DEL175 - Voedselveiligheid in een circulair water- en voedselsysteem

Introduction - Towards safer use of surface water for crop irrigation

Climate change and increasing water scarcity are putting pressure on freshwater availability for agriculture. Retaining water for longer periods, reusing treated wastewater and making greater use of surface water can help secure sufficient irrigation water. At the same time, these developments can create new pathways for microbial contamination to reach crops intended for human consumption. In this project, we investigated how the microbiological safety of irrigation water can be assessed more effectively. Rather than relying only on occasional measurements, we combined multi-year microbial monitoring, information on contamination sources, weather data and water-system modelling. The aim was to explore a more dynamic and risk-based approach that can help growers and water managers make better-informed decisions about irrigation-water use.

The challenge

Surface-water quality is influenced by many different processes and contamination sources. Treated wastewater, sewer overflows, runoff from agricultural land, animals, septic systems and other local sources can all contribute microorganisms to the water system. Rainfall, flow direction, dilution and other hydrological conditions subsequently determine how these contaminants move through the catchment. For growers, this creates an important challenge. A water sample provides information about microbial quality at one location and one moment in time, whereas contamination can occur in short-lived peaks. Current irrigation-water risk assessments therefore may not capture all periods of elevated microbial risk.

A central question in the project was: Can irrigation-water safety be assessed more effectively by combining measurements with knowledge of the water system and predictive models?

What we did

The project combined field monitoring, microbial analyses and modelling in two agricultural catchments in the Dutch province of North Brabant.

Over a period of three years, E. coli concentrations were monitored at six surface-water locations: three in the Aa of Weerijs catchment within the Brabantse Delta water authority and three around Heeswijk-Dinther within the Aa en Maas water authority. In total, the monitoring dataset contained almost 500 water samples.

In addition to E. coli, selected human pathogens were analysed during part of the monitoring programme. These measurements were used to investigate whether E. coli concentrations can also provide information about the presence of specific pathogens.

The monitoring data were subsequently linked to information about rainfall, sewer overflows and hydrological conditions. Two complementary modelling approaches were explored:

• A machine-learning model was developed to investigate whether exceedances of the E. coli threshold could be predicted using weather data and location.
• Hydrological water-system models were adapted to simulate how microbial contamination can move from potential upstream sources towards downstream irrigation-water abstraction points.

Together, these approaches allowed the project to move beyond describing microbial water quality towards understanding why peaks occur and when elevated risks may reach downstream water users.

Figuur: Study areas in North Brabant. Left: Aa of Weerijs (Brabantse Delta). Right: Heeswijk-Dinther (Aa en Maas). In total, E. coli was monitored for three years at six locations across these two catchments.

Key findings

Microbial water quality can change rapidly
The multi-year monitoring showed substantial temporal variability in E. coli concentrations. Although most measurements remained below the commonly used value of 1,000 CFU/100 mL, short-lived and sometimes very high peaks occurred at all monitoring locations. These peaks indicate that faecal contamination can enter the water system intermittently. As a result, a measurement taken on one day may give a very different impression of microbial water quality than a measurement taken only a few days later. The results therefore show that a single annual E. coli measurement is not sufficient to represent microbial water quality throughout an entire growing season

Figuur: Temporal variation in E. coli concentrations in irrigation water. Most measurements remained below the commonly used threshold of 1,000 CFU/100 mL, but short-lived and sometimes very high peaks occurred, showing that a single annual measurement does not sufficiently represent microbial water quality during the growing season.

E. coli does not tell the whole story
E. coli is widely used as an indicator of faecal contamination and plays an important role in irrigation-water risk assessment. However, an indicator organism does not necessarily behave in the same way as individual human pathogens. During one year of the project, selected pathogens were therefore measured alongside E. coli. Within this more limited pathogen dataset, no clear relationship was observed between E. coli concentrations and pathogens such as Campylobacter and Salmonella. This does not mean that E. coli is not useful. It remains an important indicator of faecal contamination. However, the results indicate that an elevated, or low, E. coli concentration cannot automatically be translated into the presence or absence of a specific pathogen.

Rainfall and sewer overflows are important factors
The project found a clear relationship between rainfall and elevated E. coli concentrations. Increases in E. coli were also associated with the activation of nearby sewer overflows. Importantly, contamination does not necessarily appear immediately downstream. There can be a delay between a rainfall or overflow event and the arrival of the microbial contamination peak at an irrigation-water abstraction point. This delay depends on factors such as distance from the source, flow velocity and hydrological conditions. This means that information from elsewhere in the catchment can be relevant for a grower’s local irrigation decision. Heavy rainfall upstream, for example, may create a microbial risk downstream even when little or no rainfall has occurred at the field itself.

From measuring to predicting
Microbial measurements remain essential, but they only provide a snapshot of water quality. The project therefore investigated whether models could complement measurements by helping to identify when and where elevated microbial risks are more likely to occur. A Long Short-Term Memory machine-learning model was trained using E. coli monitoring data, weather variables and sampling locations. In the independent test dataset, the model correctly classified 90 of 128 samples as either above or below the 1,000 CFU/100 mL threshold. The results demonstrate that weather information contains useful predictive information about microbial water-quality events. However, not every contamination event is driven by weather. Sources such as waterfowl, unexpected discharges or operational changes in the water system cannot necessarily be predicted from meteorological information alone. Predictive models should therefore complement rather than replace monitoring and local knowledge.

Understanding transport through the water system
Hydrological modelling was used to investigate how contamination can move through the surface-water network. For the Aa of Weerijs case, travel times were calculated between an upstream hypothetical contamination source and locations up to 14 km downstream. The simulations showed that travel times vary considerably throughout the year because of changing hydrological conditions. The model compared these travel times with a microbial decay timescale. During periods when contaminated water travels quickly through the system, microorganisms may reach downstream irrigation locations before substantial decay has occurred. The results therefore show that distance alone is not enough to assess microbial risk: hydrological conditions determine how rapidly contamination actually reaches downstream users.

Figuur: Hydrological model output for the Aa of Weerijs case study. The figure shows travel times from a fictive upstream contamination source to downstream locations at 7, 10, 11 and 14 km. Where the travel-time curves fall below the T99 line (6 days), E. coli-contaminated water can reach the irrigation intake before substantial decay has occurred.

Towards a more risk-based approach

The project results suggest that irrigation-water safety could be strengthened by moving from a predominantly measurement-based approach towards a more area- and risk-based system.
Three complementary elements are proposed:

1. Measure: Microbial measurements remain the foundation of the assessment. Measurements should preferably be taken close to the irrigation-water intake and close to the moment when the water will be used. Building multi-year datasets and sharing measurements between growers and water authorities can provide much more information about recurring spatial and temporal patterns than individual measurements alone.
2. Observe: Measurements should be interpreted in the context of the surrounding catchment. Relevant risk sources can include wastewater-treatment plants, sewer overflows, septic systems, livestock, manure application, recreation and bird populations. Knowledge of rainfall and upstream events can provide additional information about whether microbial contamination may be moving towards an abstraction point.
3. Predict: Weather data and water-system models can help anticipate elevated microbial risks. Combining local monitoring data with weather information could support area-specific predictions of E. coli exceedances. Hydrological modelling can additionally estimate how contamination moves through the water network and when a microbial peak could reach a particular downstream location.

Together, Measure – Observe – Predict provides a framework in which monitoring, local knowledge and modelling reinforce each other.

The results show that irrigation-water safety is not solely a question of whether a single water sample complies with a threshold. Microbial quality is dynamic and is influenced by processes throughout the catchment. In practices this means that for growers, better information on rainfall, upstream contamination sources and predicted microbial water quality could support decisions about when irrigation should preferably be avoided or when additional measurements may be useful. Water authorities already possess detailed knowledge of hydrology, rainfall, sewer-overflow operation and water-system management. Combining this information with microbial monitoring provides an opportunity to support downstream agricultural users with more timely and spatially relevant information. The project therefore provides a starting point for further development of decision-support approaches in which water-quality data and water-system information are shared between growers, water authorities and other actors in the food and water sectors.

Deltares contribution

Deltares contributed its combined expertise in water microbiology and water-system modelling to connect microbial measurements with processes occurring throughout the catchment.

The Deltares contribution included:

• design and execution of microbial field monitoring;
• analysis and interpretation of E. coli and human-pathogen measurements;
• assessment of microbial contamination patterns and potential risk sources;
• hydrological modelling of contaminant transport through surface-water systems;
• estimation of travel times between upstream sources and downstream irrigation locations;
• translation of monitoring and modelling results towards more risk-based water-management approaches.

This combination of microbiological monitoring and hydrological modelling is particularly valuable when individual measurements alone cannot capture rapidly changing environmental conditions.

Who can use these results?

The project findings are relevant to several groups involved in the use and management of surface water.

• Growers and grower organisations: use the insights to improve decisions about when and where surface water can be safely used for irrigation.

• Water authorities: use monitoring and water-system information to identify upstream microbial risks and provide more targeted information to downstream water users.

• Food-safety organisations and regulators: use the findings in discussions about how irrigation-water risk assessments could better account for temporal variability and local conditions.

• Water reuse and climate adaptation projects: apply the approach when assessing the health and food-safety implications of retaining or reusing water in increasingly circular water systems.

Project information

Publications

Final report: Voedselveiligheid in een circulair water- en voedselsysteem.pdf