably, both between the locations and throughout the swimming season. Based on microscopic analyses, we identified the presence of potentially toxic blue-green algae in all of the lakes; yet, the amounts of toxin genes and concentrations of toxins found were often low. This is a first indication that non-toxic gene species are present.


The most common toxin found in the lakes was microcystin, which is therefore the focus for this further research. This toxin was mainly found at the bathing water locations in the Oudegaasterbrekken and Plas Ter Werve, with peaks in concentrations of 10.6 and15.3 micrograms per litre, respectively. These values lie above the norm of 10 micrograms per litre for the first risk level (or a ‘Warning’) of the Dutch Cyanobacteria Protocol 2020 [3]. The toxins cylindrospermopsin and anatoxin occurred in a few samples and only in concentrations of less than 0.25 microgram per litre. For these toxins, there is currently no norm included in the Cyanobacteria Protocol 2020, but the concentrations found lie far below the indicative norm of 3 and 60 micrograms per litre for cylindrospermopsin and anatoxin, respectively, that is indicated by the World Health Organisation.

Comparison of methods

In order to test how well the various methods assess the health risk posed by blue-green algae, we studied the relationship between the specific methods and toxin concentrations. As a ‘golden standard’, we use the microcystin concentration on the basis of the chemical analysis (by means of LC-MS/MS).
The fluorescence method focuses on the total amount of blue-green algae and shows a weak relationship with the concentration of microcystins. This is in line with expectations, since not all blue-green algae produce toxins. The biovolume of potential microcystin producing blue-green algae gives a better relationship with this toxin as it is already more specific toward toxic genera and species. The concentration of microcystin genes and the concentration of microcystins, measured with the laboratory test (ELISA) show a very strong relationship with microcystin concentrations measured using the chemical analysis (LC-MS/MS).
From these results, it appears that the more specific the method focuses on toxin production (potentially microcystin producing blue-green algal species or the presence of toxin genes) or the toxin itself, the more accurate this method estimates the amount of toxins and the associated risk.

Table 1. Overview of risk groups on the basis of various methods for Oudegaasterbrekken during the bathing season 2018, with no risk (0; green), warning (1; yellow), and negative swimming advice (2; red). Risk assessment on the basis of microcystin genes has not been added due to lacking norms. Relative speed, costs, and specificity of the various methods in relation to each other vary from low (+) to high (+++).

Swimming advice

The choice of the measuring method also has consequences for the recreational swimmer, because the various methods can lead to a different swimming advice (0 = no risk; 1 = warning; 2 = negative swimming advice). This can be seen clearly for Oudegaasterbrekken (see the table), where the assessment of the total amount of blue-green algae on the basis of fluorescence for the entire bathing season issues a risk (warning or negative swimming advice), while on the basis of the microcystin concentration only a warning for one month would be issued, i.e., for a period of four months, there would be an overestimation of the risk. During that period, there were indeed blue-green algae, but they produced little to no toxins. If the biovolume of potentially toxic blue-green algae is assessed, in this case an improvement can be seen of the risk assessment, although it is still overestimated for two months. Yet, from analysis of existing data from more than 4000 data points, it appears that both fluorescence and microscopy correlate with each other, which would mean that these two methods could generally produce a comparable advice [2]. This corresponds with the current study if we include all data points. These show that in 61 percent of cases the advice ‘no risk’ is issued on the basis of fluorescence, and 52 percent on the basis of the biovolume. This therefore means that the advice on the basis of fluorescence overestimates the risk in specific cases, but that the difference between fluorescence and biovolume assessment is small.


The research shows that measuring methods for toxins give the most accurate picture of the risk, followed by assessment of the biovolume of potentially toxic blue-green algae with microscopic analysis. Measurements of the total blue-green algae biomass with fluorescence produce many overestimations of the risk. In these cases, there are blue-green algae present, but a large part does not produce toxins. When blue-green algae biomass was low, any overestimation based on fluorescence is not an issue, because the risk advice in such cases remains at level zero (i.e. no risk). Regarding the costs, measuring the total blue-green algae biomass using fluorescence is the cheapest. Toxin measurements are more expensive, especially for individual assessments of samples. The costs may decrease considerably if more samples are assessed at the same time.
On the basis of this analysis, we propose combining methods. A fast and inexpensive first risk assessment can be carried out on the basis of fluorescence measurements, if necessary followed by a more accurate measurement of the toxin concentration (ELISA or LC-MS/MS). This will prevent unnecessary warnings and unnecessary negative swimming advice in the case of low quantities of toxins. In addition, this will eliminate unnecessary anxiety amongst recreational swimmers, entrepreneurs, and administrators.


The proposed two-step procedure fits within the current Cyanobacteria Protocol 2020 used for risk assessment in the Netherlands, because in the protocol, a fluorescence measurement can be combined with a toxin measurement. However, this currently only applies to microcystins. In addition, the protocol only concerns blue-green algae in the water column during the bathing season, which is rather limited. We would propose the following recommendations for a potential evaluation of the Cyanobacteria Protocol 2020 as well as for future studies:
1. Currently, guideline values for other toxins than microcystins are lacking in the protocal. There are indicative guideline values available for cylindrospermopsin, anatoxin, and saxitoxin from the World Health Organisation [4]. The embedding of these norms and the related measurement protocols can contribute to an improved risk assessment.
2. Both at the beginning and at the end of the bathing season, we sometimes found increased quantities of blue-green algae, toxin genes, and toxin concentrations. This makes it likely that swimmers also run a risk outside of the bathing season, in particular because, with a changing climate, the water temperature will often be higher in spring and autumn, which can lead to more blue-green algae development [5].
3. The focus of this study was on blue-green algae in the water column. However, blue-green algae also live on the sediment (benthic blue-green algae), which includes species that could potentially produce very toxic saxitoxins and anatoxins. If these blue-green algae are detached from the sediment (or other substrates), they can wash up on a swimming beach in small clumps with potentially high toxin concentrations. The Cyanobacteria Protocol 2020 offers general guidelines for estimating this risk. A better understanding of the growth of benthic blue-green algae, its release from the sediment, and the toxin concentrations in clumps of these blue-green algae is necessary for a better estimation of the human health risk.

Dedmer van de Waal
(Netherlands Institute of Ecology; Amsterdam University)
Quirijn Schürmann
(Waardenburg Ecology; Amsterdam University)
Edwin Kardinaal
(Waardenburg Ecology)
Susan Sollie
(TAUW consulting engineers)
Petra Visser
(Amsterdam University)

Background picture:
Blue-green algae in the Grote Plas in Delftse Hout park (Delft).


Blue-green algae occur every summer in many Dutch bathing waters. In this study, the accuracy of methods for assessing the health risk posed blue-green algae was compared. The methods focus on all blue-green algae (fluorescence), potentially toxic blue-green algae (microscopy), toxin genes (qPCR), and toxin concentrations (ELISA and LC-MS/MS). The cheapest and fastest method was fluorescence, but this method gives an overestimation of the risk. Toxin measurements give better estimates but are often more expensive and the results take longer. We therefore propose a combination of methods: a first screening with a fast fluorescence method, after which a (corrective) toxin measurement can be carried out when norms are exceeded. Such a combination of methods must guarantee that overestimations of the health risk are limited as much as possible, while the safety of swimming waters remains guaranteed.


1. Sollie, S., Kardinaal, W.E.A. Faassen, E. J. (2020). Risk assessment of blue-green algae in recreational waters: New techniques for determining the presence of blue-green algae toxins (Risicobeoordeling blauwalgen in zwemwater: Nieuwe technieken voor de bepaling van de aanwezigheid van blauwalgtoxines). (Stowa report; No. 2020-09). Foundation for Applied Water Research (STOWA).

2. Schürmann, Q.J.F. et al. (in review). Risk assessment of toxic cyanobacterial blooms in recreational waters: A comparative study of monitoring methods.

3. Schets, F.M. et al. (2020). Blue-green algae protocol 2020.

4. Chorus, I., Welker, M., (2021). Toxic Cyanobacteria in Water: A Guide to Their Public Health Consequences, Monitoring and Management. Taylor & Francis.

5. Royal Netherlands Meteorological Institute (KNMI (2023)). KNMI’23 climate scenarios for the Netherlands (KNMI’23-klimaatscenario’s voor Nederland), KNMI, De Bilt, KNMI Publication 23-03.

^ Back to start


Determining risks

Knowledge journal / Edition 2 / 2023

Wastewater measurements support GGD Municipal Health Service during jaundice outbreak at a primary school

Now that the Covid test lanes have disappeared, the RIVM’s National Wastewater Surveillance is the ideal instrument for continuing to follow the coronavirus circulation. However, wastewater measurements can also be useful for other virus diseases, as demonstrated by an outbreak of hepatitis A at a school in Amsterdam.

During the years of Covid, the instrument of wastewater measurements developed quickly, resulting in an added value for tracking the spread of the virus and public health. Wastewater surveillance also proved it could supply reliable and useful information about virus circulation [3] at the neighbourhood level. This was the reason for GGD Rotterdam-Rijnmond to conduct further research into the usability of wastewater measurements to support public health. This was carried out together with GGD Amsterdam, KWR, Partners4UrbanWater, IMD, and Erasmus MC, and partly financed by STOWA Foundation for Applied Water Research and TKI Water technology.

Jaundice in Amsterdam

In 2022, when an outbreak of hepatitis A virus (HAV) occurred at a school in Amsterdam, the GGD wondered whether wastewater measurements could help them to gain insight into the effectiveness of their outbreak control measures. Hepatitis A is an acute viral inflammation of the liver. The disease often goes unnoticed in young children, but they can transmit the virus to adults who can often become ill.
The infection mainly occurs in tropical and subtropical countries. In the Netherlands, there are between 100 and 200 infections per year, 40 percent of which are import infections: people become infected during a visit to countries where hepatitis A is endemic [5]. Since children are often asymptomatic and since the incubation period is long (4 weeks), the virus can be present for some time before it is noticed (silent circulation).
Hepatitis A virus is transmitted via faeces and absorption via the mouth. Hepatitis A is a notifiable infectious disease; doctors and laboratories must inform the GGD of a case. The GGD’s response is usually contact tracing and implementation of transmission control measures. Just as with Covid, wastewater measurements can make HAV infections in the school and in surrounding residential areas visible. If outbreak control is effective, HAV would disappear from the wastewater of the school and neighbourhoods.

The outbreak

In October 2022, an Amsterdam hospital reported a child with jaundice to GGD Amsterdam. After contact tracing in late October and early November 2022, another three infected children were identified and, in late December 2022, there was a fifth case. All five children attended the same primary school, in different classes and both in the main building and an annex. After genetic typing, it turned out that all five were infected with the same virus strain. The sick children were kept at home to recover and to prevent spreading.
Meanwhile, GGD Amsterdam set to work. In late October, a letter was sent to the school management with the request to inform parents and carers and to ask them to contact the doctor if their children were to show symptoms of HAV. The school forwarded this letter to all parents and carers. GGD Amsterdam then quickly organised an HAV vaccination campaign at the school, to which only a few parents responded (14 percent of the children). Due to the low turnout, an information session was organised for the school and the parents. In mid-January, a sixth HAV infection was reported in the same residential area. This person did not have any link with the other cases or the school.

Illustration 1. A passive sampler is collected from the wastewater after 48 hours of sampling.

Setup of wastewater measurements

In the wastewater network around the school and annex building, two wastewater drains and two wastewater pumping stations were selected for taking representative wastewater samples. The two wastewater drains collected wastewater from both school buildings, the pumping stations from the surrounding residential areas (20,000 to 30,000 people). To show HAV, passive samplers were used, which were hung at a location for a few days in order to collect faeces containing HAV from the wastewater flow. These were 3D-printed torpedo-shaped samplers with cotton buds, which were placed and fixed in the wastewater.
The sampling took place from mid-November 2022 to March 2023. Because a fifth infection was identified at the school in late December 2022 and this child lived in a different residential area, the wastewater measurements were expanded to the wastewater pumping station of this residential area and to the wastewater treatment plant in question.
During the first measuring week in November 2022, per wastewater drain and wastewater pumping station, one passive sampler was deployed in 96 hours; in the following weeks, they were deployed for 48 hours. The samplers were transported on ice to the laboratory and examined the same day. The cotton bud from the passive samplers was processed in the same way as previously for SARS-CoV-2 [2]. The extracts were processed with a special PCR method, the digital droplet RT-PCR (derived from [4]), with positive controls.
Five passive samplers from November 2022 from the wastewater network of Rotterdam were taken as controls (so outside of the outbreak area). In these samples, no HAV was found, as expected. Extracts in which RNA from HAV was found were sent to the RIVM for sequence analysis.
From the samples, in addition to HAV, CrAssphage was also measured. CrAssphage is unique to the human gut and is used here as a measure for the amount of human faeces that was collected by the passive samplers from the wastewater [3].


In all passive samplers, CrAssphage was found in reasonably comparable amounts (5.2 ± 1.1 million gene copies per sampler), which indicated that all the samplers had collected comparable amounts of human faeces.
During the first weeks, HAV was only found in the wastewater of the annex, later also in the water from the wastewater pumping stations and the main building. 8 of the 17 samples that had been taken from November 2022 to March 2023 from the annex were positive for HAV, always in low quantities. From the main building, 1 of the 17 samples was positive (in February 2023). From the wastewater pumping stations, 2 and 7 of the 17 samples were positive for HAV, also always in low quantities. In the wastewater pumping station and the wastewater treatment plant that were added in January 2023 as a result of the fifth case of the disease, in 3 and 2 of the 10 samples respectively HAV was also found, again in low quantities.
The genetic typing of HAV in the wastewater proved to be valuable: from November 2022 to early February 2023, it was possible to genetically type HAV in 7 samples and the detected HAV appeared to be genetically identical to the strain that was found in the sick children. After late December, no new cases were reported that were linked to this outbreak. After early February 2023, this strain was no longer found in wastewater. For the GGD, this was confirmation that no more silent circulation of the outbreak strain took place in and around the school. However, it was intriguing that from early February 2023, a different HAV strain was found in the wastewater. During this period, typing of HAV in 4 samples was successful, always of this different type. The GGD did not receive any reports from people with this HAV strain. Measurement and typing of HAV in wastewater therefore indicated that silent transmission of this other HAV strain took place in the same residential area in Amsterdam.

Illustration 2. Digital droplet RT-PCR for identifying HAV in wastewater.


The close collaboration with the GGDs has shown that wastewater measurements supply valuable information, also in the case of an outbreak of a different virus disease than COVID-19, namely hepatitis A. For both infectious diseases, a (large) part of the infections is mild or without symptoms, certainly in young children. Hence, these infections can be easily missed during standard surveillance. Another similarity is that some of the infections can be serious. It is therefore important to limit transmission as much as possible.
With the outbreak at the school in Amsterdam, wastewater surveillance proved to be valuable to follow potential silent transmission, and wastewater measurements confirmed that the outbreak was over. Furthermore, it was demonstrated that early, weekly measurements with passive samplers at four (later six) locations, can be an efficient method to follow the HAV circulation without much difficulty and disturbance for the school (children) and residential areas in question.
Genetic typing was necessary during this study, both in order to confirm that with the PCR method HAV was actually identified, and to confirm that the outbreak strain was found in the wastewater. By coincidence, via the wastewater measurements, the import and potential transmission of another HAV strain came to light, just after and near the first outbreak.
For the GGD, wastewater research in the case of a Hepatitis A outbreak (and potentially also outbreaks of viral enteric and respiratory infections) can support both contact tracing and outbreak control, in showing whether and where virus circulation takes place ‘under the radar’.

With thanks to George Sips and Paul Bijkerk (GGD Rotterdam Rijnmond), Goffe Elsinga and Leo Heijnen (KWR), Jeroen Langeveld (Partners4UrbanWater), Ewout Fanoy (GGD Amsterdam), Bert Pasma (STOWA), J. Guldemeester (Erasmus MC), and Dave McCarthy, (Queensland University of Technology, Australia).

Gertjan Medema
(KWR Water Research Institute)
Remy Schilperoort
Miranda de Graaf
(Erasmus MC)
Harry Vennema
(RIVM National Institute for Public Health and the Environment)
Maarten de Jong
(GGD Amsterdam)

Background picture:
A passive sampler that has just been placed in the wastewater.


In the case of an outbreak of hepatitis A (HAV) at a school in Amsterdam, measurements of wastewater were used to follow the spread of the virus. Via sampling in wastewater drains and pumping stations, two school buildings and the surrounding residential areas were monitored. In four months, there were five cases of the disease. The same virus strain was involved in all cases. The same HAV strain occurred regularly in low concentrations in the wastewater and was no longer found after early February. This confirmed that the outbreak of that HAV strain was over. However, via wastewater measurements, ‘silent spreading’ of another HAV strain was later found in the same residential area.


1. HAVNET. Protocol Molecular detection and typing of VP1 region of Hepatitis A Virus (HAV).

2. Heijnen, et al. (2021). Droplet digital RT-PCR to detect SARS-CoV-2 signature mutations of variants of concern in wastewater. Sci Total Environ. 799:149456.

3. Langeveld, et al. (2022). Wastewater surveillance in Rotterdam-Rijnmond 2020-2022 (Rioolwatersurveillance in Rotterdam-Rijnmond 2020-2022). Water Matters, 15.

4. Persson, et al. (2021). A new assay for quantitative detection of hepatitis A virus. J Virol Methods 288:114010.

5. RIVM (2023). Hepatitis A | LCI guidelines (

^ Back to start


Outbreak of Hepatitis A

Knowledge journal / Edition 2 / 2023

Peatland inundation test in the Zwanburger polder

The peatland areas in the west and north of the Netherlands are affected by peat oxidation. The Zwanburger polder near Leiden is such a peatland polder. There, Joost van Schie wishes to transform his organic cheese farm, De Eenzaamheid, into a regenerative farm. This means working according to a system that enriches the soil, increases biodiversity, improves water management, retains carbon, and improves the ecosystem services. How can peat oxidation be stopped while preserving the landscape?

Peat degradation as a result of oxidation causes land subsidence, the emission of greenhouse gases, and leaching of nutrients. The fact that the peat layer in the Zwanburger polder can oxidate is probably due to oxygen that can penetrate into the peat layer through shrinkage cracks in the covering clay layer. The question is: if oxygen can do this, does this also apply to water? And if so: can we stop the peat oxidation process by increasing the ground water level through inundation ? This would be good for the soil, the water quality, and biodiversity (illustration 1).

A first step towards reaching an answer was done during an inundation test in the dry, warm summer period in August 2022. The project was carried out with the aid of Deltares, an international water and subsurface knowledge institute, in close collaboration with Van Schie [1] and the Rijnland District Water Board.
During this first test, it was explored whether peat can be saturated quickly, cheaply, and effectively by inundating a parcel with ditch water. If this was the case, it could potentially be a cheaper and faster alternative to other (technological) methods for increasing the groundwater level, such as underwater drainage and pressure drainage. An additional advantage: inundation is a rewetting measure that does not permanently change the soil and landscape.

Soil and groundwater system

Previous research in the Zwanburger polder charted the local soil and groundwater system, including the relationships between groundwater systems, water and groundwater quality, subsidence, greenhouse gas emissions, and biodiversity [2]. The relationship is as follows.

The soil in the polder consists of a peat layer (reed peat), sandwiched between two layers of clay: a thin top clay layer and a deep, thick clay layer to approximately five metres below ground level. Below this is a (mudflat) sand layer. From the sand layer seepage water rises, which is fed by the surrounding higher-lying canals. The winter groundwater level is close below ground level (above the ditch level) and is managed by narrow trenches of approximately 30 cm deep. In spring, these trenches are dry, and during dry summer periods the groundwater level drops further. This drop is almost entirely caused by evaporation from the grassland. Between the ditches, ‘hollow’ groundwater levels are created, which are up to 60 cm below the ditch level. This results in drying (hardening) of the topmost clay layer and peat degradation. During the wet season, this results in leaching of nutrients and eutrophication of the ditches.

Illustration 1. Schematisation of the subsurface in the Zwanburger polder. Land subsidence can be prevented by reducing peat oxidation. This also leads to a reduction in CO2 emissions, increased water quality due to less nutrient leaching, and a healthier soil with greater biodiversity.

The inundation test

The test parcel was 145 m long and 60-80 m wide. It was split into two parts: one area of 40 x 80 m to be inundated (the inundation parcel) and a reference parcel, which was not inundated. The areas were separated by a screen placed up to about 25-30 cm in depth in order to prevent leakage through the shallow clay layer (sandy clay). This worked well, but a great deal of water unexpectedly flowed under the screen through the peat layer.

In the inundation parcel, six monitoring wells with filters were installed in the non-decomposed (intact) peat. They were equipped with hydrostatic pressure sensors (so-called divers). In addition, two extra monitoring wells were installed in order to be able to take manual measurements during the test. Using bentonite and protective sleeves, it was possible to prevent potential leakages along the monitoring wells.

In the reference parcel, as reference measuring points, two monitoring wells were installed in the non-decomposed peat layer, at 60 m from the screen. In order to measure the subsurface movement, 8 metal rods were hammered 25 cm into the ground. Before, during, and after the inundation, the subsurface movement was tracked using a DGPS (a very accurate GPS system). The pressure head in the mudflat layer was measured at a distance of 200 m from the inundation parcel. For the ditch level fluctuations, measurements from the Rijnland District Water Board were used. Illustration 2 gives an overview of the test setup.

Illustration 2. Overview of the test setup, with the inundation parcel in the bottom left and the reference parcel in the top right. Inundation pump was by monitoring wells 7 and 8.

On August 3rd ditch water was pumped into the inundation parcel for 5.5 hours. At an estimated flow rate of 25 l/s, that is ± 500 m3. This is equivalent to a water layer of 15-16 cm.

Test results

Within a few hours of the pump being switched off, the inundation water had already sunk into the ground. The clay layer therefore barely formed an obstruction for the infiltration. Within a few hours, the groundwater levels in the inundation parcel rose by about 70 – 110 cm (monitoring wells numbers 2 to 8, see illustration 3). At the end of the test period, 21 days after the inundation, the groundwater level was still a few decimetres higher than before the inundation.
Unexpectedly, the inundation water also easily spread though the peat layer to the reference parcel; 36 hours after the start of pumping, the peat was completely saturated at the two reference measurement points (monitoring wells 9 and 10, see illustration 3). The drop in the groundwater levels in the inundation parcel can mainly be explained by this flow towards the reference parcel.

Illustration 3. Time series of the groundwater levels in all monitoring wells. Inundation happened on the 3rd August (between the red dotted lines).

In addition, there was also a clear short-term correlation between groundwater level fluctuations and subsurface movement. After the inundation, the ground level was 4 - 59 mm higher than at the end of the test period, which follows the same dynamics as the measured groundwater levels. This shows how the ground level is influenced by the sponge-like behaviour of the subsurface and possibly also by swell processes.
From August 1st to August 26th (the end of the trial period), 17,766 m3 was pumped out of the polder by the polder pumping station. The water extracted from the ditch during the test was only a fraction of this (2.8 %). Irrigating a parcel during the warm summer months therefore barely results in any additional stress on the ditch water level.

Conclusions, recommendations, and applicability

The test shows that the peat in the Zwanburger polder can be saturated quickly, cheaply, and effectively by irrigating (inundating) it with (surplus) ditch water. Within a few hours, the groundwater level in the entire parcel (including the reference parcel) had risen by approximately 40 cm and the peat in the inundation parcel was completely saturated. The drop in the groundwater levels in the inundation parcel after the initial rise can mainly be explained by the subsurface flow towards the reference parcel. Yet, the effect of the irrigation was still noticeable a few weeks after the inundation. This, while the inundation parcel was already accessible the morning after the inundation, which means that for a livestock farmer no delay or obstruction would occur in the grazing plan.

Long term monitoring is needed to tell whether a surface level drop can be successfully reduced by means of a good inundation plan. A follow-up study is already being worked on. In this study, a whole parcel will be inundated instead of only a part of the parcel. This is expected to have a number of effects:
- After inundation, the groundwater level will only be influenced by evaporation or potentially by run-off under the ditches to adjacent non-inundated parcels.
- It is advisable to lay a temporary (aboveground) pipe network, which will be fed by one pump, or a gravity system that uses water from surrounding higher-lying canals if the parcels are surrounded by a dike. When using a pump, solar energy can be used.
- No screens will be necessary, because parcels are usually surrounded by a small elevation as a result of dredging; this will limit the construction costs.

The follow-up study will also look at the effect the inundation has on the water quality of the adjoining ditches and the groundwater, and at the effect on the emission of greenhouse gases (especially methane emissions, because they play a greater role than CO2-emissions at groundwater levels of 20 cm or less under ground level [3]). Furthermore, research will be done into whether it is possible to increase the inflow from the ditches, instead of inundating a parcel, in view of the high permeability of the peat layer that is intersected by the ditches.

If this follow-up is also successful, it will open the door to scaling up inundation. For De Eenzaamheid farm, this could mean a scaling-up to the whole farm (40.7 ha), but a scaling-up to other farms and polders with similar characteristics will also be possible. Important conditions for success are the availability of surplus ditch water and similar soil charateristcs (high peat permeability). Water authorities can play an important role, especially in assessing whether surrounding basins or ditches can be considered as sources of ‘surplus’ water.

Looking at climate change, it is important to be economical with the available ditch water, especially in the summer. One point of attention is that, depending on the conditions, inundation with ditch water can have a negative influence on groundwater levels in surrounding parcels. Furthermore, with poor permeable soils, inundation can lead to local wet spots in a parcel, while other parts of the parcel are not affected and remain dry. If the wet spots remain swampy for a long period of time, the parcel will be inaccessible for longer and the grass could be damaged.

Nevertheless, if the correct characteristics are present, inundation as a rewetting measure has considerable advantages. It can be realised quickly, and it is a cheap measure that does not leave behind any permanent changes to the landscape. Farmers may offset the inundation costs with (higher) savings on installation and maintenance costs for drainage or spraying. Furthermore, farmers only need to inundate their parcels during times of need.

Laura Nougues
Roelof Stuurman
Henk Kooi
Joost van Schie
(De Eenzaamheid farm)

Background picture:
Inundation test in full swing


Dutch peatland areas are affected by peat degradation due to oxidation. This causes land subsidence, greenhouse gas emissions, and leaching of nutrients into the ditch water.

The Zwanburger polder near Leiden is such a peatland area. This article describes an inundation test (temporary irrigation with ditch water), which shows that − under the correct conditions (soil qualities, water management) − inundation can be a cheap rewetting measure that effectively curbs the peat oxidation. The researchers also cast an eye to further research and application in practice.


1. Wij.Land (2020). Regenerative agriculture in the Zwanburger polder (Regeneratieve landbouw in de Zwanburgerpolder).

2. Nougues, L. (2021). Limiting land subsidence of an island polder with a clay - peat subsurface.

3. Erkens, G., Melman, R., Jansen, S., Boonman, J., Hefting, M., Keuskamp, J., Bootsma, H., Nougues, L., van den Berg, M., van der Velde, Y. (2022). Subsurface Organic Matter Emission Registration Systems (SOMERS). Description SOMERS 1.0, underlying models and peatland calculation models (Beschrijving SOMERS 1.0, onderliggende modellen en veenweiderekenregels).

^ Back to start


Inundation slows peat erosion

Knowledge journal / Edition 2 / 2023

Sand transplantation of rapid sand filters for better removal of organic micropollutants for drinking water production

For decades, rapid sand filtration, or rapid filtration, has been used in drinking water purification for the removal of suspended matter and organic and inorganic material (such as methane, ammonium, iron and manganese), among other things. For a large part of these substances, the removal takes place biologically, in other words with the aid of micro-organisms in the sand filter. Rapid filtration can potentially also be used for the removal of organic micropollutants (OMPs). Can the removal of OMPs also be stimulated by means of sand transplantation?

OMPs refer to a wide range of organic substances, such as drugs, pesticides, and industrial substances, which occur in low concentrations (in the order of µg/L to ng/L) in surface water and groundwater, and which can have various physical and chemical properties. These substances must be removed as part of drinking water production. For this purpose, additions and/or adjustments to the current purification process are required.

Rapid filtration is already a part of the process in many treatment plants and could possibly also be used for the removal of OMPs, in addition to or even as an alternative for other process techniques to remove OMPs. Physicochemical techniques often require a great deal of energy or chemicals, and in the case of membrane filtration and activated carbon filtration they produce a waste stream or waste product. This does not apply to biological removal. Furthermore, biological removal can completely break down OMPs, without producing waste.

There are more and more indications that rapid sand filters can (partly) remove a wide range of OMPs biologically. This can be done by micro-organisms that use the OMPs for growth (metabolic) or by micro-organisms that oxidise the OMPs simultaneously with ammonium, methane, iron or manganese oxidation (co-metabolic) [1]. It is likely that no additional large investment is required to optimise rapid sand filters for OMP removal, which makes this solution sustainable and environmentally friendly. Furthermore, rapid filtration is often one of the first steps in drinking water purification. A large part of the OMPs is therefore already removed before the water enters the rest of the purification process.

Previous research has shown that there are considerable differences in the presence of OMPs and the removal capacity between rapid sand filters of various production locations [2]. In this project, it was therefore investigated whether OMP removal can be stimulated by means of sand transplantation (augmenting rapid sand filters with material from another filter) and/or biostimulation (adding substances to rapid sand filters, which stimulate the growth of micro-organisms).
Similar processes (bioaugmentation and biostimulation) have already been successfully used for years for the cleaning of polluted soils [3].

Column tests with sand transplantation

The experiments that we describe in this article were conducted with four columns, filled with sand from a full-scale rapid sand filter that was bad at removing OMPs. In three of the four columns, sand was transplanted (10% w/w) from a full-scale rapid sand filter that was good at removing (a number of) OMPs.

As influent for the columns, the natural surface water (influent from the full-scale filters) was used, to which a mixture of 10 OMPs was added (acesulfame, amidotrizoic acid, benzotriazole, carbamazepine, diclofenac, iopamidol, metoprolol, propranolol, gabapentin, and perfluorooctanoic acid (PFOA)), whether supplemented with nutrients (ammonium, nitrate and phosphate) and a vitamin/trace element solution or not (see illustration 1). Most OMPs had a final concentration in the influent of 10 µg/L, apart from diclofenac (1 µg/L) and amidotrizoic acid (25 µg/L). These OMPs differ in their removal during drinking water purification, as well as in their properties, such as water solubility.
The influents were dosed to the columns (3 litres per hour; contact time 0.25 hour), see illustration 1.

Illustration 1. Diagram of the column tests (left) and the setup in the experimental hall (right). In all four columns, surface water was dosed with added OMPs. In three of the four columns, sand transplantation was applied and 3 out of 4 columns received nutrients and/or vitamins/trace elements (vit/traces).

OMPs removed?

After 80 days, acesulfame, gabapentin, and metoprolol were removed very well (99%), diclofenac and propranolol were removed well (30-55%), and PFOA, carbamazepine, 1-H benzotriazole, amidotrizoate, and iopamidol were removed poorly (< 20%). The transplanted columns showed 99% removal of acesulfame after 45 days, while the non-transplanted column only achieved this after 80 days. In addition, the transplanted columns removed 99% of gabapentin and metoprolol within 24 days, while after 60 days, the non-transplanted column still did not remove any gabapentin and only removed 40% of metoprolol (illustration 2). Because the removal of acesulfame, gabapentin, and metoprolol increased over time, this shows that these OMPs were probably removed biologically, metabolically. In this process, micro-organisms use the OMPs as energy and as a carbon source in order to grow, with the aid of oxygen or nitrate.

A negative relationship was also discovered between OMP removal and ammonium oxidation, and the number of ammonium oxidating micro-organisms decreased during the column tests. This shows that OMP removal was not linked to ammonium oxidation via co-metabolism. Sand transplantation therefore appears to be potentially effective for a more rapid and/or improved removal of these substances in rapid sand filters. On the other hand, propranolol and diclofenac appear to be removed in particular by means of sorption processes, because removal is stable or even decreases due to saturation of the sand.

Illustration 2. The results of the column tests. The percentage of removal of acesulfame, gabapentin, metoprolol, diclofenac, and propranolol by the 4 different columns.

Practical significance

Before sand transplantation and biostimulation can be applied on a larger scale, more research is needed into a wide range of OMPs, long-term effects (how the filter continues to perform in the long term), and seasonal fluctuations in the water to be purified. Examples of the latter include the presence and concentrations of OMPs and nutrients and environmental factors such as the temperature of the water to be purified. The contact time with the filter may also influence OMP removal. This because, in the research described here, a number of OMPs were not removed sufficiently in 0.25 hours, but would potentially be removed with longer contact times. A number of OMPs will probably never be removable with rapid sand filters, so complementary purification steps remain essential.

Sand transplantation and biostimulation (adding nutrients and/or vitamins and trace elements) can ensure a more rapid and/or improved removal of certain OMPs. Biostimulation has been used for years at the Waternet water company, which doses phosphate to rapid sand filters in order to stimulate microbial nitrification at low temperatures. This could possibly be expanded in the future with other nutrients to stimulate microbial OMP removal.

On a large scale, sand transplantation is not the ideal form for adding OMP removing micro-organisms, especially because large volumes of sand must be transplanted. In addition, it is unknown whether at the time of transplantation the right micro-organisms are sufficiently present, and we often do not yet know what the optimum conditions are for the OMP-removing micro-organisms.

Bioaugmentation − supplementing specific micro-organisms − would be more effective, because smaller volumes of concentrated cell lines can be added. Bioaugmentation also allows specific addition of the right micro-organisms. Furthermore, it is possible to study the optimum conditions for the micro-organisms and their presence and growth are easier to monitor with molecular DNA methods.
Bioaugmentation has been used in soil pollution for years [3], but it has not yet found its way into waste and drinking water purification. If follow-up research shows bioaugmentation to be successful, it will be especially important in the drinking water sector that micro-organisms are deployed in such a way that they do not pose a risk to the water quality of the resulting drinking water. For instance, it is advisable to introduce indigenous micro-organisms that have been cultivated from sand filters, so that the natural population is not disturbed.

This research was carried out as part of the sector inquiry of the drinking water companies, in collaboration with the water companies (PWN, Waternet, Dunea, WML, and Evides).

Peer Timmers
(KWR water research institute)
Wolter Siegers
(KWR water research institute)
Paul van der Wielen
(KWR water research institute)
Marco Dignum

Background picture:
In the experimental hall: column experiment with rapid sand


Rapid sand filtration can potentially be used for the removal of organic micropollutants (OMPs). This project is a study, on a laboratory scale, of whether OMP removal can be stimulated by means of sand transplantation (augmenting rapid sand filters with material from another, better performing sand filter) and/or biostimulation (adding nutrients which stimulate the growth of micro-organisms). The results show that three of the ten dosed OMPs (acesulfame, gabapentin, and metoprolol) are biologically removed and that sand transplantation and biostimulation can be effective in removing these substances more rapidly and/or in a better day. Before large-scale application is possible, more research is required, especially into a wider range of OMPs, the long-term effects, and the seasonal fluctuations in the water to be purified. In addition, it is necessary to cultivate the micro-organisms that are responsible for OMP degradation, so that they can be augmented and monitored specifically in rapid sand filters.


1. Wang J., de Ridder D., van der Wal A. & Sutton N.B. (2021). Harnessing biodegradation potential of rapid sand filtration for organic micropollutant removal from drinking water: A review, Critical Reviews in Environmental Science and Technology, 51:18, 2086-2118, DOI: 10.1080/10643389.2020.1771888

2. Di Marcantonio C, Bertelkamp C, van Bel N, Pronk TE, Timmers PHA, van der Wielen P, Brunner AM (2020). Organic micropollutant removal in full-scale rapid sand filters used for drinking water treatment in The Netherlands and Belgium. Chemosphere. 2020 Dec;260:127630. doi: 10.1016/j.chemosphere.2020.127630. Epub 2020 Jul 12. PMID: 32758778.

3. Van Ras N., Volkers B. (2008). In-situ stimulated biological degradation: a natural solution! Dutch Centre for Soil Quality Management and Knowledge Transfer (SKB) (In-situ gestimuleerde biologische afbraak: een natuurlijke oplossing! Stichting Kennisontwikkeling Kennisoverdracht Bodem cahier).

4. Timmers P.H.A., Lousada Ferreira M., Siegers W. (2022). Bioremediation of rapid sand filters for removal of organic micropollutants (Bioremediatie van snelfilters voor verwijdering van organische microverontreinigingen, BTO 2022.042).

5. Peer H.A. Timmers et al. (2023). Bioremediation of rapid sand filters for removal of organic micropollutants during drinking water production, Water Research, 120921, ISSN 0043-1354,

^ Back to start


Purifying with sand transplants

Knowledge journal / Edition 2 / 2023

Climate-robust grass-herb mixtures in reconstructed ditches on sandy soils

Sowing wide, gradually sloping banks of reconstructed ditches and streams with a mix of flood-resistant grasses, legumes, and other herbs contributes to a climate-robust water system. At the same time, it is also possible to continue to use the precious land for agriculture. HAS green academy studied a number of plant species for their suitability and their response to flooding and drought.

Climate change is leading to longer periods of drought and more severe peak downpours. The high sandy soils in the Netherlands are sensitive to extreme drought and therefore solutions are being sought to retain (rain) water for longer, especially in the area of agriculture, both in the stream valleys and on the high peaks. In the LUMBRICUS study (, experiments were carried out with, among other things, impoundments and soil improvers. The ongoing KLIMAP project ( is researching paludiculture and drought-resistant crops as options for making water storage profitable in lowland landscapes. Other measures are cancelling the numerous plot drainages and reconstructing ditches by widening them and making them shallower with more gradual slopes (Figure 1, i.e. photograph in the background). In such a reconstructed ditch, water has longer to sink into the (water) bed. If making ditches shallower is possible with a neutral ground budget, this will guarantee feasibility.
Sowing these reconstructed ditches with grasses, legumes, and other herbs offers the possibility of still using this precious agricultural land for the production of cattle feed. In a ditch that has been made shallower, under dry, but also under normal conditions, grazing is possible.
Since the reconstructed ditches drain water less well, heavy rainfall can cause wetter conditions temporarily. The vegetation must be able to withstand this. In greenhouse experiments, HAS green academy studied which species of grasses and herbs are suitable for this.

Figure 1. A classic ditch (left) has a steep slope, a ditch made shallow has a gradual slope. In reality, a ditch made shallow is also much wider.


We conducted an experiment with twelve species: meadow fescue, common cocksfoot, tall fescue, English ryegrass, Kentucky bluegrass, common birdsfoot, red clover, white clover, yarrow, caraway, narrowleaf plantain, and chicory. These species are often found in commercial mixtures. In the experiment, the species were tested individually for sensitivity to both long periods of drought and flooding. In addition, we compared two species mixtures, one with 30 percent herbs and one with 70 percent herbs. Drought and flooding were stimulated by respectively not giving the plants water for seven days or submerging them in water for 14 days (Figure 2).

Figure 2. An impression of the greenhouse experiments.

From the experiment, it appears that temporary flooding does not have a negative effect on the survival of grasses, legumes and other herbs, the plants adapt by forming superficial roots. During the first two weeks after flooding, the dry matter yield for a number of species was lower than for plants which were given water as required. In the three weeks after the flooding, the grasses and herbs recover to such an extent that the dry matter yield is ultimately comparable to plants which were watered as required.


Unlike in the case of flooding, drought does indeed influence survival; after seven days without water, all species show considerable symptoms of drought. It was striking that three weeks after the drought period, red clover, common birdsfoot, and narrowleaf plantain had not recovered at all, and the other species had a much lower dry matter yield.
Recovery after flooding by forming extra superficial roots at the top of the bed can temporarily lead to a lower drought tolerance. They can then no longer use deeper water unless they can form deep roots again in time.

Impact of drought and flooding

The greenhouse experiments appear to indicate that temporary flooding of reconstructed ditches and lowland landscapes will not have a negative impact on the survival of most grasses, legumes, and herbs. Extreme drought has a negative effect on survival, but this effect is probably less strong in a reconstructed watercourse or lowland landscape; this will have to be further researched. In the field, plants can also form deeper roots than in the greenhouse, as a result of which better drought resistance can be developed, even if flooding can have negative consequences for deep roots. This trade-off must be studied carefully.


During the second greenhouse experiment, we compared the yield of various grass-herb mixtures. After the drought treatment, the more herb-rich vegetation appeared to yield a higher fresh and dry weight and a higher protein content.
The idea of working with mixtures is that this probably extra enhances the resilience of the vegetation (apart from potential mutual influence) [1, 3, 5].
This resilience originates from a larger diversity in root systems, which complement each other in their working and interaction with soil-dwelling organisms. A better root environment also ensures more efficient absorption and division of nutrients and, therefore, better resilience to conditions such as drought and flooding. Furthermore, the use of a mixture as diverse as possible can increase the biodiversity both below and above ground, thanks to the positive effect of a richer soil life. An additional advantage of mixed species is that the ultimate species composition can vary according to the microclimate. Species that are less resilient to drought will establish themselves proportionally more in the lower located areas; species that are less tolerant to flooding will occur more in the higher areas. The moisture gradient therefore causes a gradient in species composition, which can again lead to extra biodiversity [4].

Figure 3. Roots of meadow fescue immediately after 14 days of flooding (left), 14 days of water according to requirements (centre), and 14 days without water (right).

Implications and applications

The use of grasses, legumes, and other herbs can have a positive effect on the nutritional value of the grassland, for example, a higher protein content and medicinal properties of some herb species.
In addition to the researched species, other grasses and herbs are also potentially interesting, including dandelion and timothy grass because of their health benefits for dairy cattle, and, for example, buckwheat because of its resilience to wet and dry conditions. This is why we recommend sowing as diverse a mixture of suitable species of grasses and herbs as possible.
A risk of a diverse vegetation is the pest pressure that can occur because particular pests are attracted by some herbs [1], which can lead to nuisance in adjacent crops.
It is therefore important that the exact composition per situation is determined specifically based on the microclimate and the flora and fauna present [1, 2].
Resilience to flooding can also be realised by planting suitable trees and shrubs, although they could also result in undesired shade for the main crop. An additional disadvantage lies in the laws and regulations for such fixed vegetations and the resulting obligatory registration as landscape element (Common Agricultural Policy (CAP) from 2023). For farmers, planting trees and shrubs is therefore often still a bridge too far.


The general principle of making ditches shallower and more gradual is widely applicable. At the same time, it is important in each situation to achieve the right combination of plants and microgeography, because every agricultural plot has its own challenges and possibilities. Farmers who are aware of these principles are positive about making ditches shallower and more gradual, certainly if the connecting ditch network has already been made or will be made less shallow.
Sowing the right herb-rich grass mixtures of reconstructed ditches with very gradual slopes will ensure that the vegetation survives after a period of (anticipated) flooding, while the water will have the time to sink into the deeper soil, which will help in extreme summer drought. In this way, landscape elements that retain water will help dairy farmers with grazing but will also ensure higher biodiversity and higher buffering and filtrating capacity of the landscape.


This project was carried out by the Climate-robust landscapes lectorate (HAS green academy), together with RNOB (collaborating municipalities and water authorities of north-east Brabant). Thanks to Marnix van der Kruis and Peter Kerkhofs (RNOB), to HAS students Job Bals, Judith Leussink, Jules Terken, Tessa van Raay, Robin Straub, and Lieke Roovers, and to Martijn Bekkers and Frans van den Broek (HAS greenhouse).

Fabian E.Z. Ercan
(HAS green academy)
Mark van de Wouw
(HAS green academy)
Ellen Weerman
(HAS green academy)

Background picture:
Ditch made shallow for farmer Jos Vos from Sterksel (NB), sown with an herb-rich vegetation mix.


It is becoming increasingly important to retain water on the sandy soils, so that it is available for agriculture, nature, and people during drier periods. A possible measure is reconstructing streams and ditches: making them shallow, widening them, and making slopes more gradual. Because this leads to a decrease in water drainage, temporary wet conditions could occur due to heavy rainfall. The vegetation must be resistant to this. In a greenhouse experiment with twelve species of grasses and herbs, most species proved to be resilient to flooding, but not or less resilient to drought. Flooding leads to loss of deeper roots and the formation of new surface root systems.


1. Bais, H. P. et al. (2004). How plants communicate using the underground information superhighway. Trends in plant science, 9(1), 26-32.

2. Bardgett, R. D., Mommer, L., & De Vries, F. T. (2014). Going underground: root traits as drivers of ecosystem processes. Trends in Ecology & Evolution, 29(12), 692-699.

3. Chen, X. et al. (2020). Effects of plant diversity on soil carbon in diverse ecosystems: A global meta‐analysis. Biological Reviews, 95(1), 167-183.

4. Tölgyesi, C. et al. (2022). Turning old foes into new allies—Harnessing drainage canals for biodiversity conservation in a desiccated European lowland region. Journal of Applied Ecology, 59(1), 89-102.

5. Van Eekeren, N. (2009). A mixture of grass and clover combines the positive effects of both plant species on selected soil biota. Applied Soil Ecology, 42(3), 254-263.

^ Back to start


Grass herb mixtures tested

Knowledge journal / Edition 2 / 2023

From continual measurement to continual knowledge; optimisation of data from water quality sensors

Continual water quality data provide a great deal of useful information.
Water quality managers therefore make more and more frequent use of sensor technology and auto-analysers. However, due to errors in measurements, the raw data from water quality sensors are not immediately usable for visualisation and interpretation. In this article, we explain the state-of-the-art, as a first step towards standardisation of data processing from water quality sensors.

Water authorities and water authority laboratories are investing more and more in the application of water quality sensors in surface water. For example, water authority laboratory AQUON currently manages about 200 water quality sensors. Nowadays, they can not only measure standard parameters, such as pH, electrical conductivity, oxygen, and temperature, but also, for example, nutrient concentrations. All these continual and real-time measurements provide a great deal of valuable information, for example, about drainage, the dynamic influence of wastewater treatment plant effluent and overflows, transport processes of nutrients, or operational water management (for example, deciding to let water in or barring it on the basis of salt concentrations).

Scaling-up and reliability

Due to the sensitivity of the technology and the variable conditions in the field, there are often irregularities in the sensor data. The raw data are therefore not immediately usable for visualisation and interpretation. However, due to the scaling-up in use of water quality sensors, manual checking and correction of the data are often a great deal of work. Depending on the intended use of the data, standardised data processing methods are therefore desirable (table 1). The aim of this research was to bring together standard optimisation routines for high-frequency water quality data and to make them available to users of water quality sensors (researchers and water managers) via a wiki-site [1].

Table 1. Aims of data optimisation and the related methods


Through interviews and literature research, we charted which deviations occur frequently and which correction methods are already publicly available. Many methods have been developed in other fields. We included them in our search if they had already been applied to sensor data for water quality. For the various types of deviations in sensor data, we created optimisation routines, including the R scripts and example applications, and made these available via the wiki site [1]. The routines included there are partly (combinations of) methods that are already available, but we developed new methods for, for example, the correction of drift on the basis of laboratory measurements and for filling in gaps in measurement series.

Desk research – international experiences

There are various types of deviation possible in sensor data (illustration 1).
Incidental peaks (and troughs) occur as a result of brief disruptions or electronic instability. For example, many nitrate sensors work on the basis of absorption of UV light, which can be disrupted by a leaf or a snail.
A flatline (continual fixed value) occurs in the case of prolonged malfunctions, for example, when the sensor dries up or when the concentration falls outside of the measurement range.
Noise occurs frequently when the values to be measured fall low in the range of the measurement method; in this case, disruption soon occurs due to varying environmental conditions such as turbidity.
The cause of drift is usually (bio)fouling on the sensor, which disrupts contact with the water to be measured. Upon cleaning, a jump often occurs in the measurement series.
Gaps in measurement series occur in the case of malfunctions or the removal of deviating data.

Three general insights from the interviews and the literature:
• Data validation and correction are very complex. Faulty measurements are difficult to distinguish from real variations. This makes corrections for water quality more difficult than, for example, for groundwater levels or surface water levels and drainage.
• Techniques for automatic anomaly detection, smoothing (by determining the general line in the data through averaging) and noise correction are readily available, but still not often applied to water quality data.
• Most sensor applications for water quality are from scientific groups, who use their own scripts for data correction without reporting them. Recently, a few frontrunners, however, published method descriptions and scripts for the processing of water quality sensor data [2, 3, 4, 5].

The published data processing methods all begin with simple and objective corrections, such as removing impossible values and incidental peaks. They are then followed by more advanced correction methods whereby new data are also generated, and subjective choices become important. The raw data, the cleaned data, as well as the corrected data are then saved in different layers.

Illustration 1. Standard types of deviation in sensor data for water quality

Results – data optimisation routines

For all standard types of deviation (illustration 1), scripts and example applications were published. We collected them and made them available in a publicly accessible wiki [1].
In this article, we will not go into the relatively simple methods for detecting flatlines and noise in depth. However, we will describe the recommended methods for detecting peaks, correcting drift and jumps, and filling in gaps in measurement series.


Correcting peaks in water quality data in particular is risky because concentration peaks can also actually occur. Screening for truly impossible values is relatively easy by setting up a range, which can potentially also vary in time. For example, the range for water temperature could be 5 15 °¬C in winter and 10 30 ¬¬°¬C in summer.

A more advanced technique for anomaly detection is the feature-based approach [5]. On the basis of the measurement series, series of several features, such as log transformation, slope, and the ratio between the minimum and maximum value within a period are calculated. In order to detect anomalies in these feature series, we combine this with the isolation forest method. This looks at how easy it is to isolate a data point from the rest of the data set (in several dimensions, depending on the number of features). Illustration 2 gives an example of the result. This method works more quickly and more effectively than the k-nearest neighbour (kNN) algorithm [5], which is also used, but is not perfect either. Especially incidental, difficult-to-predict peaks, for example, in the case of wastewater overflows or other drainage, can be identified incorrectly as an anomaly. In the case of doubt, a response from other sensors (other parameters, nearby locations, water levels) can offer clarity. Sensor malfunctions are also often easier to distinguish from actual occurring concentration peaks using a higher measuring frequency.

Illustration 2. Example of anomaly detection of peaks with the feature-based approach and isolation forest method for a continual measurement series of nitrate concentrations in surface water

Drift and jumps

For the correction of drift and jumps, we have developed the DRIMP correction. The basic principle for this is that standard water quality monitoring also takes place at the sensor location. There had not yet been a method for drift correction in combination with conventional control measurements. Illustration 3 shows an example. The DRIMP correction splits the continual measurement series according to the jumps caused by maintenance moments (red dotted lines). For each part of the series, we create a linear model of the difference between the sensor value (green line) and the conventional lab analysis (red dots). The corrected series (blue line) can then be calculated using the linear model. When using this method, it is important that sufficient conventional measurements are available and that maintenance moments are registered correctly. Sampling during concentration peaks can only be used if the sampling time can be accurately linked to the sensor series.

Illustration 3. Example of correction of drift and jumps in a continual measurement series of nitrate concentrations in surface water


It may be necessary to fill in gaps in the measurement series, for example, for a good estimate of pollutant charges. Gaps in the measurement series are easy to fill in, for instance, by running through the last measured value, linear interpolation between the last and following measurement value, or filling in with the average value from the series. A time series analysis model (ARIMA) to fill in data gaps is often more accurate but works mainly for short gaps [2, 3]. For longer gaps, it helps to utilise the relationship with other continually measured parameters, for example, with linear regression between nitrate concentration and drainage or phosphate concentration and turbidity. However, in many cases the relationships are not linear and also not constant over time. In these cases, the random forest model [6] can be used, which predicts the relationship between various continually measured parameters.


The usability of random forest models for anomaly detection by continually predicting the next measurement value and then comparing this with the measured value (one step ahead predictions) is worth further research. There are also more possibilities for combining sensor data with conventional water quality measurements, for example, data assimilation techniques (such as Kalman filters). However, such automated data processing methods are merely aids; good water quality monitoring is and remains human work. Collaboration within the chain of suppliers, sensor managers, data specialists, and water quality experts accelerates the development of useful data processing methods.

For many water managers, it is essential to work towards standardisation and certification of sensor applications and of suppliers. Standardisation would make it easier to relate measurements from various measurement nets to each other and to use sensor data for reporting purposes (Water Directive, Bathing Water Directive). An independent party would then have to determine which sensors and methods of work should be used. Practically speaking, however, we are not there yet. We do not expect either that sensor measurements will replace standardised laboratory measurements. However, water quality sensors often offer a great deal of additional insight, which helps with choosing the right measures to improve the water quality.

Kevin Ouwerkerk
(Deltares knowledge institute)
Joachim Rozemeijer
(Deltares knowledge institute)
Eppe Nieuwenhuis
(AQUON water authority laboratory)
Frank van Herpen
(Water authority Aa en Maas)
Joep Appels
(MicroLAN water quality monitoring specialist)

Background picture:
Sensor measurement in the Vinkenloop (North-Brabant)


Water quality sensors provide water managers with a great deal of useful information. The sensors can measure more and more parameters (continually) and are deployed on an increasingly large scale. However, measurements often show irregularities due to disruptions in the environment. Nonetheless, due to scaling up the use of sensors, manual control and correction of sensor data are often a great deal of work. For the reliability and comparability of data, standardised data processing methods are therefore desirable. This article takes a first step towards such a standardisation and makes the collected state-of-the-art and newly developed methods publicly available to water managers in a wiki.



2. Spackman Jones, A., T.L. Jones, J. S. Horsburgh, 2022. Toward automating post processing of aquatic sensor data. Environmental Modelling and Software 151.

3. Hawkins, 2021. User guide to fine resolution (15 minute) data. Version 1.10.

4. Schmidt, L. et al., 2022: System for Automated Quality Control (Saqc) to Enable Traceable and Reproducible Data Streams in Environmental Science. SSRN Electronic Journal.

5. Talagala, P. D. et al., 2019. A feature-based procedure for detecting technical outliers in water-quality data from in situ sensors. Water Resources Research 55.

6. Barcala, V., Rozemeijer, J., Ouwerkerk, K. et al., 2023. Value and limitations of machine learning in high-frequency nutrient data for gap-filling, forecasting, and transport process interpretation. Environ Monit Assess 195.

^ Back to start


Data processing methods


Edition 1/2023

Edition 12/2022

Edition 1/2022

Edition 2/2021

Edition 1/2021

Edition 2/2020

Edition 1/2020

Edition 2/2019

Edition 1/2019

Edition 1/2018

Edition 2/2017

Edition 1/2017

Edition 2/2016

Edition 1/2016

Edition 2/2015

Edition 1/2015


Previous editions

Knowledge journal / Edition 2 / 2023


The knowledge section Water Matters of H2O is an initiative of

Royal Dutch Waternetwerk
Independent knowledge networking organisation for and by Dutch water professionals.

Water Matters is supported by

Independent institute for applied research in the field of water, subsurface and infrastructure. Throughout the world, we work on smart solutions, innovations and applications for people, environment and society.

KWR Watercycle Research Institute
Institute that assists society in optimally organising and using the water cycle by creating knowledge through research; building bridges between science, business and society; promoting societal innovation by applying knowledge.

Royal HaskoningDHV
Independent international engineering and project management consultancy that contributes to a sustainable environment in cooperation with clients and partners.

Foundation for Applied Water Research (STOWA)
Knowledge centre of the regional water managers (mostly the Water Boards) in the Netherlands. Its mission is to develop, collect, distribute and implement applied knowledge, which the water managers need in order to adequately carry out the tasks that their work supports.